Anna Duden - Environmental Impacts of Bioenergy - Ex-Ante Spatially-Explicit Assessment of Biodiversity and Hydrology Impacts of Increase

Environmental impacts
of bioenergy
Ex-ante spatially-explicit assessment of
biodiversity and hydrology impacts of
increased biomass demand
Anna Sarah Duden
AnnaDuden_BNW.indd 1 16-03-20 12:26

Environmental impacts of bioenergy
Ex-ante spatially-explicit assessment of biodiversity and hydrology impacts of
Anna Sarah Duden, January 2020
This work was carried out within the BE-Basic R&D program, which was granted a FES
subsidy by the Dutch Ministry of Economic affairs, Agriculture and Innovation (EL&I).
ISBN: 978-90-8672-910-4
Printing: Ridderprint | www.ridderprint.nl
Cover design by Elisa Calamita, persoonlijkproefschrift.nl.
Layout and design by Elisa Calamita, persoonlijkproefschrift.nl.
© A.S. Duden 2020

Environmental impacts of bioenergy
Ex-ante spatially-explicit assessment of
biodiversity and hydrology impacts of
Ex-ante ruimtelijk expliciete analyse van de

impacts van verhoogde vraag naar biomassa op
biodiversiteit en hydrologie
(met een samenvatting in het Nederlands)
Proefschrift
Ter verkrijging van de graad van doctor

aan de Universiteit Utrecht
Op gezag van de Rector magnificus,
prof. dr. H.R.B.M. Kummeling,
Ingevolge het besluit van het college voor promoties
in het openbaar te verdedigen op
24 april 2020 des ochtends te 10.30 uur
door
Anna Sarah Duden
geboren op 9 juli 1986 te
Vlissingen

Promotoren
Prof. dr. A.P.C. Faaij
Prof. dr. H.M. Junginger
Copromotoren
dr. F. van der Hilst
dr. P.A. Verweij

Contents
Abbreviations 6
Chapter 1 Introduction 9
Chapter 2 Modelling the impacts of wood pellet demand on forest 25

dynamics in the southeastern United States
Chapter 3 Impact of increased wood pellet demand on biodiversity 55

in the southeastern US
Chapter 4 Biodiversity impacts of increased ethanol production in 85

Brazil
Chapter 5 Hydrological impacts of ethanol-driven sugarcane 109

expansion in Brazil
Chapter 6 Synthesis and conclusions 127
Annexes 149
References 254
Nederlandse samenvatting 274
Acknowledgements 279
About the author 281

Abbreviations
AID Area In Drought

AF Atlantic Forest
AL Alabama
AM Amazon
AR Arkansas
CA Caatinga
CE Cerrado
CGE Computable general equilibrium model
CV Coefficient of variation
CWC Centre West Cerrado region of Brazil
dLUC Direct land-use change
EU European Union
EU-28 Member states of the European Union
FIA USDA Forest Service Forest Inventory and Analysis program
FL Florida
FSC Forest Stewardship Council
GA Georgia
GAP United States Geological Survey National Gap Analysis Program
GAEZ Global Agro-Ecological Zones
GBEP Global Bioenergy Partnership
GCM General Circulation
GDP Gross Domestic Product
GHG Green House Gasses
GLCC USGS Global Land Cover Characterization dataset
GWCM Global Water Cycle Modelling
HhLp High housing demand and low wood pellet demand scenario
HhHp High housing demand and high wood pellet demand scenario
IEA International Energy Agency
iLUC Indirect land-use change
ISI-MIP Inter-Sectoral Impact Model Intercomparison Project
IUCN International Union for Conservation of Nature
KGE Kling-Gupta Efficiency
LA Louisiana
LAI Leaf Area Index
LhLp Low housing demand and low wood pellet demand scenario
LhHp Low housing demand and high wood pellet demand scenario

MAGNET Modular Applied GeNeral Equilibrium Tool
MIRCA2000 Monthly Irrigated and Rainfed Crop Area
Mt Million metric tonnes
Mha Million hectares
MS Mississippi
NA Northern Amazon region of Brazil
NC North Carolina
NEC North East Coast region of Brazil
NER North Eastern Cerrado region of Brazil
NGO Non-governmental organisation
NLCD National Land Cover Database
NRI USDA National Resources Inventory
OB Region outside of Brazil
OECD Organisation for Economic Co-operation and Development
OK Oklahoma
PCR-GLOBWB PCRaster GLOBal Water Balance model
PLUC PCRaster Land Use Change model
PM Pampas
PN Pantanal
RCP Representative Concentration Pathway
RED European Renewable Energy
RMSE Root Mean Square Error
RPA Resource Planning Act
RSPO Roundtable for Sustainable Palm Oil
S South region of Brazil
SC South Carolina
S.d. Standard deviation
SDE+ Stimulation of Sustainable Energy Production
SE South East region of Brazil
SSP Shared Socio‐economic reference Pathway
SRI Species Richness Index
SRTS SubRegional Timber Supply model
TN Tennessee
TPO USDA Forest Service Timber Product Output
TX Texas
US United States
USDA United States Department of Agriculture
USGS United States Geological Survey
VA Virginia

Environmental
impacts of bioenergy
Ex-ante spatially-explicit
assessment of biodiversity
and hydrology impacts of
Environmental
Anna Duden
Anna Duden

Environmental
1
Environmental
Introduction
Anna Duden
Anna Duden

Chapter 1
1.1 Risks and benefits of increased use of bioenergy
Global consumption of energy is projected to increase by over 25% by 2040 (IEA,

2019) due to expected economic and population growth. Climate change, driven by
greenhouse gas (GHG) emissions, has generated a vital need to reduce the use of fossil
fuels. The use of renewable energy has grown rapidly over the past decade, but fossil
fuels still make up over 80% of global energy demand (IEA, 2019). The share of modern
bioenergy in final energy consumption is projected to increase strongly (from 5%
in 2016 to 16% in 2050)(IRENA, 2019), which is considered an important strategy
for moving towards more sustainable energy supply systems (IPCC, 2011). Apart
from a potential reduction of GHG emissions and reduced dependency on fossil fuels,
bioenergy can offer broader benefits, including diversification of (energy) markets
and local socio-economic development through income and job generation (Creutzig,
2015; Dale, 2010). In the European Union (EU), the use of bioenergy is stimulated by
the Renewable Energy Directive (REDII), which includes sustainability criteria for
bioenergy production. Increased demand for bioenergy coincides with other global
trends resulting from population and economic growth, such as increased demand for
arable land and fresh water resources (IPCC, 2019). Over 40% of the world’s ice-free
land is now in agricultural or urban use (Díaz, 2019) while agriculture is responsible
for about 70% of global fresh water use (IPCC, 2019).
Despite the potential benefits of using bioenergy, a number of risks have received
ample attention in recent years in academic, societal and political debates. These risks
relate to potential socio-economic impacts of increased biomass production (Creutzig,
2015; Duarte, 2013), potential impacts on food security (Naylor, 2007; Souza, 2015), GHG
emissions (Fargione, 2008; Searchinger, 2008; Yassa, 2015), hydrology (Gerbens-Leenes,
2009; Pfister, 2011) and biodiversity (Koh, 2008; Souza, 2015). The majority of these
impacts are related to land-use change resulting from biomass feedstock production
for bioenergy. Land-use change can be both direct and indirect (Wicke, 2012). Direct
land-use change occurs when biomass production replaces a previous land use, such as
forest or pasture, while indirect land-use change occurs when the land use displaced
by biomass production in turn expands over other land-use types (Creutzig, 2015;
Verstegen, van der Hilst, 2016).
Reducing net GHG emissions is the key aim for bioenergy deployment, but bioenergy
systems may also have other environmental impacts, for example when land-use change
results in changes in biodiversity and hydrology (Creutzig, 2015; Díaz, 2019; Junginger,
2019). Freshwater availability is already limited in some areas, with over four billion
10

Introduction
people worldwide experiencing severe water scarcity (UNESCO, 2019). Fresh water
consumption is projected to grow by 20-30% by 2050 due to the growing demand for
food, feed, and fibre (UNESCO, 2019), increasing existing stress on freshwater resources
(Gerbens-Leenes, 2009; Molden, 2007). This is exacerbated by changing patterns of
water availability due to climate change (Molden, 2007). These pressures on water
resources have resulted in major environmental concerns worldwide (Pfister, 2011).
Increased demand for bioenergy may add to these pressures, for example, when land-
use change towards bioenergy crop cultivation changes evapotranspiration of the
1
vegetation, thereby negatively influencing the hydrological cycle (Berndes, 2008).
In some locations however, this can also result in positive impacts (Creutzig, 2015;
Gerssen-Gondelach, 2017).
Another type of environmental impact relates to the effects of land-use change on

biodiversity. The decline of biodiversity is occurring rapidly, especially in hotspots
of rare and endemic species, and about a million species of plants and animals are
currently threatened with extinction (Díaz, 2019). Species extinction rates are
currently up to 1,000 times higher than natural background extinction rates, and
are projected to increase to 10,000 times higher rates in the future (De Vos, 2015).
Land-use change is considered to have the largest impact on terrestrial ecosystems,
through habitat loss and degradation (Díaz, 2019), while pressures are exacerbated
by climate change (Thomas, 2004). Changes in land use, driven by increased demand
for bioenergy, will have an impact on habitat availability for a wide range of species
(Firbank, 2008). The specific impact of bioenergy driven land-use changes will differ
per species, with some species gaining and others losing suitable habitat (Evans, 2013;
Tarr, 2017). The impacts on biodiversity are location-specific and depend on the initial
land use and associated biodiversity, they therefore can be either positive or negative
(Creutzig, 2015; Dornburg, 2012; Gerssen-Gondelach, 2017; Immerzeel, 2014; Souza, 2015;
van der Hilst, 2012).
As a result of these sustainability concerns, a range of sustainability guidelines for

bioenergy has been put forward. These sustainability guidelines exist in different
forms. Government regulations can provide binding sustainability criteria for
the sourcing of biomass. An example is the REDII, which stipulates a number of
sustainability requirements for the use of bioenergy, particularly focusing on carbon
and biodiversity impacts (European Parliament and Council, 2018). Non-governmental
sustainability schemes on the sustainability of bioenergy include guidelines developed
specifically for bioenergy, e.g. the Global Bioenergy Partnership (GBEP, 2019), or
guidelines developed for specific commodities that are also applicable to bioenergy,
11

Chapter 1
e.g. the Roundtable for Sustainable Palm Oil (RSPO, 2019) or the Forest Stewardship
Council (FSC, 2019). Quantifiable sustainability criteria need to be formulated, and
then compared to ex-ante impacts assessments of bioenergy use and production, so
that negative impacts can be avoided. Although there is a general consensus on the
key areas of concern, there is less knowledge available on how to ex-ante quantify the
environmental impacts
1.2 Quantification of environmental impacts
Increased bioenergy demand puts additional pressure on natural resources already

under high pressure. Because these environmental impacts are strongly related to
location specific conditions, it is essential for the assessment of environmental impacts
of bioenergy to include spatially explicit projections of land-use changes (Souza,
2015; van der Hilst, Hoefnagels, 2018). In order to avoid potential negative impacts of
increased bioenergy demand, and promote positive effects, it is necessary to assess the
intensity and location of these impacts. This can be done through ex-ante modelling of
potential land-use changes and subsequent environmental impacts, using scenarios
of future developments in bioenergy. Bioenergy markets have strong interlinkages
with other markets, due to synergies or competition for feedstocks, land and other
production factors (Dale, Kline, 2017). An integrated approach that includes the link
between markets, land-use and environmental impacts is required to assess the
impacts of bioenergy demand (Dale, Kline, 2017). However, in order to quantify the
environmental impact of bioenergy, the impacts that can be attributed to bioenergy
should be separated from those related to developments in other sectors in order to
quantify bioenergy-specific impacts.
Apart from direct impacts of expansion of bioenergy production, indirect impacts

can occur. Competition and interaction with other land-uses and markets, as well as
biophysical conditions such as slopes, rainfall and soil type, will determine which land
can be used for bioenergy production and its subsequent potential environmental
impacts (van der Hilst, 2013; Verstegen, van der Hilst, 2016). Global and local drivers
will result in both direct and indirect changes in land-use and land management, and
determine which land-use types are replaced and by what land-use type they are
replaced (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). Assessments
of the impact of increased bioenergy demand should cover a geographical scale that
extends beyond the bioenergy sourcing or production area, to cover a sufficiently large
study area to include these indirect effects. Additionally, the availability of natural
12

Introduction
resources varies strongly geographically. Availability of fresh water for instance, is

not spread evenly, with some regions receiving ample rainfall, while other areas are
permanently or seasonally water restricted (Hoekstra, 2012). The same is true for
biodiversity, e.g. pockets of unusually high species richness occur in certain places
(Mittermeier, 2011; Myers, 2000, 2003). As a result, the impact of land-use change due
to increased bioenergy production will also vary across the landscape (Berndes, 2008).
Therefore, a spatial analysis of environmental impacts associated to land-use change
is required.
1
Summarising, an ex-ante assessment of environmental impacts of bioenergy should
be spatially explicit, assess bioenergy impacts in the context of the development of
other sectors, and be carried out at a spatial scale that allows inclusion of most indirect
effects of market interactions and land-use change. Studies that meet all of these
requirements are scarce. To date, limited research has accounted for spatial variability
in land-use dynamics following increased bioenergy demand, while a number of
studies assesses impacts of increased bioenergy demand in a spatially aggregated
way (Dornburg, 2012; Galik & Abt, 2015; Gerssen-Gondelach, 2017; Miner, 2014). Existing
spatially explicit studies are predominantly performed at a local scale, for example at
state level or regional level (Costanza, 2017; Georgescu, 2013; Hernandes, 2013; Loarie,
2011; Oliveira, 2014; Tarr, 2017; van der Hilst, 2012), for a feedstock sourcing area
(Evans, 2013) or river basin area (Costa, 2003; Hernandes, 2018; Pereira, 2013; Scarpare,
Hernandes, Ruiz-Corrêa, Picoli, 2016), and may therefore exclude indirect impacts of
bioenergy demand. Other studies are carried out for large regions, but with limited
spatial resolution (Louette, 2010; Prestemon, 2002). A number of studies have projected
environmental impacts of agricultural expansion in general (Faleiro, 2013; Jetz, 2007),
or more general pathways of global development and climate change (Schipper, 2019),
but have not separated impacts due to bioenergy. The increased use of bioenergy is
promoted, but requires ex-ante quantification of potential environmental effects to
avoid negative impacts. Previous studies have shown both benefits and risks of using
bioenergy, and have shown that environmental impacts are context and location
specific. Spatially explicit ex-ante quantifications of potential environmental impacts
that include indirect effects are therefore required to guide sustainable growth in
bioenergy use.
13

Chapter 1
1.3 Aim and research questions
In order to address the gaps in knowledge identified and described above, the aim of
this thesis is to quantify environmental impacts of increased bioenergy demand ex-
ante in a spatially explicit manner. More specifically, this thesis addresses the following
research questions:
I. How can environmental impacts of increased bioenergy demand be

quantified ex-ante?
II. What are the impacts of bioenergy driven land-use changes on biodiversity
and hydrology?
III. How do the impacts of increased bioenergy demand compare to the impacts
related to other agricultural and forestry products?
This thesis focuses on an ex-ante assessment of increased bioenergy demand on

hydrology and biodiversity at a regional to national level for the mid-term future
(up to 2030). Because bioenergy markets tend to be strongly interlinked with other
supply chains and markets, the impact of increased bioenergy demand is assessed
while taking into account developments in other agricultural and silvicultural markets.
The research questions have been applied to two different case studies; the production
of wood pellets in the southeastern United States, and the production of ethanol from
sugarcane in Brazil. These two regions are important biomass producing regions, in
which bioenergy production is expected to expand (K. L. Abt, 2014; Calderon, 2019;
IEA, 2018; Pinchot Institute, 2013; van der Hilst, Verstegen, 2018). Ethanol and wood
pellets are the bioenergy commodities that are most widely traded internationally.
Both Brazil and the southeastern US are rich in natural resources and biodiversity
and provide important environmental services (BPBES, 2019; Díaz, 2019; Evans, 2013;
Jenkins, Alves, 2015; Noss, 2015). Apart from representing some of the major bioenergy
production systems, the choice of these particular case studies also allows for a
focus on a supply chain related to a solid bioenergy type (wood pellets) as well as a
biofuel type (ethanol). Furthermore, the cases include a supply chain in the southern
hemisphere (Brazil) as well as in the northern hemisphere (United States), related to
agricultural feedstocks (ethanol) as well as forestry feedstocks (wood pellets). Finally,
Brazil has a long history of bioenergy production and use, while wood pellet production
is a relatively new bioenergy market in the US. The case studies, differing in the type
of feedstock and geographical location, were chosen so that the method developed to
assess environmental impacts may be more broadly applicable to different bioenergy
systems.
14

Introduction
In this thesis, land-use change is defined as a change of one land-use types to another
type, and includes changes in land management (for example, a conversion of natural
pine forest into pine plantation in Chapters 2 and 3), changes from one forest type to
another (for example, a change from natural pine forest to mixed forest in Chapters
2 and 3) and changes from one crop type to another (for example, a change from
sugarcane to another crop type in Chapters 4 and 5).
1
1.4 Case studies
1.4.1 Wood pellet production in the southeastern United States

Global demand for wood pellets has grown considerably over the last decades, and as
a result, production of wood pellets increased from 1.7 Mt in 2000 to 55.7 Mt in 2018
(Calderon, 2019). The United States (US) was the biggest producer of wood pellets in
2018 (FAO, 2019), producing 8.2 Mt of wood pellets (Calderon, 2019). About 98% of all
exported pellets from the US is produced in the southeastern region of the US (Wang,
2015a). Wood pellet production in the southeastern region is projected to increase in
the near future; of total announced wood pellet producing capacity to be built in the
US in the coming years, 81% is located within the southeastern US (K. L. Abt, 2014).
The southeastern US predominantly produces industrial wood pellets, which are

used for electricity production or combined heat-power production. The EU is the
main consumer of industrial wood pellets, responsible for about 50% of global pellet
consumption in 2018 (Calderon, 2019). There has been a strong increase in the use of
wood pellets for electricity production in the EU, reaching a pellet consumption of 9.3
Mt in 2018 (Calderon, 2019). The increase in wood pellet production in the US has mainly
been driven by increased demand for industrial wood pellets in the UK, Denmark and
Belgium (K. L. Abt, 2014; Calderon, 2019; Joudrey, 2012). Wood pellet demand in Europe
is driven by policies aimed at stimulating the use of renewable energy (K. L. Abt, 2014).
For example, the SDE+ program in the Netherlands provides subsidies for the use of
renewable energy. Other policies are aimed at increasing the sustainability of biomass
supply by imposing sustainability criteria to the harvesting or sourcing of biomass.
For example, the Dutch national sustainability criteria state that bioenergy cannot
be produced on land with high carbon or biodiversity value. Apart from improving
sustainability of bioenergy production, these sustainability criteria could also have
the unwanted consequence of restricting wood pellet supply (Galik, 2016). Although
wood pellet production in the southeastern US has increased strongly over the last
decade, it is still a relatively small market compared to conventional wood markets.
15

Chapter 1
The US is the largest producer of industrial roundwood in the world (Prestemon, 2015)
and within the US, the South has the largest forestry sector (Prestemon, 2002). A forest
plot usually produces a number of different wood feedstocks, including sawlogs, pulp
logs, low-value logs, harvest residues and, during processing of logs, mill residues.
These feedstocks can be processed into different products, including sawtimber, OSB,
paper and packaging, and wood pellets. Demand for wood pellets therefore interacts
with demand for other wood products. Currently, wood pellets in the US are made out
of the following feedstock types: 19% sawmill residues, 23% manufacturing residues,
35% other residues (including bark, logging residues and unmerchantable wood) and
23% roundwood (US EIA, 2019). The type and volume of wood products available, and
subsequent revenues and environmental effects, are influenced by forest management
and the interaction between the wood pellet market and other forestry markets (Dale,
Kline, 2017). Due to the strong linkages between different wood markets, demand
for wood pellets cannot be studied in isolation; an integrated analysis is needed that
includes the interactions between different wood markets.
Land use in the southeastern US is dynamic, and there is a continuous flux in the
landscape between forest and non-forest land uses (Miner, 2014; Wear, 2011).
The majority (86%) of forest area in the southeastern US is privately owned by
corporations and families, who are responsible for 96% of large roundwood harvesting
(Oswalt, 2014b; USDA FS, 2016b; Wear, 2013b). Changes in demand for wood product
feedstocks such as timber and pulpwood affect feedstock prices. Increased prices for
wood feedstocks in turn may stimulate landowners to increase investment in forestry
(Daigneault, 2012; Dale, Kline, 2017), resulting in more intensive management (Berndes,
2010; Cintas, 2017; Olesen, 2016; Plantinga, 1993; Schulze, 2012) or an increase in forest
establishment (Hardie, 2000; Lubowski, 2008; Wang, 2015a). Increased demand for
wood products, including wood pellets, has been projected to lead to a change in forest
area (R. C. Abt, 2009a, 2013a) and a shift from natural forest to pine plantation (Evans,
2013). An increased demand for wood pellets may thereby exacerbate the existing
trend of loss of natural forest, while at the same time promote the establishment of pine
plantations (R. C. Abt, 2009b, 2013b; Evans, 2013). While high-value sawtimber and pulp
markets are expected to continue driving major forest management decisions, wood
pellet demand may contribute to increased feedstock prices and thereby influence
land-owner decisions (Dale, Kline, 2017). The resulting land-use and management
changes may result in indirect land-use changes.
Urbanization has been projected to be the largest driver of deforestation in the

southeastern US in the near future (Wear, 2013b). When strong markets for forest
16

Introduction
products motivate private forest owners to keep their land in forest cover, it may
shift the expansion of urban areas towards agricultural land rather than forests
(Wear, 2013b). Southern pine plantations are among the most intensively managed
forests (Fox, 2007). Shifts in forest area and management may have environmental
impact, for example on maintenance of carbon stocks and biodiversity conservation
(Chen, 2006; Evans, 2013; Tarr, 2016; Zhu, 2010). As the demand for wood pellets
from the southeastern US has grown, there is a growing debate about the potential
environmental effects of wood pellet production (Dale, Kline, 2017; Parish, 2018).
1
Concerns about increasing wood pellet production have been raised for several reasons
including its potential contribution to changes in land use and forest management
(Costanza, 2015, 2017; Prestemon, 2002) and subsequent impacts on biodiversity
(Evans, 2013; NRDC, 2015; Olesen, 2016; Pelkmans, 2014; Tarr, 2016) and greenhouse gas
emissions (Booth, 2018; Colnes, 2012; Junginger, 2019; Yassa, 2015). This thesis focuses
on projected land-use changes in the southeastern US as a result of increased wood
pellet demand (Chapter 2), and on subsequent impacts on biodiversity (Chapter 3).
The southeastern US has a large variation in terrain, soil types, and a range of
precipitation and temperature patterns, supporting forests with high primary
production and rich biodiversity (Evans, 2013). The region has high concentrations of
endemic species (species that occur only in this region) ( Jenkins, Van Houtan, 2015),
with 43% of all amphibian species, 28% of reptile species, 2% of breeding bird species
and 6% of mammal species being endemic to the region (Noss, 2015). The North
Atlantic Coastal Plain (a region within the southeastern US) has been identified as a
global biodiversity hotspot (Noss, 2015). Over the last two centuries, species habitats
in the region have been severely altered due to land-use and land management change,
driven by logging of most primary forest by initial settlers, suppression of natural fire
patterns, agricultural expansion and contraction, a shift from natural forest to pine
plantation forestry, suburban development, and spread of invasive species (Evans,
2013). Despite these major changes, today’s vast forests in the southeastern US still
maintain populations of the majority of native species (Wear, 2013a). However, a shift
from natural forest to plantation forests has been identified as a major factor of habitat
loss for a range of specialized forest species, including species of conservation concern
(Evans, 2013). While pine plantations can provide habitat for some species, a change in
forest management can result in changes in structural diversity and habitat suitability,
the impact of which will vary across species (Rupp, 2012). The potential impacts of
increased demand for wood pellets from the southeastern US on biodiversity have been
identified as one of the major EU policy risks regarding wood pellet imports (Olesen,
2016). The study of these impacts is considered as one of the 40 scientific priorities to
17

Chapter 1
inform US conservation managers and policy makers (Fleishman, 2011). A number of

studies have assessed the potential impact of wood pellet production on biodiversity in
the southeastern US using a spatially explicit approach (Evans, 2013; Galik & Abt, 2015;
Tarr, 2016). Spatially explicit studies on the impacts of woody biomass production on
biodiversity in the southeastern US have assessed the impacts for certain restricted
areas in the wood pellet sourcing regions (Evans, 2013; Galik, 2016; Tarr, 2016), thereby
excluding indirect effects beyond those sourcing areas.
1.4.2 Ethanol production in Brazil

The agricultural and forestry sectors in Brazil are amongst the largest in the world
(FAO, 2019; OECD, 2015) and Brazil is the world’s third largest exporter of agricultural
products (OECD, 2015). The most important crops produced in Brazil (in terms of area
and net production value) are sugarcane, soybean and corn (FAO, 2019), the area of
cultivation of these crops has increased strongly over the last three decades (FAO, 2019;
OECD, 2015). The agricultural output of Brazil has doubled between 1990 and 2015,
while livestock production tripled, partly due to increased productivity, but also due
to an expansion of agricultural land of 34 Mha between 1990 and 2012 (OECD, 2015).
Crop production in Brazil has been projected to continue to grow in the coming years
(OECD, 2015).
Brazil currently has the largest area of sugarcane cultivation in the world (FAO,
2019), and is the second largest producer of ethanol (IEA, 2018). Sugarcane in Brazil
is cultivated both for sugar and ethanol production, with the sugar market using 36%,
and the ethanol market 64% of total sucrose content from sugarcane in 2018 (USDA
Foreign Agricultural Service, 2019b). The largest share of Brazil’s ethanol production is
destined for the domestic market (IEA, 2018; USDA Foreign Agricultural Service, 2019a).
In Brazil, ethanol is available as a fuel either unblended or blended with fossil transport
fuels (IEA, 2018), and biofuel made up a 20% share of transport fuel in Brazil in 2017
(IEA, 2018). The strong position of ethanol in the Brazilian fuel market is the result of a
long history of supportive policies that stimulated sugarcane-based ethanol production
to increase energy security, as well as promote rural development and decrease the
dependency on fossil fuels (OECD, 2015; Verdade, 2012). Ethanol production increased
from 28 to 34 billion litres between 2010 and 2019 (UNICA, 2019; USDA Foreign
Agricultural Service, 2019a). Ethanol production in Brazil is projected to continue to
increase in the future, in part due to the planned implementation of the RenovaBio
policy framework in 2020 (IEA, 2018). This policy will, amongst other things, raise
the domestic blending target by 2030 (OECD, 2015), which is expected to result in
an increase in demand, investment, capacity and production of ethanol (IEA, 2018).
18

Introduction
As a result, Brazil is projected to show the strongest growth in biofuel production

worldwide between 2018 and 2023 (IEA, 2018); Brazilian ethanol production is
projected to increase to 36.5 billion L by 2023 (IEA, 2018) and 54.2 billion L in 2030
(van der Hilst, Verstegen, 2018).
The projected increase in ethanol production is expected to be realized through

an increase in yield as well as through the expansion of sugarcane cultivation area.
Sugarcane area in Brazil is projected to increase from 10.1 Mha in 2012 to 13.6 Mha
1
in 2030 (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). The projected
additional 35 thousand km2 of sugarcane production area will result in direct land-use
changes (when sugarcane replaces an area previously occupied by another land use)
and indirect land-use changes (when the expansion of sugarcane induces changes in
land use elsewhere) (Adami, 2012; Lapola, 2010; van der Hilst, Verstegen, 2018; Verstegen,
van der Hilst, 2016). Brazil has an abundance of natural resources, and these resources
are vital to the resilience of the Brazilian agriculture, energy, forestry and fisheries
sectors (MMA, 2015). However, increasing pressure on natural assets may result in
increased environmental impacts. Satisfying the increased demand for food and
biomass while minimizing the impacts of agricultural production is therefore a huge
challenge for the future (Pfister, 2011).
Brazil is recognized as one of the most biodiverse countries in the world (IPBES, 2018;
Jenkins, Alves, 2015) and contains about 15-20% of all species worldwide (BPBES, 2019).
Of the over 115,000 species found in Brazil that have been described by science, at
least 421 are known to be threatened with extinction (IUCN, 2019). The Cerrado and
Atlantic Forest biomes are both biodiversity hotspots (Myers, 2000) under severe
threat (Hunke, 2015; Sano, 2010). Natural vegetation in the Cerrado biome has been
reduced dramatically in the last few decades due to agricultural expansion, to the point
that over 40% of its original extent has been converted to agricultural use (Hunke,
2015; Klink, 2005; Zuurbier, 2008). Only about 2% of the remaining natural vegetation
in the Cerrado biome is legally protected (Carvalho, 2009; Klink, 2005). The Cerrado is
home to 16 mammal, 29 bird, 35 reptile and 47 amphibian species that are endemic
to the region (Diniz-Filho, 2008). The tropical and semi-deciduous Atlantic Forest is
home to a large number of endemic species ( Jenkins, Alves, 2015), but only about 12%
of its original extent remains (Beca, 2017; de Abreu Bovo, 2018). The remaining areas of
Atlantic Forest are highly fragmented (Beca, 2017; de Abreu Bovo, 2018). Furthermore,
Brazil contains almost 70% of the Amazon rainforest biome (Kirby, 2006). The Brazilian
Amazon has lost about 18% of its original forest area and has recorded the highest
levels of deforestation in 2017/2018 since 2008 (MMA, 2018).
19

Chapter 1
Expansion and intensification of agriculture have been identified as the main drivers
of biodiversity loss, and projected ongoing agricultural expansion is expected to
exacerbate the pressure on biodiversity (Norris, 2008). The projected expansion of
sugarcane is projected to lead directly or indirectly to habitat loss (van der Hilst,
Verstegen, 2018), and concerns have been raised on the potential impact of sugarcane
expansion on biodiversity (Naylor, 2007). A number of studies have projected
substantial future biodiversity impacts of agricultural expansion in Brazil (Faleiro,
2013; Jetz, 2007; Kennedy, 2016; Strassburg, 2017). However, it remains unclear which
proportion of these impacts can be attributed specifically to an increase in ethanol
demand, and how these biodiversity impacts vary over different regions in Brazil.
Analysis of the impacts of ethanol-driven sugarcane expansion on biodiversity has
been identified as a priority for future research (Verdade, 2012), in view of the paucity
of studies that assess the potential impacts of sugarcane expansion on biodiversity in
Brazil (Walter, 2014). This thesis focuses on biodiversity impacts in Brazil of increased
ethanol demand in Chapter 4.
Brazil has abundant freshwater resources (Martinelli, 2008) but water availability is
unevenly distributed (Flach, 2016; Hernandes, 2013; Moreira, 2007), and several regions
in Brazil are facing water scarcity problems (Scarpare, Hernandes, Ruiz-Corrêa, Kolln,
2016). The southeast of Brazil, where the majority of sugarcane production is located,
was hit by a severe drought in 2014 and 2015. This drought put the water supply of the
São Paulo metropolitan area at risk. São Paulo is South America’s largest metropolis
and home to over 20 million people, as well as the location of the generation of a third of
Brazil’s gross domestic product (GDP) (Nobre, 2016). The rapid development of the São
Paulo metropolitan region over the last few decades has increased its vulnerability to
drought by increasing water demand and through destruction of the natural systems
that aided in the recharge of water reservoirs (Nobre, 2016). Reduced rainfall and high
temperatures in 2014 and 2015 resulted in dangerously low storage levels in the main
water reservoirs in the region, jeopardizing the water supply of 28 million people
(Melo, 2016; Nobre, 2016). This drought had severe effects: it reduced agricultural
production and resulted in a loss of about US$ 5 billion to the agricultural sector, it
increased the number of forest fires by 150%, it increased electricity prices due to
reduced energy production of hydro-power plants, and hampered the functioning of
schools, hospitals and businesses due to lack of water (Nobre, 2016).
Increased demand for Brazilian sugarcane may increase the existing pressure on
water resources through changes in evapotranspiration and a possible increase in
water needs for irrigation (Fachinelli, 2015). The potential negative impact on water
20

Introduction
resources has been identified as one of the key sustainability challenges related to
sugarcane production in Brazil (de Cerqueira Leite, 2009; Guarenghi, 2016). A number
of spatially explicit studies have assessed the hydrological impacts of sugarcane
expansion in Brazil. However, by focussing on a single river basin (Costa, 2003; Pereira,
2013; Scarpare, Hernandes, Ruiz-Corrêa, Picoli, 2016) or a particular region in Brazil
(Georgescu, 2013; Hernandes, 2013; Loarie, 2011; Oliveira, 2014), these studies do not
include the potential effects of indirect land-use change due to sugarcane expansion.
Chapter 5 of this thesis focuses on hydrological impacts of increased ethanol demand
1
in Brazil.
1.5 Thesis outline
The research questions described in paragraph 1.3 are addressed in chapters 2 to 5.

Research question I, which focuses on the methodological challenges of quantifying
environmental impacts of increased bioenergy demand, is addressed for two different
types of environmental impacts (biodiversity and hydrology) in two different case
studies (the US and Brazil). Research question II, aimed at quantifying environmental
impacts of increased bioenergy demand, is also applied to both case studies (the US and
Brazil) for two types of environmental impacts (on biodiversity and water). Finally,
research question III compares the projected environmental impact of increased
bioenergy demand to the projected impact of increased demand for other products
produced in the same landscape, this is done for the US and the Brazil case study. Table
1.1 provides an overview of which research questions, case studies and environmental
impacts are addressed in each chapter.
Chapter 2 uses a combined modelling approach to assess land-use changes in the

southeastern US following projections of wood pellet demand up to 2030 (addressing
research question I), and subsequently compares the land-use changes due to increased
wood pellet demand to scenarios of timber demand, as well as demand for other
agricultural and forestry products (addressing research question III). Chapter 3 builds
on the results of Chapter 2 and provides a new approach for the spatial assessment
of environmental impacts of land-use change (addressing research question I). This
method is applied to the assessment of biodiversity impacts of the land-use change
projections in the southeastern US for different scenarios of wood pellet demand
developed in Chapter 2, thereby addressing research question II. The biodiversity
impacts of increased wood pellet demand are then compared to scenarios of timber
demand, as well as demand for other agricultural and forestry products (addressing
21

Chapter 1
research question III), in line with Chapter 2. Chapter 4 presents the application of
the method presented in Chapter 3 to a different case study: the production of ethanol
from sugarcane in Brazil. Chapter 4 assesses the biodiversity impacts of increased
ethanol demand and subsequent sugarcane expansion in Brazil (addressing research
question II). The land-use projections used in this chapter are based on previous
research (van der Hilst, Verstegen, and Woltjer 2018; Verstegen et al. 2016). In Chapter
4, the impacts due to increased ethanol demand are compared to a reference scenario
based on increased demand for other (agricultural) land uses (addressing research
question III). Chapter 5 adapts an existing hydrological model and couples it to a model
of projected sugarcane expansion and other land-use change in Brazil (addressing
research question II) to allow for the assessment of the hydrological impacts of
increased ethanol demand (addressing research question I).
Table 1.1: Overview of the thesis chapters, and the topics, research questions, case studies and
environmental impacts they address.
Research
Case Environmental
Chapter Chapter title question
study impact
I II III
Modelling the impacts
of wood pellet demand
2 on forest dynamics in x x US x
southeastern United
States
Impact of increased
wood pellet demand
3 x x x US Biodiversity
on biodiversity in the
southeastern US
Biodiversity impacts
4 of increased ethanol x x Brazil Biodiversity
production in Brazil
Hydrological impacts of
5 ethanol-driven sugarcane x x Brazil Hydrology
expansion in Brazil
22

Introduction
23

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
2
Modelling the impacts of wood
and hydrology
pellet demand impacts
on forestof
dynamics in
increased biomass
southeastern Environmental
demand
United States
Anna Duden
A.S. Duden, P.A. Verweij, H.M. Junginger, R.C. Abt, J.D. Henderson, V.H. Dale, K.L. Kline, D.
Karssenberg, J.A. Verstegen, Andre Faaij, F. van der Hilst
assessment
This chapter has been published in the of biodiversity
journal Biofuels, Bioproducts and Biorefining.

Anna Duden

Chapter 2
Abstract
The export of wood pellets from the southeastern United States (US) has grown
significantly in recent years, following rising demand from Europe. Increased wood
pellet demand could lead to spatially variable changes in timberland management
and area in the US. This study presents a modelling approach to assess the effects of
increasing wood pellet demand (an additional 11.6 Mt by 2030) on land-use dynamics,
taking into account developments in other wood product markets as well as expected
changes in other land uses. An economic model for the forest sector of the southeastern
US (SRTS) was linked to a land-use change model (PLUC) to identify potential locations
of land-use change following scenarios of demand for pellets and other wood products.
Projections show that in the absence of additional demand for wood pellets, natural
timberland area is projected to decline by 450-15,000 km2 by 2030, mainly through
urbanization and pine plantation establishment. Under the high wood pellet demand
scenario, more (2,000-7,500 km2) natural timberland area is retained and more (8,000-
20,000 km2) pine plantation is established. Shifts from natural timberland to pine
plantation occur predominantly in the Atlantic coastal region. Future work will aim to
assess the effect of projected transitions in natural timberland and pine plantations on
biodiversity and carbon storage. This modelling framework can be applied for multiple
scenarios and land-use projections to identify locations of timberland area changes
for the whole southeastern US, thereby informing the debate about potential effects
of wood pellet demand on land-use dynamics and environmental services.
26

Modelling the impacts of wood pellet demand on forest dynamics in southeastern United States
2.1 Introduction
Global wood pellet production has grown considerably over the last decade; from 1.7
Mt in 2000 to 29.7 Mt in 2015 (European Biomass Association, 2016). The European
Union (EU) is the main user of wood pellets, responsible for about 80% of global pellet
consumption in 2015 (European Biomass Association, 2016). Specifically, there has
been a strong increase in the use of wood pellets for electricity production in the
EU-28, reaching a pellet consumption of 7.3 Mt in 2015 (European Biomass Association,
2016) with about 4.6 Mt (63%) being imported from the southeastern United States
(US) (Dale et al., in press). Wood pellet production has risen strongly in the US in
2
recent years, mainly driven by an increase in European demand for wood pellets
( Joudrey, 2012). As a result, US exports to the EU have increased (Lamers, 2012) and
the US is now the main supplier of wood pellets to Europe (FAO, 2015). The southern
part of the US produces 98% of exported pellets from the US (Wang, 2015b). Future
increases in pellet production in this region are expected: of total announced wood
pellet producing capacity to be built in the US in the coming years, 81% is located
within the southeastern US (K. L. Abt, 2014). However, future development in pellet
production in the region is uncertain; projections range from around 5.6 Mt (Pinchot
Institute, 2013) to 16 Mt (K. L. Abt, 2014) pellet production in 2020. The feedstock
used to produce wood pellets originates from timberland, which is comprised of two
subsets: natural timberland1 and planted timberland (pine plantation)2.
An increased demand for wood pellets from the southeast of the US could lead to
changes in both timberland area and management in this region. Changes in demand
for wood product feedstocks such as timber (large roundwood) and pulpwood (small
roundwood, which is also the main feedstock for wood pellet production (Galik, 2016;
Stephenson, 2014a)) affect feedstock prices. High prices for wood feedstocks can lead
to an increase in timberland establishment (Lubowski, 2008; Wang, 2015b). Strong
markets for forest products can motivate private forest owners to keep their land
in forest cover, which may shift the expansion of urban areas – projected to be the
1 Natural timberland is defined as ‘Productive forests composed of trees established by natural

regeneration of existing seed sources, root suckers, stump sprouts, etc. Establishment may be
either afforestation on land that until then was not classified as forest or by reforestation of land
classified as forest after a disturbance or following harvest’ following the Forest Inventory and
Analysis National Program (see for instance (Oswalt, 2014b)).
2 Planted timberland is defined as ‘Productive forests composed of trees established through plant-
ing and/or seeding of native or introduced species. Establishment may be either afforestation
on land that until then was not classified as forest, or by reforestation of land classified as forest
after a disturbance or following harvest’ following the Forest Inventory and Analysis National
Program (see for instance (Oswalt, 2014b)).
27

Chapter 2
largest driver of deforestation in this region in the near future – towards agricultural
land rather than timberland (Wear, 2013b). Shifts in prices have also been shown to
lead to changes in forest management. High prices for small roundwood promote
forest investment and increased management intensity in terms of shorter rotations
(Plantinga, 1993; Schulze, 2012), higher density planting, increased thinning practices
and fertilizer use (Berndes, 2010). Currently, southern pine plantations are among the
most intensively managed forests in the world (Fox, 2007). The largest share (86%) of
timberland in the southeastern US (both natural and planted) is privately owned by
corporations and families, who are responsible for 96% of large roundwood harvesting
(USDA FS, 2016b; Wear, 2013b). Market forces can influence private landowner decisions
on forest management and harvest (Aguilar, 2014; Conrad, 2011; Young, 2015).
Changes in timberland area and management can have an effect on the provision
of ecosystem services, such as biodiversity conservation and carbon storage. The
increased production of wood pellets has fuelled debate about its effects on potential
land-use change (e.g. (Costanza, 2015, 2017; Prestemon, 2002)) and its implications for
biodiversity (e.g. (Cubbage, 2006; Evans, 2013; Pelkmans, 2014; Tarr, 2016)) and carbon
stocks (e.g. (Hanssen, 2017; Jonker, 2014; Meyer, 2015; Olesen, 2016; Pinchot Institute,
2013)). As a result of uncertainty about the sustainability of the use of wood pellets,
several European countries have put forward sustainability guidelines (Lamers, 2015)
which could restrict the supply of wood pellets from the US to Europe (Galik, 2016). The
South is thought to be the largest carbon sink across the conterminous United States
(Chen, 2006). Changes in forest area determine variation in carbon storage across
the landscape. Forest management also influences carbon stocks, as was shown for
different plantation management strategies in the southeastern US ( Jonker, 2018).
Additionally, while natural forests generally provide a more suitable habitat for
a wider range of species than pine plantations, pine plantation establishment can
provide habitat for a number of forest species and increase connectivity and landscape
diversity (Brockerhoff, 2008; Miller, 2009). Net benefits of managed timberlands depend
on the reference system and the assumed alternative management practices. These
effects might not be distributed evenly over the region (Olesen, 2016). Historical trends
in land-use change vary across the landscape in the southeastern US (Brown, 2005).
Timberland area and management dynamics in the southeastern US are expected
to be non-uniform due to the variation in spatially explicit characteristics such as
potential agricultural yields, water availability and urbanization pressure (Drummond,
2010). Therefore, changes in timberland area and management intensity, as well as
subsequent environmental impacts are expected to be location specific. In order to
answer questions related to the sustainability of pellet production in the southeastern
28

US, it is necessary to take into account spatial variability. To date, limited research
has taken into account spatial variability in land-use dynamics following increased
pellet demand for the whole southeastern region. Previous studies on the effect of
wood pellet demand on land use have provided insights into potential changes in land
use, but were mostly not spatially explicit (e.g. (K. L. Abt, 2012, 2014; R. C. Abt, 2009b;
Galik, 2016; Wang, 2015b)). Those analyses that were spatially explicit were conducted
either at low resolution (state level) on a large scale (several states in the southeast)
(Prestemon, 2002), or provided detailed information of high resolution (100 m or less)
but on a small local scale (woodshed (Evans, 2013) and state (Costanza, 2017)).
2
The aim of this research is to create a spatial assessment of the effect of increasing
wood pellet demand on land-use dynamics, while taking into account demand for
other wood products (i.e. saw timber and pulp and paper), as well as development
of other land uses (e.g. urban land and cropland). This study does not quantify the
environmental impacts of wood pellet production, but provides an important first step
for environmental impact assessment by providing projections of potential land-use
change. By identifying where and how much timberland area is projected to change
under different pellet demand scenarios, we can inform the debate about the potential
effects of wood pellet demand on land-use dynamics.
2.2 Methods
The southeastern US is defined as in previous studies of the forestry sector in the

area (R. C. Abt, 2000, 2009b; Galik & Abt, 2015; Galik, Abt, 2015) and includes the
states of Florida, Georgia, South Carolina, North Carolina, Virginia, Tennessee,
Alabama, Mississippi, Louisiana and Arkansas, as well as the western parts of the
states of Texas and Oklahoma. The landscape patterns in the southeastern US today
were created through large scale deforestation for agriculture in the 18th and 19th
centuries, followed by land abandonment and subsequent forest regeneration in the
late 19th–early 20th century (Taverna, 2005). The area of natural timberland in the
southeastern US decreased by almost 20,000 km² between 1990 and 2010, while the
area of pine plantation increased by more than 50,000 km² (USDA FS, 2016b). These
land-use changes have been attributed to the driving forces of urbanization, increased
commercial forestry and competition with agriculture (Napton, 2010). Between 2001
and 2011, urbanization resulted in over 7,000 km2 additional developed land in the
southeastern US, while agricultural land decreased by over 2,000 km2 and timberland
decreased by almost 19.000 km2 (Homer, 2007, 2015).
29

Chapter 2
A modelling approach was applied to assess potential future changes in land use in the
southeast of the US resulting from an increasing wood pellet demand between 2010
and 2030 and taking into account developments in other wood markets (i.e. saw timber
and paper products) and dynamics of other land uses. The methodology consisted of
the combination of a partial-equilibrium economic wood market model (SRTS) (R. C.
Abt, 2000; Galik, 2009b) and a spatial land-use change model (PLUC) (Van der Hilst,
2012; Verstegen, 2012). The wood market model determined the volume and area of
timberland needed to satisfy an exogenously set demand scenario for wood products.
The land-use model spatially allocated the projected changes in timberland area, as
well as changes in other land uses. The development in demand for non-forest land (e.g.
urbanization, agriculture) was exogenously set, based on the 2010 Resources Planning
Act Assessment of the US Forest Service (USDA FS, 2012) and included in the PLUC
model. Figure 2.1 provides an overview of the modelling approach. The spatial model
generates maps of annual projected land use between 2010 and 2030 at a 2-by-2-km
resolution. These maps can be used to model and project environmental impacts in a
spatially explicit way for the entire southeastern region of the US.
Wood market analysis Spatial land-use analysis

Input Model Output Input Model Output
Forest
Land use map
inventory data Price,
inventory, Suitability
Definition of
removal data factors
feedstock types
Timberland area projections for 5 forest Projected land
Demand SRTS types PLUC use changes
scenario forest
products Other land use
Demand and projections
price
elasticities Allocation
order
Figure 2.1: Overview of the methodology, showing a combination of the SRTS (in the component
‘wood market analysis’) and the PLUC model (in the component ‘spatial land use analysis’). For each
model, input data and output data are listed. Forest area projections are both an output of the SRTS
model and an input to the PLUC model.
Scenarios
Demand from different markets for different wood feedstocks is divided into
softwood (from pine and mixed forests) and hardwood (from mixed forests, upland
and lowland hardwood forests), as well as large (timber) and small roundwood
(pulpwood). Between 2006 and 2011 in the US, 19-25% of wood harvested was used
as sawtimber, 49-55% was used by the pulp and paper sector, 15-17% was used
to produce wood based panels and 9-11% of wood harvested was used for energy
30

production (Stephenson, 2014b). Because an acre of timberland generally contains both

small and large roundwood, harvests of small and large roundwood are often linked.
The future development of wood markets is uncertain, therefore, four scenarios for
future demand for woody biomass were constructed to assess the consequences of
potential future market development on land use. The wood pellet sector is strongly
influenced by developments in other larger wood using sectors, such as the saw timber
and pulp and paper markets. The scenarios therefore included projections for wood
pellet markets in conjunction with pulp and paper markets and the domestic housing
market in the US.
2
A scenario of relatively high future pellet demand was derived from projections
of future EU imports of wood pellets (Cocchi, 2011) and the proportion of imports
originating from the US (Boie, 2016). This scenario assumed a gradual increase in
pellet demand from the southeast of the US, reaching a plateau of an additional 11.6
Mt in 2020 from a starting point of about 0.5 Mt in 2010 (based on (K. L. Abt, 2014;
Poyry, 2014)). This scenario was compared to a scenario with no increased demand
for wood pellets from 2010. The demand for wood pellets was translated into demand
for small roundwood3. Two scenarios for rates of recovery of the domestic housing
market, which crashed in 2008, were constructed based on Ince & Nepal (Ince, 2012).
All scenarios include a slowly growing demand for small roundwood from the pulp and
paper sector of under 2% a year up to 2030, also based on Ince & Nepal (Ince, 2012).
Figure 2.2 provides an overview of demand volumes in the different scenarios (see
also Annex 1). The two scenarios for the wood pellet market and the two scenarios
for the US domestic housing market were combined to formulate 4 scenarios based
on a combination of demand for small roundwood (from the pellet and paper sector)
and large roundwood (from the housing sector, see Figure 2.3): HhHp for high housing
and high pellet demand, HhLp for high housing and low pellet demand, LhHp for low
housing and high pellet demand and LhLp for low housing and low pellet demand.
3 This was done by assigning 20% of demand for pellets to hardwood small roundwood feedstock,
and 80% to softwood small roundwood feedstock. The small roundwood category also includes
15% and 30% (for softwood and hardwood respectively) harvest residues from the large round-
wood category.
31

Chapter 2
Demand scenarios for different feedstocks

180
160
Feedstock demand (Mt) 140
120
100
80
60
40
20
0
2010 2020 2030
SwSr - high SwSr - low SwLr - high SwLr - low

HwSr - high HwSr - low HwLr
Figure 2.2: Overview of demand scenarios for the different feedstocks, with demand in tons per year
between 2010 and 2030. The changes between 2010 and 2011 reflect the TPO adjustment (see Annex
2). SwSr = Softwood small roundwood, SwLr = Softwood large roundwood, HwSr = Hardwood small
roundwood, HwLr = Hardwood small roundwood.
+85 *106 t
High
Housing
HhLp HhHp
scenario scenario
Pellets Pellets
+0 t +11.6 *106 t
Low High
LhLp LhHp
Housing
scenario scenario
Low
+24 *106 t
Figure 2.3: Overview of scenarios used in the analysis. ‘High housing’ stands for a high demand for
large roundwood. The volumes presented on either side of the vertical arrow depict the additional
demand for large roundwood in 2030. ‘High pellets’ stands for a high demand for small roundwood.
The volumes presented on either side of the horizontal arrow depict the additional demand for
small roundwood in 2030.
Wood market analysis

The influence of the demand for different wood products on the area of timberland
was assessed using an economic model of the forestry market in the US South: the
Sub Regional Timber Supply (SRTS) model. The SRTS model was developed by Abt et
al. (R. C. Abt, 2000; Galik, 2009b) to assess timber supply and demand in the US South
32

on a medium timescale (5-25 years). SRTS is a partial equilibrium model, i.e. it takes
into consideration one sector (in this case the forestry sector, which contains several
markets), but not the dynamics in other sectors. About 14% of total timberland in the
area is found on public lands (USDA FS, 2016b). Harvest decisions on public lands (such
as national forests) are not driven by market forces, and only 4% of timber harvest
takes place on public lands (USDA FS, 2017a). Therefore public lands were excluded
from the analysis (as in (R. C. Abt, 2000; Galik, 2009a; Wear, 2013b)). Timberland area
included in the analysis follows the USDA definition of timberland: ‘forest land that is
producing or is capable of producing crops of industrial wood’ (USDA FS, 2016a). SRTS
was run in demand mode, which means that harvest, price and forest areas respond
2
endogenously to an exogenous development in demand.
SRTS simulates the economic market for wood products based on Forest Inventory &
Analysis (FIA) inventory data, definitions of feedstock types and demand scenarios
for forest products. The SRTS model calculates timberland dynamics for spatial units
with the size of a FIA survey unit level, which consists of a number of counties and is on
average 25,000 km2 in size (area ranges from 11,296 to 48,558 km2). Forest inventory
data used in SRTS includes data on area per forest type, stand age and removals and
is based on FIA (USDA FS, 2016b) and Timber Product Output (TPO) data (USDA FS,
2017b) (see Annex 2 for a more detailed explanation of TPO adjustment). SRTS also
requires a definition of feedstock types, which defines the class boundary between
small and large roundwood. It also defines a percentage of large roundwood that
ends up in the small roundwood pool, such as tops and branches4. Finally, elasticity
values are required to calculate price responses of the feedstock types. The output of
SRTS includes projections of large and small roundwood price, forest inventory and
removals, as well as expected timberland area for five different forest types: pine
plantation, natural pine, mixed forest, upland hardwood and lowland hardwood.
SRTS was run with a time step of 1 year from 2010 to 2030. We used the version of
SRTS based on FIA data for start year 2010 (SRTS version 28b). In SRTS, changes in
timberland area are calculated at each time step for five different forest types. An
overview of the processes in SRTS is shown in Figure 2.4. Input data on demand from
the saw timber, pulp and paper and wood pellet markets are compared to inventory
data. At the first time step, inventory data from the FIA are used as input. In subsequent
time steps, these inventory data are adapted based on calculated removals (item 1 in
Figure 2.4) and regrowth (2). Regrowth is a slow process and therefore has a time
4 This percentage was 15% for softwood and 30% for hardwood.
33

Chapter 2
lag in the model. Changes in timberland inventory trigger changes in small and large
roundwood prices based on elasticity values (3). Small and large roundwood prices
are used to calculate timber land rents based on a Net Present Value calculation5 (4).
Timber rents are compared to agricultural rents to calculate the area per forest type
that is projected to be established or lost (5). Price responsiveness is assumed to differ
per forest type, with pine plantations being most price responsive6. The projected
changes in timberland area affect the inventory of small and large roundwood; in the
case of timberland area expansion, this process occurs with a time lag (6). SRTS creates
yearly projections of timberland area per forest type, which were used as land-use
projections in a spatial land-use analysis.
Spatial land use analysis

Future land use was projected spatially using the PCRaster Land Use Change (PLUC)
model. PLUC is a spatial land-use change model developed by Verstegen et al. (Van der
Hilst, 2012; Verstegen, 2012) and often applied in case studies on land-use dynamics
and bioenergy (e.g. (Diogo, 2014; van der Hilst, 2014; Van der Hilst, 2012)). PLUC
spatially allocates land-use transitions per time step based on current land use, yearly
projections of demand for different land uses for the region, the suitability of land for
different land uses and the allocation order. See also earlier work by Verstegen et al.
for a more detailed description of PLUC (Verstegen, 2012). In PLUC, land-use classes
are divided into dynamic, passive and static. Dynamic land-use classes can actively
expand or contract based on changes in relative demand, leading to a net change in area
for competing dynamic land classes. Passive land-use classes only change as a result
of changes in the area of dynamic land-use categories. Static land uses do not change
over time. The allocation order determines in which order land uses get assigned a
location of expansion or contraction.
5 Large roundwood starts off 4x the price of small roundwood, and is discounted at 4% for 10 years.
6 Pine plantation was assumed to be 2.5 more price responsive than other forest types. Lowland
hardwood was assumed to be 0.5 times less price responsive because of hydrological and geo-
graphical restrictions in allocation.
34

Timber market Pulp & paper Pellet market Agricultural

demand market demand demand rent
Input
data
Elasticity Large roundwood Small roundwood Elasticity

values inventory (t) inventory (t) values
I I
Large roundwood Small roundwood
removals removals
processes
Large roundwood Small roundwood

Model
inventory (t+1) inventory (t+1)
VI
III II III
VI 2
Large roundwood price Small roundwood price
IV
Timber rent
V
Timberland area
Output
data
Pine Natural Mixed Upland Lowland

plantation pine forest forest hardwood forest hardwood forest
Figure 2.4: Linkages and feedback mechanisms in SRTS model. The type of the relationship between
two model components is shown using plus (positive) and minus (negative) signs. Dashed lines show
feedbacks subject to a time lag.
PLUC was adapted to the US southeast case study by including projections of the
expansion or contraction of eight dynamic land-use classes (see Annex 3 for land use
projections used in PLUC). These are urban land, cropland, pasture, pine plantation,
natural pine forest, mixed forest, upland hardwood forest and lowland hardwood
forest. Water and federal land are static land-use classes. Non-forest vegetation
was included as a passive land-use class, this land-use class includes all non-forest
vegetation land uses including wetlands and shrub lands (see Annex 4 for a more
detailed definition of land use classes). PLUC was further adjusted for the southeastern
US context by defining the allocation order (see Table 2.1 and Annex 5) and including
land-use specific suitability factors applicable to the US case. A literature study was
done to identify potential suitability factors. A binary logistic regression analysis was
then carried out to select from these potential suitability factors, those factors, which
significantly explained patterns of historical land-use transitions in the southeastern
US between 2001 and 2011. Historical land use transitions were determined using
35

Chapter 2
land cover data from the National Land Cover Database (NLCD) from 2001 and 2011
(Homer, 2007, 2015); this timeframe was chosen because of availability of spatial data.
Each suitability factor was assigned a weight for each of the land-use classes; see
Table 2.1. Weights were calculated as the difference in R 2 created by leaving the factor
out of the model. A higher R 2 value means more of the variation in historical land
use transitions was explained by the combined suitability factors. The relationship
between the suitability factor and the transition into each land use was determined to
be either positive or negative. Due to a lack of spatial data that distinguishes between
pine plantation and natural pine forests, suitability factors for these land-use classes
were the same. Annex 6 provides a description of the potential and final suitability
factors.
PLUC requires input of a land-use map for the start year (2010), which in this case was
an initial land-use map developed specifically for this purpose. The initial land-use map
was created using the method of Verstegen et al. 2016 (Verstegen, van der Hilst, 2016).
As input data for this method, data of the USDA National Resources Inventory (USDA,
2013) of land-use type per state in 2010 were combined with spatial input data from
the National Land Cover Database (NLCD) (Homer, 2007, 2015) and the National Gap
Analysis Program (US Geological Survey Gap Analysis Program (GAP), 2011). Annex 4
describes how the land-use categories of these data sources were translated into the
categories used in this analysis. Projections of timberland area, including plantations,
were based on SRTS output. Projections of urban, cropland and pasture demand were
adapted from RPA 2010 scenario A1B (Wear, 2011). RPA scenario A1B was selected
because it is based on similar projections of future GDP growth as the scenario used
in the SRTS analysis (Ince, 2012).
The model was run from 2010 to 2030 with a time step of one year at a resolution of 2
by 2 km and results were visualized using Aguila (Pebesma, 2007) and ArcGIS software.
36

AnnaDuden_BNW.indd 37
Table 2.1: Parameters included in the sensitivity analysis with their upper (high) and lower boundaries (low). Source lists the rationale behind choosing these
values. Rw = roundwood
Softwood Softwood Hardwood Hardwood

Parameter Input type Source
small rw large rw small rw large rw
Demand price elasticity

Default 0.3 0.3 0.3 0.3 Default values SRTS
Low 0.2 0.2 0.2 0.2 Based on literaturea
High 0.5 0.5 0.5 0.5 Based on literatureb
Default values SRTS, for softwood also see
Supply price elasticity

Default 0.3 0.5 0.3 0.5
literaturea
Based on literature: small roundwoodc,
Low 0.23 0.42 0.23 0.42
large roundwoodd
High 0.35 0.55 0.35 0.55 Estimatione
Supply inventory
Default values SRTS, for softwood also see
Default 1 1 0.7 0.7
elasticity
literaturea
e
Estimation , for hardwoods based on
Low 0.9 0.9 0.6 0.6
literaturea
High 1 1 0.8 0.8 Estimatione
Large to small
roundwood (%)
Default 15% 30% Default values SRTS
Low 13.5% 27% Estimatione
High 16.5% 33% Estimatione
Cut-off point small and

large roundwood (inch)
Default 9 11 Based on literaturef
Alternative value 11 X Estimationg
Alternative value 13 X Estimationg
Agricultural rent (%
increase)
Default 2.28 Based on literatureh
Low 0 Estimatione
High 3.81 Based on literaturei
37
2
16-03-20 12:27
Default 9 11
point literaturef
small and Alternati Estimatio
11 X
large ve value ng
roundwo Alternati Estimatio
13 X
od (inch) ve value ng
Based on
Chapter 2 Default 2.28
Agricultu literatureh
ral rent Estimatio
Low 0
(% ne
Validation and sensitivity
increase) Based on
High 3.81
The spatial allocation of the different land uses by the adapted PLUC model was literature
i
validated by running the model for the time frame 2000-2011 using NLCD data as
input (Homer, 2007). Modelling resultsof
The spatial allocation forthe
2011 were compared
different land uses to the
by observed
the adapted 2011
PLUC model was
NLCD map (Homer,validated by running the model for the time frame 2000-2011 using NLCD data as
2015). The accuracy of the results was quantified by calculating the
input (Homer, 2007). Modelling results for 2011 were compared to the observed
Standardized Root Mean Squared Error (RMSE) between observation and modelled
2011 NLCD map (Homer, 2015). The accuracy of the results was quantified by
results per land-use class on different spatial levels: ecoregion, state, survey unit and
calculating the Standardized Root Mean Squared Error (RMSE) between observation
county level. The RMSE was calculated as follows:
and modelled results per land-use class on different spatial levels: ecoregion, state,
survey unit and county level. The RMSE was calculated as follows:
./ ("(#$%&'(&))+"(,#)&--&)))/
01.
RSME =
"(#$%&'(&))/3
With “A(observed)” being the area per spatial unit of a certain land use class in the
With “A(observed)” being the area per spatial unit of a certain land use class in the
NLCD map in 2011 (Homer, 2015) and “A(modelled)”
NLCD map in 2011 (Homer, 2015) and “A(modelled)” the area per spatial unit of a
the area per spatial unit of a
certain land use class in the PLUC results for 2011. N is the number of spatial units:
certain land use class in the PLUC results for 2011. N is the number of spatial units: 5
5 ecoregions, 12 states, 51 survey units and 949 counties. A map showing the
ecoregions, 12 states, 51 survey units and 949 counties. A map showing the different
different spatial levels can be found in Annex 7.
spatial levels can be found in Annex 7.
A sensitivity analysis was carried out for the assumptions made in SRTS. The
parameters included in the sensitivity analysis are related to product definition,
A sensitivity analysis was carried out for the assumptions made in SRTS. The
parameters included in the sensitivity analysis are related to product definition,
30
elasticities of price and inventory and agricultural rent. Table 2.2 shows all variables
included in the sensitivity analysis and their ranges. SRTS has a static definition of
feedstock types throughout the runs. In reality however, these definitions are expected
to shift and vary with relative changes in prices for large and small roundwood
depending on local circumstances such as proximity to pellet mills. Therefore, the cut-
off point between large and small roundwood was varied in the sensitivity analysis. In
addition, assumptions are made in SRTS about the percentage of large roundwood that
is expected to be of low quality and as a result not usable as saw timber; this includes
tops and branches, crooked or diseased trees (see Table 2.2 for an overview of SRTS
input parameters). Consequently, a percentage of large roundwood is expected to end
up in the small roundwood pool. The influence of this assumption was tested in the
sensitivity analysis. The influence of assumptions about demand and supply elasticities
was tested because a range of potential values was found in previous studies (see
Table 2.2). Finally, the effect of agricultural rent developments on timberland area was
tested, because the developments of agricultural rents are uncertain.
38

Table 2.2: Suitability factors per land-use type with corresponding weights as used as input in the
PLUC model. The table also provides the allocation order of the land use types. The R 2 per land use
shows the amount of variation in historical land-use change that was explained by the combination
of significant suitability factors.
pine forest
plantation
hardwood
hardwood
Cropland
Lowland
Pasture
Natural
Upland
Urban
Mixed
forest
Pine
Suitability factors
Allocation order 1 2 8 3 6 7 5 4
Elevation (m) -0.09 -0.04 -0.04 -0.12 -0.02
Slope (degrees) -0.03 -0.04 -0.04 -0.04 +0.10 -0.02 2
Rainfall (mm) +0.01 -0.05 +0.09 +0.09 -0.14
Yield (kg DW/ha) +0.03 -0.22 +0.05 +0.05 -0.12
Growing season
-0.10 +0.19 +0.19 +0.18 +0.09 +0.01
(days)
Soil type 0.15 d 0.62 d 0.15 d 0.15 d 0.22 d
Distance to main
-0.03 +0.79
roads (km)
Distance to urban
-0.79 +0.05 +0.05 -0.03
area (km)
Distance to river
+0.03 +0.03 +0,05 -0.05
(km)
Distance to wood
-0.07 -0.04 -0.04 -0.05
mills a (km)
Distance to main
-0.10 -0.04 -0.04 -0.16
ports (km)
Density of self b -0.01 +0.15 +0.08 +0.08 +0.20 +0.04
Density of urban
+0.10 -0.05 -0.05 -0.05 -0.08 -0.07
areac
Density of forest
-0.21 -0.14 +0.15 +0.15 -0.20 -0.13 +0.13
areac
R2 0.644 0.324 0.100 0.272 0.401 0.275 0.343
2.3 Results
The combined analysis using the SRTS and PLUC models results in projections of
timberland area per forest type for the four different wood market scenarios and
subsequent projections of changes in land use. First, a description of projected changes
in timberland area is given, followed by an overview of spatial projections of land-use
transitions. Land-use changes related to timberland are then discussed in more detail.
39

Chapter 2
Timberland area projections

Modelling results from SRTS provide projections of timberland area per year for 5
forest types; pine plantations and 4 types of natural timberland. Total timberland
area is higher in 2030 compared to 2010 in all scenarios but the LhLp scenario.
Increased pellet demand leads to projected higher timberland area compared to no
additional pellet demand in 2030. However, increases in timberland area consist
mainly of an increase in pine plantation area. Pine plantation area has an upward
trend under the high housing scenarios, while in the low housing scenarios, it declines
first before it starts increasing (see Figure 2.5). The area of natural timberland is
projected to decrease by about 500 (HhHp scenario) to 15,000 km2 (LhLp scenario)
by 2030, especially natural pine and mixed forest. Changes in total timberland area
(including pine plantation) between 2010 and 2030 range from an increase of 4.8%
(HhHp scenario) to a decrease of 1.3% (LhLp scenario). Projected changes in total
timberland area are mostly due to a change in plantation area, which increases by
4-19% under the different scenarios (LhLp and HhHp respectively). When demand
from the housing market is assumed to be high, additional wood pellet demand leads
to projected establishment of approximately 8,000 km2 additional pine plantation (a
change of 4.7%) and avoided loss of just under 2,000 km2 (a 0.1 to 0.5% decrease in
total area) natural timberland by 2030 when compared to low pellet demand. For the
low housing demand scenarios, increased wood pellet demand leads to the projected
establishment of almost 20,000 km2 of pine plantation (a change of 11.2%) and about
7,500 km2 less natural timberland lost by 2030 (a 1.5 to 3.0% decrease).
40

High housing - high pellet

High housing - low pellet scenario scenario
120 120
Timberland area (%)
Timberland area (%)

115 115
110 110
105 105
100 100
95 95
90 90
2010 2015 2020 2025 2030 2010 2015 2020 2025 2030
Low housing - low pellet scenario Low housing - high pellet scenario
120 120
2
Timberland area (%)
Timberland area (%)

115 115
110 110
105 105
100 100
95 95
90 90
2010 2015 2020 2025 2030 2010 2015 2020 2025 2030
PP NP MF UH LH TT
Figure 2.5: Timberland area per scenario in percentages with 2010 as base year. Forest type abbre-
viations - PP: planted pine forest, NP: natural pine forest, MX: mixed forest, UH: upland hardwood
forest, LH: lowland hardwood forest, TT: total.
Land-use transitions
Timberland area projections from SRTS were combined with projections from other
land uses, which were used as input for PLUC to produce yearly maps of projected land
use between 2010 and 2030 for the 4 wood market scenarios. Urban area is projected to
expand around all main cities, particularly in the northern part of Alabama, Georgia and
South Carolina, the eastern part of Tennessee and the western part of North Carolina
(see Figure 2.6). Establishment of new pine plantation is projected to occur mainly in
the coastal parts of Virginia, North Carolina, South Carolina, Georgia, Alabama and
Mississippi. This is expected due to suitability of this area because of proximity to
existing forests, as well as suitable soil types and a longer growing season. New natural
timberland regenerates in the coastal regions as well, often alongside pine plantations.
The different scenarios provide similar patterns of land-use transitions (see Annex 8
for land-use transition results for the HhLp, LhHp and LhLp scenarios). Pine plantations
are established mainly at the expense of natural pine and pasture (see Figure 2.7). Some
pasture is replaced by regeneration of natural pine and mixed forest. Some of the pasture
area lost is restored by a transition from non-forest vegetation into new pasture area,
particularly in Florida in the HhHp scenario. The two scenarios with high pellet demand
lead to the largest cumulative amount of land use transitions. This is due to the larger
41

Chapter 2
projected increase in timberland area for the high pellet demand scenarios compared to
the low pellet demand scenarios. The additional increase in timberland area in the high
pellet demand scenarios is established mainly at the expense of pasture area.
No transition
To urban
To pasture
To pine plantation
To natural timberland
To non-forest vegetation
Figure 2.6: Transitions in land use between 2010 and 2030 for the HhHp scenario. Colours show the
land uses in 2030. The natural timberland types were aggregated into one class.
Land use transitions > 4,000 km²

180,000
Rest
160,000
Cropland to nfor veg
Pine plantation to nfor veg
140,000
Pasture to nfor veg
Area of transition (km²)
Pine plantation to pasture

120,000
Nfor veg to pasture
Cropland to pasture
100,000
Natural pine to pine plantation
Pasture to pine plantation
80,000
Mixed forest to pine plantation
Pasture to natural pine
60,000
Mixed forest to natural pine
Nfor veg to natural pine
40,000
Pasture to mixed forest
Natural pine to upland hardwood
20,000
Upland hardwood to urban
Pasture to urban
0
HhHp HhLp LhHp LhLp
Figure 2.7: Area of transitions in land use (in km2) between 2010 and 2030 for the different scenarios.
The graph shows all transitions larger than 4,000 km2 in area separately, all smaller transitions
were aggregated into the category ‘rest’.
42

Pine plantations
Even in absence of strong demand from the housing or pellet market, an increase in
pine plantation area is expected in most states (see Figure 2.8). This increase becomes
considerably larger however with increased pellet demand or high housing demand.
Urbanization led to the loss of about 1,850 km2 of pine plantation, with the largest
losses occurring in Alabama and Georgia. Besides Oklahoma and Texas, all states in
the southeastern US show a net gain in pine plantation in all scenarios, with many
states showing a strong increase. The HhHp scenario leads to the strongest increase
in pine plantation in all states except for Oklahoma, Tennessee and Texas, which do not
have a lot of pine plantation to begin with. When comparing the high and low pellet
2
scenarios it becomes clear that pellet demand has a larger influence on pine plantation
establishment in the absence of a strong housing market (i.e. the difference between
the LhLp and LhHp scenario is larger than the difference between the HhHp and the
HhLp scenarios). It also shows that even if the housing market is relatively strong,
pellet demand can still have an added influence on pine plantation establishment,
as can be seen from the difference between the HhHp and the HhLp scenarios. Pine
plantation loss occurs mainly in the low housing scenarios, but this loss is offset by
gains in plantation area in most states.
For all scenarios, transitions into pine plantation occur mainly in the coastal parts of
Virginia, North Carolina, South Carolina, Georgia, Alabama and Mississippi (see Figure
2.9), where most of the pine plantations occur today. The new areas of pine plantations
are spread out over this region, and are established mostly at the expense of natural
pine – and to a lesser extent at the expense of pasture. Results for all scenarios
show a similar pattern (see Figure 2.8 and Annex 10). While the establishment of
pine plantation is scattered across the region, some areas show relatively large
increases in pine plantation area at the local level. Large differences in pine plantation
establishment can be seen between the scenarios on a local scale, particularly along
the Gulf coast. These locations seem to be suitable for pine plantations mostly because
of the proximity of existing natural timberland and plantation forest area, as well as a
longer growing season. We can conclude that establishment of pine plantation between
2010 and 2030 varies strongly over the region, but in many states net pine plantation
area will increase considerably, mostly at the expense of natural timberland.
43

44
Area of land use transitions to and from pine plantation for 2010-2030
Alabama Arkansas Florida Georgia Louisiana Mississippi North Oklahoma South Tennessee Texas Virginia
Chapter 2
Carolina Carolina
6,000
5,000
Area of pine
4,000
3,000
plantation gained (km²)

2,000
1,000
-1,000
-2,000
Area of pine
plantation lost (km²)
-3,000
Natural pine Mixed forest Upland hardwood Lowland hardwood Non-forest vegetation Pasture Urban Net area change
Figure 2.8: Area in km2 of land use transitions involving pine plantations between 2010 and 2030, per state and for the different scenarios. Above the x-axis
the area of pine plantation gained, colours show the land use type in 2010 that was replaced by pine plantation. Below the x-axis the area of pine plantation
lost, colours show the land use type that replaced pine plantation by 2030. Black bars show the net gain or loss of pine plantation per state for each scenario.
16-03-20 12:27
Land-use transitions – pine plantations

No transition
Loss of pine plantation
Upland hardwood forest to pine plantation
Lowland hardwood forest to pine plantation
Non-forest vegetation to pine plantation
Pine plantation
Figure 2.9: Land use transitions between 2010 and 2030 involving pine plantations for the LhHp
(upper panel) and LhLp (lower panel) scenario. Circles show a close-up of the coastal area of Mis-
sissippi and Alabama.
Natural timberland
In the absence of strong demand for wood products, natural timberland declines or
stays relatively stable (see Figure 2.10). Increased pellet demand leads to a larger
area of natural timberland lost to pine plantation, as well as a larger area of forest
regeneration at the expense of pasture land. The amount of natural timberland lost
to urbanization is 16.400 km2, with most transitions occurring around major cities
(see Figure 2.11 and Annex 9). Arkansas and Texas show very little transition of
natural timberland to new urban land. For most states, the largest share of natural
timberland lost transitions into pine plantations. This result is particularly prominent
for the coastal parts of Virginia, North Carolina, South Carolina, Georgia, Alabama
45

Chapter 2
and Mississippi, where the largest increases in pine plantation area are expected.
For Florida and Oklahoma, a relatively large area of natural timberland is lost to new
pasture areas (up to about 2,500 km2 and 1,000 km2 respectively in the HhHp scenario).
This is mostly due to loss of small remnants of upland hardwood in these areas. In the
states Arkansas, Oklahoma, Texas and Virginia, a net gain in natural timberland area
is projected for all scenarios. In Louisiana and Tennessee, natural timberland area
increases in all but the LhLp scenario. Natural timberland is gained mostly at the
expense of pasture area and to a lesser extent non-forest vegetation. Reforestation
occurs throughout the southeastern US, but mostly in the eastern states of Georgia,
Arkansas and Texas. This could be explained by a combination of proximity to existing
forest area, steeper slopes, and relatively low suitability for other land uses. The states
of Alabama, Florida, Georgia, Mississippi, North Carolina and South Carolina show a
net loss of natural timberland area under all scenarios.

Accuracy of results was evaluated by validating the spatial allocation of land-use
change for the adapted PLUC model as well as a sensitivity analysis for the SRTS model.
RMSE values were calculated for 4 spatial levels (see Table 2.3). PLUC results are most
accurate (i.e. a lower RMSE value) for the land-use types of lowland hardwood, urban
and cropland. This can be explained by the relatively high ranking in the allocation
order of urban area and cropland, which means that these land uses get first pick of
most suitable locations. Another explanation is the relatively high R 2 values in the
regression analysis of the suitability factors for urban area and lowland hardwood,
which means the model is able to capture spatial patterns that determined allocation
of the land use better. Results are less accurate for mixed forest, pine forest, pasture
and non-forest vegetation. This result can be explained by the relatively low ranking
in the allocation order of mixed forest, pasture and non-forest vegetation, as well as
low R 2 values of the suitability factors for pasture and pine forest. The accuracy of
PLUC is reduced with increasing spatial resolution but remains relatively high for all
land-use classes except non-forest vegetation up to a survey unit level - and for urban
and lowland hardwood forest even up to a county level (similar to (Verstegen, van
der Hilst, 2016)). Because PLUC was able to successfully replicate historical land-use
changes up to the survey unit level, we assume that our projections of future land
use can be considered robust up to that level, assuming that future condition are the
same as the past.
46

Area of land use transitions to and from natural timberland for 2010-2030
Alabama Arkansas Florida Georgia Louisiana Mississippi North Oklahoma South Tennessee Texas Virginia
Carolina Carolina
8,000
6,000
Area of natural
4,000
forest gained (km²)

2,000
-2,000
-4,000
Area of natural
forest lost (km²)
-6,000
-8,000
Urban Pine plantion Pasture Non-forest vegetation Cropland Net area change
Figure 2.10: Area in km2 of land use transitions involving natural timberland between 2010 and 2030, per state and for the different scenarios. Above the
x-axis the area of natural timberland gained, colours show the land use type in 2010 that was replaced by natural timberland. Below the x-axis the area of
natural timberland lost, colours show the land use type that replaced natural timberland by 2030. Black bars show the net gain or loss of natural timberland
per state for each scenario.
47
2
16-03-20 12:27
Chapter 2
Land-use transitions – natural timberland

No transition
Reforestation
Natural timberland to urban
Natural timberland to pine plantation
Natural timberland to pasture
Natural timberland to non-forest vegetation
Natural timberland
Figure 2.11: Land use transitions between 2010 and 2030 involving natural timberland for the
LhHp (upper figure) and LhLp (lower figure) scenario. Circles show a close-up of central Arkansas
and eastern Oklahoma.
A sensitivity analysis was done for the main assumptions made in SRTS related to
parameters on product definition, elasticities and agricultural rent. Of the different
forest types, pine plantation proved to be most sensitive to shifts in parameter values,
particularly for elasticities (see Figure 2.12 and Annex 11). This result is due to the
assumption that pine plantation is more price responsive than natural timberland
forest types. The cut-off point for the division between small and large roundwood
had a smaller effect on pine plantation area, with shifts leading to up to a 3% change
in the area of pine plantation. Variation in agricultural rents and the percentage of
large roundwood that ends up in the small roundwood pool had no significant impact
on timberland area.
48

Agricultural rent increase Inventory & price elasticities

1.15 1.15
Area change (indexed)

Area change (indexed) 1.1 1.1
1.05 1.05
1 1
0.95 0.95
0.9 0.9
0.85 0.85
low default high low default high
Agricultural rent increase Elasticity values
Feedstock cut-off point % large rw to small rw

1.15 1.15
2

1.1 1.1
1.05 1.05
1 1
0.95 0.95
0.9 0.9
0.85 0.85
9 11 13 low default high
Cut-off point product definition (inches)
% to small roundwood value
PP NP MX UH LH TT
Figure 2.12: Sensitivity of total timberland area in the HhHp scenario to a) cut-off point between
small and large roundwood, b) elasticity values for demand and supply c) annual increase in ag-
ricultural rent d) the percentage of large roundwood ending up in the small roundwood pool. For
low, default and high values for elasticities, agricultural rent and % to small roundwood value, see
Table 2.3. Results of the sensitivity analysis for the HhLp, LhHp and LhLp scenarios can be found in
Annex 11. The abbreviation rw stands for roundwood. Abbreviations in the legend: pp = pine plan-
tation, np= natural pine, mx= mixed forest, uh= upland hardwood, lh= lowland hardwood, tt= total.
Table 2.3: Coefficient of variation of the RMSE values for the area of different land use types in
2011 at different spatial levels. The change in area (in km2) for the input data is also given. ∆
area = Absolute area change NLCD 2001-2011 (km2). RMSE = root mean squared error.
Ecoregion State Survey unit County

Land use types ∆ area Standardized RMSE
Urban 7,309 0.01 0.02 0.04 0.12
Cropland 2,299 0.00 0.05 0.10 0.35
Pasture 13,256 0.05 0.09 0.14 0.36
Pine forest 13,942 0.02 0.08 0.13 0.25
Mixed forest 7,224 0.07 0.07 0.12 0.16
Upland hardwood 8,626 0.04 0.05 0.10 0.18
Lowland hardwood 2,699 0.00 0.01 0.03 0.08
Non-forest vegetation 13,696 0.13 0.15 0.33 1.04
49

Chapter 2
2.4 Discussion
This study presents a modelling approach that can be used to project changes in land
use in a spatially explicit way based on scenarios of future demand for wood pellets
and other wood products. This method was demonstrated for wood pellet production
in the southeastern US. Results show that an additional demand for wood pellets (of
11.6 Mt, from a starting point of about 0.5 Mt in 2010) and other wood products can
have a considerable impact on timberland area in 2030. Increased wood pellet demand
led to projections of establishment of approximately 8,000 km2 (high housing demand)
to almost 20,000 km2 (low housing demand) additional pine plantation area compared
to stable pellet demand, which amounts to an increase of 4.7% and 11.2% in pine
plantation area respectively. It also resulted in projections of avoided loss of natural
timberland of almost 2,000 km2 (high housing demand) or to a loss of just over 7,500
km2 (low housing demand), which amounts to 0.4% and 1.5% of the current natural
timberland area respectively. Our projected timberland area changes are modest
compared to previous research on the southeastern US, which predict a change in
pine plantation area in the range of -11% to +67% and a change in timberland area of
-10% to +3%, depending on scenario assumptions, timeline and geographical extent
(K. L. Abt, 2014; R. C. Abt, 2009b; Galik, 2012; Prestemon, 2002; Wear, 2013b)(see Annex
12 for an overview of previous findings on timberland and pine plantation expansion
in the southeastern US).
The innovative component of this study is the spatial allocation of these changes in
timberland area. Projected establishment of new pine plantations occurs mainly in
the coastal parts of Virginia, North Carolina, South Carolina, Georgia, Alabama and
Mississippi. This is in line with earlier findings on historical dynamics (Drummond,
2010) as well as future projections (Prestemon, 2002) for this region. According to our
results, pine plantations are often established at the expense of natural timberland;
mainly natural pine and mixed forest. Projected forest regeneration occurs throughout
the southeastern US but is more concentrated in the eastern parts of Georgia, Arkansas
and Texas. Natural timberland cover was reduced in the areas where pine plantation
expands, as well as around all major cities as a result of urbanization. Natural
timberland cover was reduced in 6 states, while in 6 other states the area increased.
This finding differs from Prestemon and Abt, who find a loss of natural timberland
for all states in their base scenario between 1995 and 2040 (Prestemon, 2002).
Similarly however, they find a strong loss of natural timberland in North Carolina,
South Carolina, Georgia and Alabama (as well as Florida), as a result of pine plantation
expansion and urbanization.
50

A number of limitations of the modelling approach need to be taken into consideration

when observing the findings presented here. Validation of the model shows that these
results are robust up to a survey unit level (on average 25,000 km2 in size). However,
the PLUC model is based on an extrapolation of historical land-use trends used to
project future trends in land use. This may not reflect potential changes in drivers
of land-use change over time (Verstegen, Karssenberg, 2016). The link between the
SRTS and PLUC model presented here is based on input-output, without any feedback
mechanism between the two models. This means that changes in forest area were
only spatially allocated after SRTS was run, so spatial input data such as suitability
factors did not influence the outcome of SRTS runs. In reality, there might be spatial
2
variation in a number of input variables of SRTS, such as the cut-off point between
small and large roundwood, and agricultural rent. A more robust integration of both
models would improve the strength of results (Wicke, 2015).
A number of assumptions used in the individual models might also have influenced
our results in a significant manner. In the SRTS model, forest productivity was fixed
throughout the modelling period. However, in the presence of strong forest product
markets, forest productivity could increase over time (K. L. Abt, 2012; Prestemon, 2002),
which could result in a lower expected increase in timberland area and a smaller shift
from natural timberland to pine plantations. These increases in productivity could
be the result of changes in forest management related to, for instance, harvest cycle
or planting densities ( Jonker, 2018). Furthermore, the SRTS model is based on the
assumption that land or forest owners respond to market prices. It has however been
shown that other factors - including property size and age, profession and income of the
land owner, as well as life events, such as the need to raise money for health, education,
or retirement - also play a role in forest management and ownership decisions in
the southeastern US (Aguilar, 2014; Butler, 2017; Joshi, 2009; Lubowski, 2008). Finally,
the SRTS model was also shown to be sensitive to assumptions on elasticity values.
The values used in this study are default values for SRTS and in line with previous
studies, but choosing different elasticity values would influence forest area projections
significantly. Higher elasticity values of feedstock demand price, supply price and
supply inventory lead to lower changes in timberland area, particularly for pine
plantation.
The results of the PLUC model could be improved by resolving data constraints,
for instance by the availability of more recent spatial data, particularly for pine
plantations. It was impossible to perform a regression analysis for pine plantation and
natural pine independently, because these categories were combined in the historical
51

Chapter 2
spatial land-use data. Therefore, planted pine and natural pine were pooled in the
regression analysis, while these two land use types may have different suitability
factors. Pine plantation establishment could be more dependent on anthropogenic
factors such as accessibility and distance to wood using mills, and, as a result, have a
different allocation pattern than natural pine. A forthcoming update of the GAP land-
use map could resolve this issue by providing additional spatial data on pine plantation
locations. Furthermore, PLUC results at the local level provide an indication of the
spatial patterns in that area; however pixel values should not be considered precise
predictions.
The results presented here include a distinction between pine plantations and natural
timberland types. The projected shift from natural timberland to pine plantations
is expected to have significant sustainability impacts, particularly for biodiversity
and carbon storage. Even though natural timberland regenerates rapidly in most
areas, it may take a few decades before new forests reach mature levels of carbon
storage and biodiversity. However, the transition of pasture into pine plantation forest
could increase carbon storage in the landscape. Environmental impacts are expected
to differ across the landscape according to the dominant local land use transition
patterns and heterogeneity in biophysical characteristics.
2.5 Conclusion
This study demonstrates a spatially explicit modelling approach to assess the effect
of an increasing wood pellet demand on land-use dynamics, while taking into account
demand for other wood products (i.e. saw timber and pulp and paper), as well as
development of other land uses (such as urban land and cropland). The modelling
approach presented here could be useful for determining land-use change patterns and
identifying the most prominent land-use transitions. This modelling approach can be
applied for other scenarios of demand for forest products as well as other projections
of land use. In this study, it was applied to determine the effect of increased demand
for wood pellets (of 11.6 Mt) and other wood products by 2030. We conclude that an
increase in wood pellet demand leads to an increase in pine plantation establishment
and an increase in both loss (to pine plantation) and gain (through regeneration) of
natural timberland. This results in a lower loss of natural timberland under high pellet
demand scenarios. In some areas, increased demand for wood pellets is projected
to result in an increased natural timberland cover, but in most locations it will lead
to a substantial shift from natural timberland to pine plantations. Future research
52

could compare the delivery of environmental services for different stand ages for
natural timberland and pine plantations, to understand potential impacts of a shift
in forest types. The results on land use transitions developed in this study will be
used as input for further research on environmental impacts for carbon storage and
biodiversity conservation. Projections created using this model and follow-up analyses
of environmental impacts can inform the debate about the sustainability of wood
pellets.
53

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
3
Impact of increased wood pellet demand
Environmental
on biodiversity in the southeastern US
Anna Duden
A.S. Duden, M.J. Rubino, N.M Tarr, P.A. Verweij, A.P.C. Faaij, F. van der Hilst
This chapter has been published in the journal Global Change Biology Bioenergy.
Anna Duden

Chapter 3
Abstract
Increasing wood pellet exports from the United States are projected to lead to changes
in land use and timberland management, including a shift from natural timberland to
pine plantations. These projected changes may impact biodiversity. This study aims
to quantify potential biodiversity impacts of increased wood pellet demand in the
southeastern US in a spatially explicit manner. We determined differences according
to an index of potential species richness (for total, threatened and endemic species
and four taxonomic groups) between scenarios of high and low demand for wood
pellets, while taking into account potential developments in other wood markets and
other land uses.
Increased demand for wood pellets was projected to cause both positive and negative
biodiversity impacts. Negative shifts in total potential species richness were projected
for areas in Florida, coastal Virginia and North Carolina, and parts of the Gulf Coast.
Positive shifts in total potential species richness were projected in parts of Oklahoma
and Arkansas. In some locations, the direction of change differed per taxonomic group,
highlighting the importance of analysing different taxonomic groups. Shifts in potential
species richness due to increased wood pellet demand were considerably smaller
compared to the changes due to other drivers, such as urbanization and increased
timber demand. Biodiversity impacts due to wood pellet demand should therefore be
considered in the context of other drivers of land-use change and biodiversity loss.
Our results provide information that allows policy makers, industry and NGOs to
focus on areas of concern and take appropriate mitigation measures to limit negative
biodiversity impacts and promote positive impacts. The spatially explicit approach
presented in this study can be applied to different regions and drivers of land-use
change, to show how projected demand for an internationally traded commodity may
lead to impacts on land use and biodiversity in the procurement region.
56

Impact of increased wood pellet demand on biodiversity in the southeastern US
3.1 Introduction
Exports of wood pellets from the United States (US) have grown from close to zero in
2008 (Dale, Parish, 2017) to 4.7 Mt per annum in 2016 (Copley, 2017). Of all exported
pellets from the US, about 98% is produced in the southeastern US (Wang, 2015a).
The wood pellet producing sector in the US is largely driven by demand from the
European Union (EU) (K. L. Abt, 2014; Joudrey, 2012), which is in turn determined by
policies within the EU member states (K. L. Abt, 2014). Future development in wood
pellet production in the southeastern US is uncertain, partly because it relies heavily
on European policies. Projections of wood pellet production in the region range from
~6 Mt (Pinchot Institute, 2013) to ~16 Mt per annum in 2020 (K. L. Abt, 2014). An
increased demand for wood products, including wood pellets, has been projected to
3
lead to a change in timberland7 area (Chapter 2 and (R. C. Abt, 2009a, 2013a)) and a
shift from natural timberland8 to pine plantation (Chapter 2 and (Evans, 2013)). These
changes in timberland area and management may exacerbate changes due to trends in
other wood using markets such as the US domestic housing market and the pulp and
paper market (Chapter 2). Changes in forest area driven by wood markets occur in
addition to forest losses due to urbanization, which is identified by the US Department
of Agriculture (USDA) Forest Service as a major risk to forest sustainability in the
southeastern US (Wear, 2013b). The area of pine plantations in the southeastern US
increased by over 54 thousand km2 between 1990 and 2013 (an average increase of
2% per year), reaching almost 180 thousand km2 in 2013 (USDA FS, 2017a). In the same
time period, natural timberland decreased by almost 34 thousand km2 (0.1% per year)
to 620 thousand km2 in 2013 (USDA FS, 2017a). An increased demand for wood pellets
may exacerbate the existing trend of loss of natural timberland, while at the same time
promote the establishment of pine plantations. These projected shifts in land use and
forest management due to increasing demand for wood pellets could impact (either
positively or negatively) biodiversity in the southeastern US (Parish, 2018).
7 Timberland is defined as ‘forest land that is producing or is capable of producing crops of indus-
trial wood and not withdrawn from timber utilization by statute or administrative regulation.
(Note: Areas qualifying as timberland are capable of producing in excess of 20 cubic feet (1.4 cubic
meters) per acre (0.4 hectare) per year of industrial wood in natural stands. Currently inaccessible
and inoperable areas are included.)’ following the US Forest Service e.g. (Oswalt, 2014a)
8 Natural timberland is defined as ‘Productive forests composed of trees established by natural
regeneration of existing seed sources, root suckers, stump sprouts, etc. Establishment may be
either afforestation on land that until then was not classified as forest or by reforestation of land
classified as forest after a disturbance or following harvest’ following the Forest Inventory and
Analysis National Program (see for instance (Oswalt, 2014a)).
57

Chapter 3
The sustainability of wood pellet production has been questioned and discussed
extensively for several reasons including its potential contribution to changes in
land use and forest management (Chapter 2 and (Costanza, 2015, 2017; Prestemon,
2002)) and impacts on biodiversity (Evans, 2013; Olesen, 2016; Pelkmans, 2014; Tarr,
2016). Land-use change and subsequent habitat loss have been the major cause of
biodiversity loss in terrestrial ecosystems over the last 50 years (CBD, 2010; Millenium
Ecosystem Assessment, 2005; Pimm, 2000; WWF, 2012), and are projected to continue
to drive biodiversity losses in the future ( Jetz, 2007; Sala, 2000). The southeastern
US is rich in endemism ( Jenkins, Van Houtan, 2015) and a large part of the region,
the North Atlantic Coastal Plain, has been identified as a global biodiversity hotspot
(Noss, 2015). Two important land-use trends which may lead to biodiversity loss are
expected in the southeastern US in the coming decades: urbanization (McKinney, 2008;
Price, 2006) and expansion of pine (Chapter 2). Pine plantations in the southeastern
US are currently among the most intensively managed forests in the world (Fox, 2007).
Some field studies have found lower species diversity in plantation forests compared
to natural forests, although other studies have also found similar or higher diversity
in forest plantations (Carnus, 2006; Stephens, 2007). Biodiversity in pine plantations
strongly depends on plantation age (Carnus, 2006), forest management (Carnus, 2006;
Miller, 2009) and abiotic factors (Miller, 2009). Biodiversity impacts of plantation
establishment may differ according to the initial land use and spatial heterogeneity
of abiotic factors (Immerzeel, 2014; Tarr, 2016). As a result, it is important to assess
impacts of land-use change due to wood pellet demand on biodiversity in a spatially
explicit manner.
Several European countries have discussed sustainability guidelines that may limit
the sourcing of wood pellets from the southeastern US (Galik, Abt, 2015). Biodiversity
loss has been identified as one of the major EU policy risks regarding the wood pellet
imports from the region (Olesen, 2016). The effect of biomass production on ecosystems
has been identified as one of the 40 scientific priorities to inform US conservation
managers and policy makers (Fleishman, 2011). The European Renewable Energy
Directive (RED) aims to increase the share of renewable energy, while ensuring that
‘areas of high conservation value are preserved’ and ‘the impacts of forest harvesting
on biodiversity are minimised’ (European Commission, 2017). Therefore, to fulfil the
requirements of the RED, it is necessary to identify areas of high conservation value
throughout sourcing areas of the southeastern US, and determine potential impacts
of wood pellet production on biodiversity. No uniform guidelines currently exist on
how to quantify these biodiversity impacts. Apart from the political need for impact
assessments of increasing wood pellet demand, biodiversity has an intrinsic and
economic value that requires conservation.
58

A number of studies have assessed the potential impact of wood pellet production on
biodiversity in the southeastern US using a spatially explicit approach. Some studies
showed overlap of wood pellet sourcing areas with areas of high biodiversity value.
For example, Evans et al. (2013) examined the overlap of habitats for 16 conservation
priority species with sourcing areas surrounding six wood pellet mills in Virginia,
North Carolina, and Georgia (Evans, 2013). Procurement areas of existing and proposed
pellet plants were also found to overlap with Important Bird Areas (Olesen, 2016). The
latter study focused on the entire southeastern US but only included areas important
for bird species. In some locations, Galik & Abt (2015) found an overlap of over 40%
between presently harvestable forest areas and areas of potential harvest restrictions
based on proposed EU sustainability guidelines for wood pellets (Galik & Abt, 2015).
This study spanned the entire southeastern US, but focused on areas that could be
3
excluded according to the EU sustainability criteria without determining potential
impacts outside these excluded areas. Finally, Tarr et al. (2016) assessed the impact
of land-use transitions due to bioenergy production in North Carolina on the habitat
area of 16 indicator species in a spatially explicit manner. Wood pellet production
was projected to lead to habitat increases for some species but decreases for others,
indicating that land-use transitions will lead to both winners and losers (Tarr, 2016).
However, Tarr et al. (2016) assessed just one state (North Carolina) within the wood
pellet producing region. Studies have identified regions of potential biodiversity risk
due to feedstock harvesting for wood pellets, including coastal Georgia and Virginia
(Olesen, 2016), Florida and the Atlantic coastal area (Galik, 2016). Species at risk
included threatened species such as the Cerulean Warbler (Setophaga cerulean) and
Eastern Spotted Skunk (Spilogale putorius) (Evans, 2013; Tarr, 2016). The area of dense,
shrubby vegetation may increase in bioenergy scenarios, providing additional habitat
for some species including the southeast endemic Oak Toad (Anaxyrus quercicus)
(Tarr, 2016). Spatially explicit studies on the impacts of woody biomass production
on biodiversity in the southeastern US have assessed the impacts either for a limited
number of species (Evans, 2013; Olesen, 2016) or for a restricted area in the wood pellet
sourcing region (Galik, 2016; Tarr, 2016). Using broad taxonomic groups as indicators
has been identified as a research need in the study of biodiversity impacts of bioenergy
(Fletcher, 2011; Immerzeel, 2014; Riffell, 2011b, 2011a; Verschuyl, 2011), partly because
of contradictory findings for selected species groups as indicators (Dauber, 2003).
The aim of this study is to identify and quantify potential future biodiversity impacts of
an increased wood pellet demand in the southeastern US in a spatially explicit manner.
This study focuses on the whole southeastern US in order to include all areas where
pellets will likely be sourced in the near future. We assess the impact of scenarios
59

Chapter 3
with and without increased wood pellet demand between 2010 and 2030 (Chapter 2)
on projected potential species richness for all terrestrial vertebrate species reported
for the region.
3.2 Methods
General approach
This study makes use of projections of land use up to 2030 following scenarios of
demand for wood pellets and other wood products derived from Chapter 2. In order
to determine the impact of an increasing demand for wood pellets on biodiversity,
we created potential species richness maps for the southeastern US by summing
all terrestrial vertebrate potential habitat maps available at the USGS Gap Analysis
Program database (McKerrow, 2018). We then generated an index based on location-
specific associations between land-use types and potential species richness by
overlaying potential richness maps with land-use maps based on Chapter 2 (Figure
3.1). Finally, we projected the species richness index in 2030 according to projections
of land-use change. To determine the potential impact of increased wood pellet demand
on biodiversity, we focussed the analyses on: 1) spatial variation in species richness
index within and among different land-use types, 2) projected baseline changes in
species richness index between 2010 and 2030 in the absence of increasing wood
pellet demand, and 3) potential impacts of scenarios of increased pellet demand on
biodiversity.
Land-use
projections
Wood pellet demand scenarios
Species richness
Index projections
Species habitat maps
Species
richness map
Figure 3.1: Overview of general approach to create projections of species richness index. Rectangles
symbolize maps, the stack of rectangles symbolizes individual species habitat maps of 811 species
included in the analysis. Arrows symbolize adaptations to the maps; 1) modelling of wood pellet
demand scenarios using the economic wood market model SRTS and land use change model PLUC
to create different land use projections up to 2030 (see Chapter 2), 2) summing of species habitat
maps taken from US-GAP (McKerrow et al., 2018) to create maps of potential species richness, 3)
spatial analysis using a moving window approach to create spatially explicit projections of species
richness index up to 2030.
60

Pellet demand scenarios and land-use projections

In Chapter 2, we created land-use maps for 2010 and 2030 under different scenario
assumptions using a combination of the Subregional Timber Supply (SRTS) model (R.
C. Abt, 2000) and the PCRaster Land-Use Change (PLUC) model (Van der Hilst, 2012;
Verstegen, 2012). The scenarios consisted of four alternative trends of developments of
wood markets (including wood pellets) in the southeastern US, as well as projections
of developments in different land-use types. The wood market projections included
a trend of increasing wood pellet demand (12.1 Mt by 2030) and a trend of stable low
wood pellet demand (0.5 Mt by 2030, see Figure 3.2) (Chapter 2). Because the wood
pellet sector in the southeastern US is strongly influenced by developments in the
domestic housing (Chapter 2), the biggest timber consuming sector, trends of high or
low domestic housing demand were also included in the scenarios. Scenarios consisted
3
of a combination of demand projections for wood pellets (high or low) and demand
projections for timber (high or low), resulting in 4 different scenarios (see Figure
3.2): HhHp (high housing demand, high pellet demand), HhLp (high housing demand,
low pellet demand), LhHp (low housing demand, high pellet demand) and LhLp (low
housing demand, low pellet demand). Land-use projections were created in the PLUC
model and based on projections of timberland (both natural and planted) area based
on the wood market scenarios, as well as projections for agricultural land, pasture and
urban land. The PLUC model allocates land use according to the projected area per
land use, an allocation order of the different land uses and suitability maps for each
land use. The resulting land-use projections were at a 2 x 2 km resolution and include
11 different land-use types. These include four anthropogenic land-use types (urban,
cropland, pasture and pine plantation); four natural forest types (natural pine, mixed
forest, upland hardwood forest and lowland hardwood forest); non-forest vegetation,
which comprises all natural land-use types not classified as forest; and two land-use
types for which projections are stable over time (federal land and water). Between
2010 and 2030, 9.4 to 12.3% of the area is projected to be subject to a transition in
land use (Chapter 2). The most common land-use transitions are natural pine to pine
plantation, pasture to natural pine and non-forest vegetation to pasture. We refer to
Chapter 2 for a more detailed description of the wood product demand trends and the
land-use projections.
61

Chapter 3
400
350
Feedstock demand (Mt)
300
250
200 Pellet
demand
150 Housing
demand
100
Pulp & paper
50 demand
0
Base year Low Low High High
housing housing housing housing
low pellet high low pellet high
pellet pellet
2010 2030 2030 2030 2030
Figure 3.2: Demand for wood feedstock from the southeastern US for the US domestic housing
market, the pulp and paper industry, and the wood pellet sector, in 2010 and as assumed in the four
different scenarios (LhLp, LhHp, HhLp, HhHp) in 2030 (Chapter 2).
Species richness data

Species richness is an important determinant of ecosystem function and forest
productivity (Liang, 2016). Ecosystems with higher species richness have shown
to be able to sustain higher levels of multiple ecosystem functions (Lefcheck, 2015;
Soliveres, 2016; Zavaleta, 2010a) and to maintain these ecosystem functions while
subjected to changes in environmental conditions or loss of species (Folke, 1996; Potter,
2014). Threatened species are those considered to be at relatively high risk of (local)
extinction in the near future (Dobson, 1997; Steck, 2007), and as such provide a group
of species of conservation concern (Boyd, 2008). For this study, threatened species
were defined as those species listed as 1) ‘vulnerable’, ‘endangered’ or ‘critically
endangered’ according to the IUCN Red List (IUCN, 2019), and/or 2) listed ’threatened’
or ‘endangered’ by the Endangered Species Act (U.S. Fish and Wildlife Service, 1979), and/
or 3) listed globally in the category of ‘critically imperilled’, ‘imperilled’ or ‘vulnerable’
by NatureServe (Master, 2012). When species consisted of several subspecies with
an individual conservation status, these subspecies were included separately in the
analysis. Endemic species are considered priorities for conservation because of their
uniqueness and extinction risk (Kier, 2001; Laffan, 2003). Endemism is the restriction
of a species’ range to a defined area, meaning it occurs uniquely within the boundaries
of that area (Laffan, 2003). In this study, endemic species were defined as those species
that have their entire potential habitat area within the southeastern US.
62

The GAP dataset includes potential habitat maps for all terrestrial vertebrate species
in the US at 30 x 30 m resolution (McKerrow, 2018). GAP potential species habitat
maps were created using a deductive modelling approach in which potential species
habitat is determined by selecting only areas within its range, i.e. where a species
could be expected to persist according to scientific literature. Potential habitat is
determined by selecting areas within the species’ range that are considered suitable
based on land use (as classified in the GAP land-use map), hydrology, elevation, patch
size, human disturbance, canopy cover and forest edge effects. The GAP land-use
map was derived from 1999-2001 Landsat satellite imagery at 30 x 30 m resolution
and contains 280 different land-use types within our study region (USGS, 2011). The
GAP potential habitat maps of individual species reflect potential distribution. We
generated maps of potential species richness by summing potential habitat maps of
3
individual species (McKerrow, 2018). While these individual GAP species potential
habitat maps were not assessed for accuracy, previous iterations of GAP modelling
efforts related to species potential distribution projects at state and regional level
using nearly identical deductive methodologies, produced accuracy rates of 65- 80%
(Aycrigg, 2015; Maxwell, 2005, 2009; McKerrow, 2006). The latest update of the complete
dataset has undergone a USGS data review and has been included in McKerrow et al.
(McKerrow, 2018). Since abundance data are not available for all species in the study
area, we were not able to assess biodiversity indicators based on species abundance,
such as the Simpson Index (Simpson, 1949), Shannon index (Shannon-Wiener, 1950) or
the Mean Species Abundance (Alkemade, 2009).
Biodiversity indicators
Many different biodiversity indicators exist and no single indicator can measure
the full complexity of biodiversity (Noss, 1990). The impact of wood pellet demand
on biodiversity was quantified in this study using a combination of the following
biodiversity indicators:
• Total potential species richness (of all terrestrial vertebrate species, and of
the individual taxonomic groups of amphibians, birds, mammals and reptiles)
• Threatened potential species richness
• Endemic potential species richness
Species richness is widely used (Fleishman, 2006) and is well established as an
indicator for biodiversity because of its relative ease to calculate and interpret
(Dobson, 1997; Lamb, 2009; Orme, 2005; Williams, 1994). A downside of using total
species richness as a biodiversity metric is its inability to differentiate among species
with different conservation values based on e.g. endemism or sensitivity to threat
(Fleishman, 2006). Therefore, we selected total species subsets using endemic and
63

Chapter 3
threatened species as separate biodiversity indicators. We assessed total potential

species richness, as well as potential species richness specific for different taxa,
because spatial richness patterns and response to environmental and land-use change
have been shown to vary among taxa in the United States (Gregory, 2005; Hof, 2011;
Jenkins, Van Houtan, 2015; McKerrow, 2018). Using species’ habitat distribution data
from the United States Geological Survey’s Gap Analysis Project (GAP) (McKerrow,
2018), all terrestrial vertebrate species (amphibians, birds, mammals, and reptiles)
whose habitat lies within the study area were included in the analysis (See Chapter
2 for a detailed description of the study area). A total of 811 terrestrial vertebrate
species were reported for the study region, including 135 threatened species and 140
endemic species (see Table 3.1). A13 provides a definition and a list of threatened and
endemic species in the southeastern US.
Table 3.1: Species count for total, threatened and endemic species in the southeastern US, separated
per taxonomic group.
Total Threatened Endemic

Mammal 123 36 23
Bird 372 19 3
Reptile 145 33 39
Amphibian 171 47 75
Total taxa 811 135 140
Spatial analysis
Our aim was to estimate changes in potential species richness driven by the projected
land-use change in the southeastern US (Chapter 2). To do this, maps of potential
species richness were combined with maps of land-use types by assigning spatially
variable values of potential species richness (McKerrow, 2018) to each land-use type,
in a spatial analysis described in more detail below and in A15. Due to the difference
in resolution and years of origin of the GAP data (30 x 30 m, 2001) and PLUC land
use output (2 x 2 km, 2010), we had to develop a method to align these two maps
(see Figure 3.3). This method consists of a spatial analysis which assigned values of
potential species richness from the GAP habitat suitability data to land-use projections.
Land use projections were based on modelling with the PLUC model at a county-sized
resolution (on average approximately 1520 km2 or 39 x 39 km) because sensitivity
analyses indicated that PLUC output finer than that resolution was less accurate
(Chapter 2). Our method links two spatial data sets with different spatial resolutions
to produce results at a resolution coarser than either one, while retaining location and
land use-specific information on potential species richness.
64

Potential species richness maps were created by aggregating all individual species
habitat maps from the GAP database (step 1 in Figure 3.3). As a result, potential species
richness of a cell is defined in this study as the number of species for which the cell
contains suitable habitat according to the GAP data. The resulting potential species
richness maps reflect the land use in 2001 as characterized by the GAP program (as
well as other factors including elevation and hydrology). Because the PLUC land-
use projections and GAP habitat maps utilize different land use classifications, we
reclassified the 280 GAP land-use types to match PLUC land-use types (see A14 for a
more detailed description of the reclassification process) (step 2). Spatially variable
land-use specific richness values were then determined by creating separate potential
species richness maps for each land use, using the reclassified GAP land-use map and
the potential species richness map (step 3, see A15 for a more detailed description). A
3
species richness index (SRI) for each PLUC cell (2 x 2 km) for 2010 was then calculated
using a spatial neighbourhood analysis (step 4, see A15 for a more detailed description
of the neighbourhood analysis). The process was repeated for the 2030 PLUC land-use
maps. This resulted in 2 x 2 km resolution maps of SRI for each biodiversity indicator
(total, endangered, and endemic potential species richness and the four taxonomic
groups) and scenario (HhHp, HhLp, LhHp and LhLp scenarios, step 5 in Figure 3.2).
The species richness index map for 2010 created in step 4 was used to assess
differences within and among projected SRI values per land-use type, by calculating
average SRI and ranges of SRI for each land-use type (step 6). Species richness index in
2010 was then compared to species richness index in 2030 for the low pellet demand
scenarios (HhLp and LhLp), to assess the baseline changes in SRI occurring between
2010 and 2030 in the absence of increased wood pellet demand (step 7). This was done
by subtracting the 2010 SRI map from the 2030 SRI maps, creating ΔSRI2010-2030
(see Equation 1). The resulting maps of species richness index change (ΔSRI2010-2030)
were then divided into positive and negative changes, and aggregated (mean) to 39
x 39 km resolution. SRI values in 2030 for scenarios with and without an increased
pellet demand were then compared to determine the differences in species richness
index attributable to increased wood pellet demand, see Equation 2 (step 8). This
was calculated both for the high housing and low housing demand scenarios. Each
of the resulting maps of the differences in species richness index (ΔSRIpellets2030)
was then divided into positive and negative changes, and aggregated (mean) to 39 x
39 km resolution.

65

Chapter 3
Eq. 1) ΔSRI2010-2030 = SRI2030-LP – SRI2010

Eq. 2) ΔSRIpellets2030 = SRI2030-HP – SRI2030-LP
ΔSRI2010-2030: Change in richness between 2010 and 2030 for low wood pellet demand
scenario
ΔSRIpellets2030: Difference in richness between high and low wood pellet demand
scenarios by 2030
SRI2010: Potential species richness index in 2010
SRI2030-LP: Potential species richness index in 2030 in the low wood pellet demand
scenario (HhLp or LhLp)
SRI2030-HP: Potential species richness index in 2030 in the high wood pellet demand
scenario (HhHp or LhHp)
INPUT
Map of modelled land use Species habitat maps GAP 2001 land cover map Map of modelled land use
(for 2010, 2x2 km) (based on 2001 land cover, 30x30 m) (for 2001, 30x30 m) for 4 scenarios (for 2030, 2x2 km)
1) 2)
Species richness map Reclassified land cover map

(based on 2001 land cover, 30x30 m) (for 2001, 30x30 m)
3)
Map of species richness per land-use type

ANALYSIS
(based on 2001 land cover, 30x30m)
4) 5)
Map of SRI per land-use type Map of SRI per land-use type
(for 2010, 2x2 km) for 4 scenarios (for 2030, 2x2 km)
6) 7) 8)
OUTPUT
Richness per Map of SRI changes 2010-2030 Map of SRI differences

land-use type (per scenario/indicator, 39x39 km) due to pellets
(for 2010) (per indicator, 39x39 km)
Figure 3.3: Overview of the method of the spatial analysis to determine species richness index in
2030. Rectangles symbolize maps, while arrows indicate adaptations to the map. SRI = species
richness index.
66

The SRI output maps were aggregated from a resolution of 2 x 2 km to 39 x 39 km to

reflect uncertainty of the input map as well as uncertainty introduced in the analysis.
This means that SRI calculations at 2 x 2 km are based on averages at a 39 x 39 km
window in the analysis, after which the maps are aggregated to a 39 x 39 km resolution
for visualization of the results. Because our spatial processing methods introduced
uncertainty by averaging and aggregating richness, while assuming land-use change to
be the sole driver of changes in potential species richness, our results were expressed
as ‘index’ of species richness. Therefore, species richness index values do not signify
absolute numbers of species projected to be lost or gained at a given location. Rather,
they give an indication of the possible intensity of shifts in potential species richness
that is based on what we know about the spatial distributions of potential species
richness in each of the land cover classes. The unit of the potential species richness
3
index is average number of species per 30 x 30 m, which is the original resolution of the
species habitat maps. However, as mentioned above, this is provided at a 39 x 39 km
resolution. All subsequent mentions of species richness refer to this species richness
index or SRI. When visualizing changes in species richness index, we provided figures
for species richness increases and species richness decreases separately. Due to the
aggregation to a large cell size (39 x 39 km) and the fact that land use transitions
resulting in SRI losses and as well as those resulting in SRI gains coincided in some
areas, both positive and negative changes were projected within some single 39 km
cells. As a result, simply averaging SRI values could result in a loss of information. For
instance, if within a cell large positive shifts and large negative shifts in SRI occur, the
average may even out and result in a low average SRI change.
3.3 Results
The potential impact of increased demand for wood pellets on biodiversity is discussed
in four steps. First, we show spatial variation in potential species richness index in
2010. We then discuss Species Richness Index (SRI) values in 2010 per land-use
type. Next, we discuss projected changes between 2010 and 2030 in SRI without
an increased pellet demand. Finally, we examine the additional impacts due to an
increased demand for wood pellets.
Species richness in the southeastern US

The mean total species richness index in the southeastern US in 2010 was 108 species
(standard deviation = 24), the total SRI ranged from 48 to 183 (see Table 3.2). Of the
four taxonomic groups, birds had the highest SRI (mean = 62) and amphibians the
67

Chapter 3
lowest (mean = 7). The threatened and endemic SRI ranged from 0 to 15 (threatened
species) and 0 to 20 (endemic species). The total SRI in the southeastern US in 2010
was generally highest along the Coastal Plains (Figure 3.4a), with highest values mostly
overlapping with the presence of lowland hardwood forest (Figure 3.4b). Some areas
with high SRI occurred in the western part of the study area, corresponding with
species-rich pasture land (see A16 for SRI patterns per land-use type). The high SRI
in Tennessee and surrounding areas to the south and east are due to high densities of
species rich mixed forest and upland hardwood forest in this area. Low species richness
index in eastern Arkansas, northern Louisiana and western Mississippi reflects the
abundance of cropland. Low SRI in eastern North Carolina and Virginia is the result
of high densities of cropland and pine plantations. Threatened species richness index
was relatively high in the northern part of the study area, in Tennessee and Arkansas
(see A17 for SRI for different indicators). This region contains concentrations of
upland hardwood forest and mixed forest, two land-use types with the highest SRI
of threatened species. Endemic SRI was strongly concentrated in the south-eastern
part of the study region, particularly in Florida and Georgia, a pattern illustrating the
gradient of increasing SRI towards the southeastern part of the study area within all
land-use types (A16). SRI for the different taxonomic groups generally showed similar
patterns. The patterns described above were more pronounced for amphibians, which
showed high SRI values in lowland hardwood forest but much lower values in other
land-use types. Mammal SRI was more concentrated in the northern part of the study
region due to high density of upland hardwood forests.
Table 3.2: Minimum (min), maximum (max) and mean potential species richness index in the
southeastern US in 2010 for seven biodiversity indicators. S.d. = standard deviation.
Statistics Mean Min Max S.d.

Threatened 6 0 15 3
Endemic 2 0 20 4
Mammal 22 5 39 6
Bird 62 27 104 10
Reptile 16 3 43 7
Amphibian 7 0 36 7
Total 108 48 183 24
68

a) b)
Urban Pine plantation

Water Natural pine forest
Total species richness index Protected area Mixed forest
Cropland Upland hardwood forest
Pasture Lowland hardwood forest
Non-forest vegetation
Figure 3.4a: Projected total species richness index in the southeastern US for 2010. White areas contain Federal lands or open water, which were not projected
to change over time and were not included in the biodiversity analysis. Figure 3.4b: Land use in the southeastern US for 2010. Black lines show state bound-
aries, abbreviations indicate state names. AL = Alabama, AR = Arkansas, FL = Florida, GA = Georgia, LA = Louisiana, MS = Mississippi, NC = North Carolina,
OK = Oklahoma, SC = South Carolina, TN = Tennessee, TX = Texas, VA = Virginia.
69
3
16-03-20 12:27
Chapter 3
Potential species richness in land-use classes

Total species richness index was highest in lowland hardwood forests (mean = 156)
followed by the other natural forest types (natural pine (mean = 115), mixed forest
(mean = 117), upland hardwood forest (mean = 118)) and pasture (mean = 117; Figure
3.5 and Table 3.3). Natural forests (apart from lowland hardwood forest) and pasture
showed comparable mean and range in potential species richness index. As a result,
land use changes from forest to pasture are projected to result in either increased
or decreased potential species richness, depending on the potential species richness
values for these land use types in a particular area. Natural forest types showed a
large range of total SRI over the study area (range = 52 (minimum mixed forest) –
183 (maximum lowland hardwood)), reflecting the spatial variation between and
within these land-use types. Non-forest vegetation had the largest range of total SRI
(range = 48 – 172). Urban area showed relatively high total species richness index
compared to the other anthropogenic land-use classes (mean = 103). Cropland and pine
plantation had the lowest total SRI (cropland (mean = 81), pine plantation (mean = 79).
Pine plantation showed little variation in total SRI (range = 64 – 93). Similar patterns
of total species richness index per land-use type were found for the taxonomic groups
(Table 3.3, A18). Total SRI of pine plantations and cropland was low, this was due
to low index values for all taxonomic groups. Lowland hardwood forest had high
total SRI across all taxonomic groups, especially amphibians. Bird species showed a
slightly different pattern compared to other taxonomic groups, with relatively high
SRI in urban areas, cropland, pasture and non-forest vegetation. In fact, bird species
richness index was so high in urban areas that a transition from forest to urban in
some locations led to a projected positive change in potential bird SRI. The highest SRI
for both threatened and endemic species was found for natural forest types, with the
exception of endemic species in upland hardwood forests, which was relatively low.
The lowest SRI values for threatened and endemic species occurred in urban areas,
cropland and pine plantation.
70

200
180
160
Total species richness index
140
120
100
80
60
40
20 3
0
urban cropland pasture non-forest pine natural mixed upland lowland
vegetation plantation pine forest hardwood hardwood
Land use type
Figure 3.5: Potential total species richness index per land-use type in the southeastern US for 2010.
Boxplots show median values (middle horizontal bar) and first (upper limit) and third quartile (lower
limit) values, while vertical bars show maximum and minimum values.
Species richness changes between 2010 and 2030 without additional wood
pellet demand
We assessed the (positive and negative) changes in potential species richness index
between 2010 and 2030 due to drivers other than an increasing wood pellet demand
to assess which changes in species richness index would be projected to occur even
in the absence of projected increase in wood pellet demand. The changes in total SRI
ranged from -21 to 37 (mean increase = 1.1, mean decrease = -2.0) for the high housing
scenario (HhLp). Changes in threatened species richness index ranged from -2 to 1
(mean increase = 0.1, mean decrease = -0.2), while changes in endemic species richness
index ranged from -4 to 5 (mean increase = 0.1, mean decrease = -0.3).
Decreases in total, threatened and endemic species richness index were projected to
occur along the Gulf Coast, particularly in Mississippi, Alabama and Florida, as well as
along the Atlantic Coast, particularly in South Carolina (Figure 3.6). These decreases
are mainly due to decreases in reptile species richness index (Figure 3.7). Decreases in
total SRI were also projected for the area around the Ouachita Mountains in Arkansas,
this is mainly due to a decrease in SRI of birds and reptiles.
71

Table 3.3: Mean and standard deviation (S.d.) of species richness index within land-use classes of the southeastern US for seven biodiversity indicators.
72
Pine Natural Mixed Upland Lowland Non-forest
Species richness Statistic Urban Cropland Pasture
Chapter 3
plantation pine forest hardwood hardwood vegetation

Mean
Total
S.d.
103.0 81.1 116.9 79.3 114.5 117.2 118.1 156.4 105.6
Mean
9.0 5.9 8.4 4.6 16.3 14.1 6.9 17.2 10.8
Threatened
S.d.
3.1 3.5 5.8 4.5 8.2 8.5 9.1 7.6 4.7
Mean
0.8 1.1 1.1 1.3 2.3 1.7 1.7 1.6 1.5
Endemic
S.d.
0.4 0.4 1.3 0.4 5.9 3.3 1.0 6.3 2.5
Mean
0.7 0.7 2.3 0.7 4.9 4.4 2.6 6.5 3.2
Mammal
S.d.
20.3 13.1 22.6 18.1 22.6 26.3 29.8 29.7 19.2
Mean
2.0 2.4 2.0 1.7 3.2 3.8 2.5 2.8 5.9
Bird
S.d.
67.3 57.1 71.1 50.1 57.4 59.3 59.8 76.7 69.0
Mean
6.0 4.1 4.3 3.4 7.0 5.9 4.4 5.5 10.8
Reptile
S.d.
11.5 9.3 17.1 9.2 26.4 22.2 18.6 26.6 13.0
Mean
1.8 1.2 2.2 0.9 5.4 6.1 3.5 5.9 4.0
Amphibian
S.d.
4.0 1.6 6.1 1.9 8.1 9.4 9.9 23.5 4.4
1.1 0.9 2.4 0.8 3.4 5.6 2.0 6.2 3.4
16-03-20 12:27
Additionally, threatened species richness index was projected to decrease in the

Appalachian area on the eastern borders of Tennessee, this was likely primarily
due to a lower SRI of threatened mammals and reptiles. The most common land
use transitions between 2010 and 2030 for the high housing demand and low pellet
demand scenario (HhLp) were transitions from pasture and natural forest to urban
area, pasture and natural forest to (other types of) natural forest, natural forest and
pasture area to pine plantations and non-forest vegetation to pasture area (Chapter
2)(Duden, 2017). The most common land use transitions between 2010 and 2030 that
resulted in a reduced SRI were transitions from upland hardwood forest and pasture
to urban, and from pasture to pine plantation (see A20 for more information on specific
land use transitions and related changes in SRI).
3
Increases in SRI were generally low in the study area, apart from the southern tip of
Florida, where relatively larger increases in species richness index were projected. This
was due to increasing species richness index for all taxonomic groups. Southwestern
Louisiana showed a decrease in bird SRI, but an increase in mammal, reptile and
amphibian SRI (Figure 3.7). Results for the low housing scenario (LhLp) showed a
similar pattern (A19), indicating the same areas where the intensity of SRI changes
is relatively high. Increases in total species richness index were the result of land-use
transitions into non-forest vegetation (from cropland, pasture and pine plantation)
and to natural forest (from non-forest vegetation, pasture and other natural forest
types). Some transitions to urban area also resulted in some areas showing positive
changes in total SRI, particularly for birds.
73

Chapter 3
Scenario: High housing demand

Change in potential species richness 2010-2030 due to other drivers
(2030HhLp – 2010 richness)
Increase Decrease
(index) (index)
Total species
Threatened species
Endemic species
Figure 3.6: Projected changes in species richness index (ΔSRI2010-2030 = SRI2030-LP – SRI2010)

between 2010 and 2030 without additional wood pellet demand for different species richness indica-
tors at 39 x 39 km resolution for the high housing scenario. Positive (ΔSRI>1) and negative (ΔSRI<1)
changes were separated before aggregation. Increase = window (1500 km2) average of positive
changes in SRI. Decrease = window (1500 km2) average of negative changes in SRI. Grey cells show
areas where no change in species richness were projected to occur. AL = Alabama, AR = Arkansas,
FL = Florida, GA = Georgia, LA = Louisiana, MS = Mississippi, NC = North Carolina, OK = Oklahoma,
SC = South Carolina, TN = Tennessee, TX = Texas, VA = Virginia.
74


Change in potential species richness 2010-2030 due to other drivers
(2030HhLp – 2010 richness)
Increase Decrease
(index) (index)
Mammal species
3
Bird species
Reptile species
Amphibian species
Figure 3.7: Projected changes in species richness index (ΔSRI2010-2030 = SRI2030-LP – SRI2010) per
taxonomic group between 2010 and 2030 without additional wood pellet demand for the high hous-
ing scenario. Positive (ΔSRI>1) and negative (ΔSRI<1) changes were separated before aggregation.
Increase = window (1500 km2) average of positive changes in SRI. Decrease = window (1500 km2)
average of negative changes in SRI. Grey cells show areas where no change in richness was projected
to occur. AL = Alabama, AR = Arkansas, FL = Florida, GA = Georgia, LA = Louisiana, MS = Mississippi,
NC = North Carolina, OK = Oklahoma, SC = South Carolina, TN = Tennessee, TX = Texas, VA = Virginia.
75

Chapter 3
Impact of wood pellet demand on biodiversity

We assessed the projected difference in potential species richness index by 2030
between the scenarios with and without an increased wood pellet demand (see
equation 2). Projected differences in total species richness index due to an increased
wood pellet demand ranged from -11 to 5 (mean increase = 0.4, mean decrease = -1.0)
for the high housing (Hh) scenarios. For the low housing (Lh) scenarios, difference in
total SRI ranged from -23 to 33 (mean increase = 0.7, mean decrease = -1.4) (see A21 for
results of the low housing scenario). Small differences in SRI were found throughout
the study region, particularly in the Coastal Plains (Figure 3.8), while some areas show
comparatively larger differences due to an increased pellet demand. In the high housing
scenario, impacts of increased wood pellet demand lead to relatively large decreases
in total SRI in patches in the southern tip of Florida, due to low richness index of bird,
reptile and amphibian species (Figure 3.9). Threatened species richness index, on the
other hand, showed gains in this area, which is mainly due to high richness index values
of mammal species. Endemic SRI is also low in the southern tip of Florida when wood
pellet demand increases. Species richness index was also low due to increased wood
pellet demand in the coastal areas of northeastern Florida and southern Alabama.
This is mainly caused by a low SRI of reptiles.
When the demand for housing is assumed to be low, a small area of eastern Oklahoma
shows relatively large positive differences in total SRI by 2030 due to increased wood
pellet demand (see A21). This area showed high species richness index of birds under
the high pellet and low housing scenario. Threatened species richness index in the
Ouachita area in Arkansas is also relatively high, coinciding with high mammal and
bird SRI. In the low housing scenario, relatively large negative differences were found
in total, threatened and endemic species richness index in coastal Virginia –coinciding
with low species richness index of birds - and North Carolina – coinciding with low
species richness index of mammals and amphibians. Negative differences in total SRI
increased demand for wood pellets were mainly the result of transitions from natural
forest, pasture and non-forest vegetation to pine plantations in the eastern and southern
coastal regions, as well as transitions from species rich pasture to natural forest in the
western states in the study area (see A22 for an overview of land use transitions for
the different pellet scenarios). Shifts from natural forest to pine plantation occurred
throughout the study area but were more prevalent in southern Alabama and Mississippi,
as well as in North and South Carolina. Positive differences in total SRI were caused
by transitions to natural forest and pasture occurring along the Coastal areas. The
differences in SRI values were also high in some areas where a transition from natural
pine forest and mixed forest to hardwood forest was projected to occur.
76


Change in potential species richness 2010-2030 due to increasing pellet demand
(2030HhHp – 2030HhLp richness)
Increase Decrease
(index) (index)
Total species
3
Threatened species
Endemic species
Figure 3.8: The difference in species richness index (ΔSRIpellets2030 = SRI2030-HP – SRI2030-LP)

for total, threatened and endemic species richness by 2030 due to increased wood pellet demand for
the high housing scenario. Positive (ΔSRI>1) and negative (ΔSRI<1) changes were separated before
aggregation. Increase = window (1500 km2) average of positive changes in SRI. Decrease = window
(1500 km2) average of negative changes in SRI. Grey cells show areas where no change in richness
was projected to occur. AL = Alabama, AR = Arkansas, FL = Florida, GA = Georgia, LA = Louisi-
ana, MS = Mississippi, NC = North Carolina, OK = Oklahoma, SC = South Carolina, TN = Tennessee,
TX = Texas, VA = Virginia.
77

Chapter 3

Change in potential species richness 2010-2030 due to increasing pellet demand
(2030HhHp – 2030HhLp richness)
Increase Decrease
(index) (index)
Mammal species
Bird species
Reptile species
Amphibian species
Figure 3.9: Projected differences in species richness index (ΔSRIpellets2030 = SRI2030-HP – SRI2030-

LP) per taxonomic group by 2030 with additional wood pellet demand for the high housing sce-
nario. Positive (ΔSRI>1) and negative (ΔSRI<1) changes were separated before aggregation. In-
crease = window (1500 km2) average of positive changes in SRI. Decrease = window (1500 km2)
average of negative changes in SRI. Grey cells show areas where no change in richness was projected
to occur. AL = Alabama, AR = Arkansas, FL = Florida, GA = Georgia, LA = Louisiana, MS = Mississippi,
NC = North Carolina, OK = Oklahoma, SC = South Carolina, TN = Tennessee, TX = Texas, VA = Virginia.
78

3.4 Discussion
An increasing demand for wood pellets was projected to cause both positive and
negative biodiversity impacts in the southeastern US, and these impacts were location
specific. The effect of an increased wood pellet demand on the species richness index
was projected to be relatively small and was not restricted to a particular area. We
found that the SRI was relatively high in the Coastal Plains, particularly for the land-
use type lowland hardwood forest. The Coastal Plains were also the region in which
biodiversity impacts were likely to occur. Our study therefore agrees with earlier
findings indicating the Coastal Plains as a region of potential biodiversity impacts
of wood pellet demand (Galik, 2016; Olesen, 2016). Negative shifts in total SRI were
projected to occur in southern Florida, along the coast of Virginia and North Carolina
3
and along parts of the Gulf Coast. Positive shifts in total SRI were projected to occur
in parts of Oklahoma and Arkansas. Threatened species showed positive shifts in SRI
values in Tennessee, while endemic species index values showed negative shifts in
coastal South Carolina.
Some of the areas showing large shifts in SRI do not source wood pellets. The projected
land use change in these areas is the result of indirect land use changes following wood
pellet demand driven expansion of natural and planted forest. The area of negative
SRI changes was larger for reptiles than for birds, mammals and amphibians. In some
areas, the direction of change differed per taxonomic group, and positive shifts were
projected for one taxonomic group while negative shifts were projected for the other.
The coastal area of Virginia in particular showed potential SRI shifts for several
taxonomic groups due to increased pellet demand when housing demand was assumed
to be low; positive impacts for mammals and reptiles and negative impacts for birds.
When housing demand was assumed to be high, the southern tip of Florida showed
potential for positive biodiversity impacts for mammals but potential negative impacts
for birds, reptiles and amphibians. This shows that impacts on biodiversity differ per
taxonomic group and highlights the importance of analysing different taxonomic
groups in biodiversity impact assessments.
When compared to the changes due to other drivers, including urbanization and
increased demand for timber, the shifts in potential species richness due to an
increased demand for wood pellets were small. Furthermore, the differences in SRI
between scenarios of high and low demand for wood pellets were smaller when the
demand from the housing market was assumed to be high. This shows that the impact
of wood pellet demand on biodiversity is influenced by developments in other sectors.
79

Chapter 3
The scenarios for wood pellet demand run from 2010 to 2030 (Chapter 2). Between
2010 and the present, actual wood pellet demand has been closest to the trajectory of
the ‘high pellet demand’ scenario.
We found that, depending on the location, a specific land-use transition can lead to
both increases and decreases in species richness index. This was for example the case
for transitions from pasture to non-forest vegetation and from pasture to natural pine
forest. This finding emphasizes the added value of a spatially explicit assessment of
biodiversity impacts. Areas of positive biodiversity impacts overlapped partly with
areas of negative impacts due to data aggregation. In these areas, land-use transitions
that result in increases in SRI are projected to occur alongside transitions that result
in decreases in SRI. Land-use transitions from natural timberland forest to urban area
and pine plantations are responsible for the majority of the negative shifts in total
SRI. Of these transitions, conversions from natural pine forest to pine plantation and
from upland hardwood forest to urban area are most prevalent. Conversely, positive
changes in SRI occurred when land was converted to natural forest or pasture, with
transitions of non-forest vegetation and cropland to pasture being most common.
These transitions accounted for about half of the areas with SRI increases. Therefore,
we recommend land managers and planners to minimize biodiversity impacts by
avoiding conversion of natural habitats to pine plantations, i.e. by focusing wood pellet
sourcing from existing plantations.
The aggregation of data in this study has influenced the results. The PLUC land-use
classification of the GAP data (280 land-use types) was aggregated to the classification
of the land-use projections (11 land-use types). The number of land-use types included
in an analysis has shown to influence the assessment of species richness (Lawler, 2004).
Reclassification was however unavoidable, as it was unfeasible to create land-use
projections of all 280 land-use types up to 2030. Reclassification resulted in a loss
of detail, but the variation in potential species richness between the GAP land use
classes is included indirectly in the analysis by using local averages of potential species
richness for that land use. Furthermore, land use projections contain high levels of
uncertainty, that may influence subsequent environmental impact assessments
(Prestele, 2016). Land use projections were found to be most accurate for the types
lowland hardwood forest, urban and cropland, results were less accurate for mixed
forest, natural and plantation pine forest, pasture and non-forest vegetation. The
accuracy of PLUC is reduced with increasing spatial resolution and sensitivity analyses
indicated that PLUC output at a resolution higher than a county-sized level is less
accurate (Chapter 2). This necessitated the aggregation of species richness index to a
80

window size comparable to the average county in the study region (about 1500 km2),
in order for the visualization of the results to be consistent with the uncertainty in
the land use projections. This resulted in a loss of detail. Species habitat maps also
introduce additional uncertainty, previous GAP modelling efforts had accuracy rates
of 65-80%. The effects of uncertainty of habitat maps on the overall pattern of our
results may however be limited, due to aggregation of the potential habitat maps from
30 m to 1500 km2.
Data limitations necessitated a number of assumptions in the analysis. Firstly, we

assume that the species richness index at a certain location does not vary over time
if the land use remains the same. However, species’ ranges could shift over time due
to drivers other than land-use change, most notably due to climate change. Future
3
research may want to incorporate both land use change and climate change as drivers
of biodiversity changes (Titeux, 2016). The temporal scope of this study is short, so no
large changes due to climate change are expected. Secondly, the method applied in this
study is based on the assumption that if land-use changes in a certain location, changes
in species richness index will reflect the difference in richness of the two land uses in
that location. By using land use as the determinant factor for species richness, issues
of landscape heterogeneity and connectivity are not included. However, landscape
configuration is often noted as an important determinant of regional species richness
(Dale, 2000; Dauber, 2003). In addition to land use and landscape configuration, other
factors besides land use and landscape configuration, such as hydrology or human
disturbance, are expected to affect presence or absence of species. These factors
were not explicitly modelled in this analysis but were considered implicitly, as they
are expected to influence local spatial variation in species richness index values.
Furthermore, these factors are assumed to stay stable over time between 2010
and 2030. Finally, the species specific time-lag, for instance due to regeneration or
migration, necessary for the area to actually reach that potential species richness,
is disregarded in this approach. This is a disadvantage of using potential rather than
actual species richness. However, the GAP database provides uniform biodiversity data
in the form of potential habitat maps for a large geographical area and allows us to
include a large number of species in the analysis. Field observations of species richness
before and after land-use transitions would be helpful to validate the assumptions
made in this study.
This study used potential species richness index as an indicator of biodiversity. Species
richness alone is often considered to be insufficient as an indicator to determine
priority areas of conservation because it provides no information of species’ functional
81

Chapter 3
roles and contribution to ecosystem processes, which is necessary to reflect the

concept of biodiversity (Fleishman, 2011; Lamb, 2009). Furthermore, two cells in a
map showing a similar value of potential species richness may differ completely in
species composition, i.e. while alpha diversity is similar, beta diversity may be very
different. In addition, a location that shows similar potential species richness over time
may in fact have undergone changes in species composition due to species turn-over.
This is an inherent shortcoming of using potential species richness as an indicator
for biodiversity. However, by also including selected groups of species in this study
(according to taxonomy, level of threat and endemism), we are able to provide some
information on which species groups may be impacted in which areas. In areas where
biodiversity shifts due to wood pellets are projected, future studies should assess
actual species richness and habitat suitability at a higher resolution. Some projected
changes in SRI may partly reflect the assumptions about heterogeneity within land
cover classes in the GAP data. For example, the low species richness in pine plantations
reflects the assumption that all pine plantations that are identifiable with remotely-
sensed imagery are intensively managed, as many unmanaged plantations would
have mixed pine-hardwood canopies. In reality, pine plantations in the region have
a range of management intensities. Unfortunately, we were unable to include the
impacts of intensification of existing management plantations in our analysis, because
projections of intensification of existing plantations for the study region were not
available. Furthermore, the GAP habitat suitability models do not distinguish between
different management intensities in pine plantations. Forest management can have
either positive effects, for instance due to thinning (Gottlieb et al., 2017; Verschuyl,
Riffell, Miller, & Wigley, 2011), or neutral or negative impacts, for instance due to
residue removal (Fritts, 2016; Gottlieb, 2017; Grodsky, Moorman, Fritts, Castleberry,
2016; Grodsky, Moorman, Fritts, Hazel, 2016; Riffell, 2011a)(Fritts et al., 2016; Gottlieb
et al., 2017; Grodsky, Moorman, Fritts, Hazel, et al., 2016; Grodsky, Moorman, Fritts,
Castleberry, & Wigley, 2016; Riffell, Verschuyl, Miller, & Wigley, 2011b) on species
richness. Therefore, impacts on species richness due to pine plantation establishment
may be overestimated if new plantations are not managed intensively. On the other
hand, urban areas had a relatively high species richness compared to the other
anthropogenic land-use classes, which may be explained by the fact that the land-
use type ‘urban’ includes the GAP land-use types ‘developed: low intensity’ and
‘developed: open space’, which includes suburban areas that contain a variety of
vegetative communities and structures, including very small patches of forest. ‘Non-
forest vegetation’ had the largest range of SRI, a reflection of the land-use type’s
heterogeneous composition which includes non-forest and non-anthropogenic land
uses such as marshes, dunes and grasslands. This means that a land-use transition
82

involving non-forest vegetation can result in very different changes in SRI, depending
on the location.
The spatially explicit method developed and presented in this study provides a way
to combine available spatial datasets of land-use projections and biodiversity. The
method is applicable to different regions and for different drivers of land-use or
biodiversity changes. Because of the uncertainty and aggregation of data in this study,
it is not possible to provide solid answers about the exact magnitude and location of
potential biodiversity impacts related to developments in the wood pellet market in
the southeastern US. More data and research is therefore required to serve as a basis
for policy development. However, our study provides trends in biodiversity related
to increased wood pellet demand. These trends do not indicate strong additional
3
biodiversity impacts compared to the impact of other drivers, apart from a small
number of areas where higher changes in potential species richness were projected,
including the coastal area of Virginia and the southern tip of Florida. These results
provide information that allows policy makers, industry and NGOs to identify areas of
concern where appropriate mitigation measures may need to be taken to limit negative
biodiversity impacts, and areas where positive impacts can be promoted. Positive
impacts could be promoted by stimulating land-use transitions that may increase
biodiversity in particular areas such as the southern tip of Florida. Avoiding negative
impacts could be achieved through restrictions of conversion of lowland hardwood
forest, particularly in high biodiversity areas such as the Coastal Plains. Our results
show that it is not straightforward to determine whether wood pellet production
is good or bad news for biodiversity. Biodiversity impacts are location specific and
different taxonomic groups may show different responses. Because biodiversity
impacts are spatially variable, it will be vital to monitor the origin of wood pellets
that are being produced in the southeastern US in order to quantify wood pellet related
biodiversity impacts. Shifts in species richness index values due to wood pellet demand
were considerably smaller than biodiversity impacts caused by trends in other sectors.
Therefore, biodiversity impacts due to wood pellet demand should be considered in
the context of other drivers of land-use change and biodiversity loss.
83

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
4
Biodiversity
and hydrology impacts of of
impacts increased ethanol
production in Brazil
increased biomass Environmental
demand
Anna Duden
A.S. Duden, P.A. Verweij, A.P.C. Faaij, D. Baisero, C. Rondinini, F. van der Hilst
This chapter has been published in the special issue Bioenergy and Land of the journal
Land.
Anna Duden

Chapter 4
Abstract
Growing domestic and international ethanol demand is expected to result in increased

sugarcane cultivation in Brazil. Sugarcane expansion currently results in land-use
changes mainly in the Cerrado and Atlantic Forest biomes, two severely threatened
biodiversity hotspots. This study quantifies potential biodiversity impacts of increased
ethanol demand in Brazil in a spatially explicit manner. We project changes in potential
total, threatened, endemic and range-restricted mammals species richness up to 2030.
Decreased potential species richness due to increased ethanol demand in 2030 was
projected for about 19,000 km2 in the Cerrado, 17,000 km2 in the Atlantic Forest and
7,000 km2 in the Pantanal. In the Cerrado and Atlantic Forest, biodiversity impacts of
sugarcane expansion were mainly due to direct land-use change, in the Pantanal they
were largely due to indirect land-use change. The biodiversity impact of increased
ethanol demand was projected to be smaller than the impact of other drivers of land-
use change.
This study provides a first indication of biodiversity impacts related to increased

ethanol production in Brazil, which is useful for policy makers and ethanol
producers aiming to mitigate impacts. Future research should assess the impact of
potential mitigation options, such as nature protection, agroforestry or agricultural
intensification.
86

Biodiversity impacts of increased ethanol production in Brazil
4.1 Introduction
Brazil is one of the major producers and exporters of agricultural and forestry products
in the world (FAO, 2019; OECD, 2015). Crop production in Brazil is dominated by the
production of sugarcane (30% of net production value), soybean (29%) and corn (4%)
(FAO, 2019) and the area dedicated to the cultivation of these crops has increased
rapidly over the last 30 years (FAO, 2019). Brazil currently has the largest area of
sugarcane cultivation in the world (FAO, 2019), and is the second largest producer
of ethanol (IEA, 2018). The historical increase in sugarcane cultivation is primarily
the result of Brazilian policies focused on stimulating the production of sugarcane-
based ethanol in order to increase energy security, promote rural development and
decrease the dependency on fossil fuels (Verdade, 2012). Due to growing domestic and
international demand, Brazilian ethanol production is expected to increase from 33.3
billion litres in 2018/2019 (UNICA, 2019) up to 54.2 billion litres in 2030 (van der Hilst,
4
Verstegen, 2018) based on the global outlook of IEA and OECD (IEA, 2014).
Today, sugarcane is grown mainly in the Cerrado and Atlantic Forest biome in the
southeast of Brazil, and has recently also expanded further into the northwestern
part of the Cerrado biome (Verdade, 2012; Walter, 2014). The projected increase in
ethanol production is expected to require additional land for sugarcane cultivation,
resulting in changes in land use which are both direct, when sugarcane replaces
an area previously occupied by another land use, or indirect, when the expansion
of sugarcane induces changes in land use elsewhere (Adami, 2012; Lapola, 2010; van
der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). About 35 thousand km2 of
sugarcane expansion is projected to occur between 2012 and 2030, mostly in the
Cerrado and Atlantic Forest biomes (van der Hilst, Verstegen, 2018; Verstegen, van der
Hilst, 2016). Sugarcane is expected to replace mainly cropland and natural grassland
(Lapola, 2010; van der Hilst, Verstegen, 2018), as well as shrubland (Verstegen, van
der Hilst, 2016). Additionally, the projected expansion of sugarcane is expected to
result in indirect land-use changes affecting an area ranging from about 20 thousand
km2 (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016) to 78 thousand km2
(Lapola, 2010). The three biomes where direct and indirect land use change related to
sugarcane expansion were projected were the Cerrado, Amazon, and Atlantic Forest
biomes (Lapola, 2010; van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016),
resulting in loss of forest (Lapola, 2010; van der Hilst, Verstegen, 2018) as well as other
natural vegetation (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). The
Cerrado and Atlantic Forest biomes are both biodiversity hotspots (Myers, 2000) under
severe threat (Hunke, 2015; Sano, 2010).
87

Chapter 4
Brazil is recognized as one of the most biodiverse countries in the world (IPBES, 2018;
Jenkins, Alves, 2015) and contains about 15-20% of all species worldwide (BPBES, 2019).
Mammal species are vulnerable to habitat loss, and Brazil has been identified as one
of the top 10 of countries with high projected mammal decline by 2050 (BPBES, 2019;
Visconti, 2011). The tropical and semi-deciduous Atlantic Forest is home to a large
number of endemic species ( Jenkins, Alves, 2015). Currently, only about 12% of the
original vegetation cover of Atlantic Forest is left, and the remaining forest areas
are highly fragmented (Beca, 2017; de Abreu Bovo, 2018). The Cerrado biome consists
of savannah vegetation and contains a large share of endemic plant and vertebrate
species (Diniz-Filho, 2008; Strassburg, 2017). Only about 2% of the Cerrado biome is
legally protected (Carvalho, 2009; Klink, 2005), and over 40% of its original extent has
been converted to agricultural use (Hunke, 2015; Klink, 2005; Zuurbier, 2008). Almost
70% of the Amazon rainforest falls within the borders of Brazil (Kirby, 2006), and the
Brazilian Amazon forest has lost 18% of its original extent and has recently shown the
highest levels of deforestation in a decade (MMA, 2018). For a quarter of all Amazonian
mammal species, further deforestation of the Amazon is projected to result in the loss
of over 40% of their range by 2050 (Soares-Filho, 2006).
Past expansion and intensification of agriculture have been identified as the main
driver of species decline, and with ongoing agricultural expansion, biodiversity
impacts are expected to become exacerbated in the future (Norris, 2008). Conversion
of native vegetation to agriculture is one of the main drivers of habitat loss in the
tropics (Gibbs, 2008; Green, 2005). Expansion of sugarcane leads directly or indirectly
to habitat loss (van der Hilst, Verstegen, 2018), and has therefore been linked to impacts
on biodiversity (Naylor, 2007). A number of studies have projected substantial future
biodiversity impacts of agricultural expansion in Brazil. For example, studies assessing
the projected effects of different scenarios of agricultural expansion pinpoint the
Cerrado and Atlantic Forest as regions with large potential range contractions for a
number of bird species ( Jetz, 2007) and non-flying mammals (Faleiro, 2013) by 2050,
and potentially resulting in local extinction of 140–191 bird and mammal species
(Kennedy, 2016) as well as 480 endemic plant species (Strassburg, 2017). However, it
remains unclear which proportion of these impacts can be attributed specifically to
an increase in ethanol demand and subsequent sugarcane expansion, and how these
biodiversity impacts vary over different regions in Brazil. Despite the strong projected
increase in sugarcane production, the historical impact of agricultural expansion in
Brazil and the importance of Brazil in global biodiversity conservation, it has been
noted that there is a paucity of studies that assess the potential impact of sugarcane
expansion on biodiversity in Brazil (Walter, 2014). The impact of ethanol-driven
88

sugarcane expansion on biodiversity has therefore been identified as a priority for

future research (Verdade, 2012).
Biodiversity within a land-use type was found to vary spatially due to spatial
heterogeneity in biophysical conditions (Chapter 3), which implies that a conversion of
one land use type to another may result in different impacts on biodiversity, depending
on the location. Therefore, it is important to assess potential biodiversity impacts
in a spatially explicit manner. The aim of this study is to assess potential impacts of
ethanol-driven sugarcane expansion on biodiversity in Brazil. To this end, we perform
a spatially explicit assessment of changes in potential mammal species richness
following scenarios of land-use change up to 2030. We focus on total, threatened,
endemic and range-restricted mammal species, and include the whole of Brazil in order
to assess biodiversity impacts of both direct and indirect land-use changes related to
increased ethanol demand.
4
4.2 Materials and Methods
General approach
This study combines land-use change projections for Brazil from previous studies (van
der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016) with species distributions
derived from habitat suitability data (Rondinini, 2011) in order to assess impacts of
ethanol-driven land-use change in Brazil on potential mammal species richness.
The spatial modelling approach developed in Chapter 3 was applied to determine

potential mammal species richness in 2012 and 2030 for a reference and an ethanol
scenario. In the ethanol scenario, demand for ethanol is assumed to increase to meet
global ethanol mandates from 24×109 litres in 2012 to 54×109 litres in 2030, while
in the reference scenario, ethanol demand remained static at the 2013 level. Both
scenarios also included changes in demand for other agricultural and silvicultural
commodities. In this analysis, we combined land-use projections (at 5 km resolution)
and species habitat suitability maps (at 300 m resolution) to create maps of projected
species richness index (at 5 km resolution) (Figure 4.1). Habitat suitability maps for
all individual Brazilian mammal species were summed to produce a map of potential
mammal species richness in Brazil (Figure 4.1).
We determined potential species richness for total, threatened, endemic and range-
restricted mammal species. Potential species richness in 2012 was compared to
89

Chapter 4
potential species richness in 2030 in the reference scenario to determine biodiversity

impacts due to increased demand for agricultural and silvicultural commodities other
than ethanol. Potential species richness in 2030 in the ethanol scenario was compared
to potential species richness in 2030 for the reference scenario in order to quantify
the impact of increased ethanol demand on biodiversity.
Biodiversity indicators
Species richness is a frequently used indicator for biodiversity status (Gestich, 2019;
Myers, 2000; Sierra, 2018). In this study, we assess biodiversity impact expressed as
changes in potential species richness of mammals. We focus on species richness of
mammals because habitat loss and degradation have impacted mammal communities
in such a way that globally, 25% of all mammal species are currently threatened with
extinction (Díaz, 2019; IUCN, 2019; Visconti, 2011). Mammals are also one of the most
intensively studied taxa (Rondinini, 2011), for which spatial habitat data is available
globally at a high resolution (Rondinini, 2011). Furthermore, mammal species can
modify vegetation structure and thereby influence species diversity and composition
in other taxonomic groups (Sinclair, 2003). Land-use changes may affect different
mammal species in different ways. In this study, we used the following potential
mammal species richness indicators: total species richness, threatened species
richness, endemic species richness, and restricted range species richness. We used
potential threatened, endemic and range restricted species richness as indicators
because these species are considered conservation priorities due to vulnerability
and high extinction risk. Threatened species are considered to be already at risk of
extinction, and are therefore considered a conservation priority (Given, 1993). Endemic
species are those species whose distribution is restricted to a unique geographical area
(Laffan, 2003) and are considered conservation priorities because of their uniqueness
and high risk of extinction (Kier, 2001; Laffan, 2003). Species with restricted ranges
are considered a conservation priority because they are more vulnerable to human
impact in comparison to species with large ranges (Ceballos, 2006).
We selected all mammal species whose potential habitat overlapped partly or entirely
with Brazil, which was the case for 610 species (Table 1; for a list of species see Annex
28). Threatened mammal species were defined as ‘Critically Endangered’, ‘Endangered’
or ‘Vulnerable’ according to the IUCN Red List of Threatened Species (IUCN, 2019)
or the Brazilian National List of Threatened Fauna Species (Lista Nacional Ofical de
Espécies da Fauna Ameaçadas de Extinção (MMA, 2014)). Endemic species were defined
as having 100% of potential habitat within Brazil, in line with previous research
( Jenkins, Alves, 2015). Restricted range species were identified as those species with
90

a total potential habitat area smaller than 50,000 km2, in accordance with earlier
studies (Eken, 2004; Rodrigues, 2004). Based on our definition, there is an overlap in
threatened, endemic and range-restricted species. This means that a certain species
can, for example, be both threatened and range-restricted.
Table 1. Numbers of mammal species included in this study9; total number of species, threatened
species, endemic species and range-restricted species in Brazil, as derived from habitat suitability
data (Rondinini, 2011).
Species group indicator Number of species

All mammal species 610
Threatened mammal species 107
Endemic mammal species 93
Restricted-range mammal species 120
Land-use projections
4
The projected land-use changes in Brazil following scenarios of future ethanol demand
and developments in demand for other agricultural an silvicultural commodities, were
taken from previous studies (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst,
2016). We compared two scenarios: an ethanol scenario and a hypothetical reference
scenario. The ethanol scenario assumes fulfilment of current and planned ethanol
mandates worldwide, which results in Brazilian ethanol production to more than
double, from 24×109 litres in 2012 to 54×109 litres in 2030 (Verstegen, van der Hilst,
2016). The reference scenario assumes static ethanol demand at the level of 2013
(27×109 litres) up to 2030, but in this scenario the area of sugarcane still increases
due to increased demand for sugar (Verstegen, van der Hilst, 2016). Both scenarios
also assume increased production of crops, livestock and wood products by 2030
(van der Hilst, Verstegen, 2018). Trends of agricultural and silvicultural production
were based on the Shared Socio‐economic reference Pathway SSP2 (a ‘Middle of the
Road’ scenario), which projects socio-economic developments up to 2100 in line with
historical trends (O’Neill, 2017).
The land-use projections were based on a combination of the global computable

general equilibrium (CGE) model MAGNET (Modular Applied GeNeral Equilibrium
Tool) with the PLUC (PC Raster Land Use Change) model, which were run for the
period 2012 to 2030. MAGNET was used to model the development in demand and
supply of different commodities, including ethanol, on a global scale (Kavallari, 2014).
9 The species group indicators are not mutually exclusive, for example, a species can be both en-
demic and range-restricted.
91

Chapter 4
It provided projections of the amount of land in Brazil needed to meet increased

demand for ethanol up to 2030, as well as that of other agricultural and silvicultural
commodities (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). Projected
land requirements between 2012 and 2030 for crops, livestock production, forest
plantations and sugarcane were then spatially allocated using the PLUC model (Van
der Hilst, 2012; Verstegen, Karssenberg, 2016). PLUC was developed to project land-use
change based on temporal developments in demand for various land uses and spatially
explicit suitability of land for various land uses (Verstegen, Karssenberg, 2016). PLUC
produced annual maps of projected land use in Brazil between 2012 and 2030 at 25 km2
resolution for the following land-use types: grass and shrubs, natural forest, planted
forest, crops (excluding sugarcane), sugar cane, planted pasture, rangeland, abandoned
agricultural land, urban, water, and bare soil (Verstegen, van der Hilst, 2016).
The largest share of projected land-use change between 2012 and 2030 in both
scenarios consists of cropland expansion (excluding sugarcane, see Annex 23 for an
overview of land-use changes), which increases by about 320,000 km2 between 2012
and 2030 in both scenarios. This occurs predominantly in the Cerrado biome, at the
expense of natural forest, rangeland and grass/shrubland. Sugarcane expansion also
occurs mainly in this region, mostly replacing cropland (van der Hilst, Verstegen, 2018).
Sugarcane area increases by 1,400 km2 in the reference situation and 36,775 km2 in the
ethanol scenario. In addition, the ethanol scenario projects for 20,150 km2 of indirect
land-use changes scenario as a result of increased ethanol demand (Annex 23).
Species habitat suitability

In this study, biodiversity assessments were based on habitat suitability data for
terrestrial mammal species (Rondinini, 2011). Habitat suitability data is available
for 95% of all terrestrial mammal species at a global scale, covering 5,027 species
in total (Rondinini, 2011). The maps of species habitat suitability used in this study
had a resolution of 300 m, and distinguished three habitat suitability classes; highly
suitable, moderately suitable and not suitable (Rondinini, 2011). Only highly suitable
habitat, defined as primary habitat which can sustain the species, was included as
suitable habitat in this study, in line with Visconti et al. (Visconti, 2011). These habitat
suitability maps, developed by Rondinini et al. (Rondinini, 2011) and used as input data
in this study, provide habitat suitability per species based on information on habitat
suitability from the IUCN Red List (IUCN, 2019) on species’ range, species minimum and
maximum elevation, preferred habitat types (forest, shrub land, grassland, bare land
and artificial), required proximity to water bodies and tolerance to human disturbance
(Rondinini, 2011), as well as spatial data on land cover (based on Globcover version 2.3
(Bontemps, 2011)), water bodies and elevation.
92

Spatial analysis of land use and biodiversity

We assessed potential biodiversity impacts of increased ethanol demand using a
spatial neighbourhood analysis described in Chapter 3, which combines land-use data
with potential species richness data to create projections of biodiversity impact under
different scenarios of land-use change. Potential species richness was defined as the
sum of all species for which highly suitable potential habitat occurs in a cell. A map of
total potential mammal species richness was created by adding up habitat suitability
maps of all mammal species in Brazil. In a similar way, maps of threatened, endemic
and restricted-range mammal species richness were created by adding up suitability
maps of these subsets of species. This method was developed specifically to quantify
the impact of land-use change on biodiversity and was applied previously to a different
case study (Chapter 3).
Classification of habitat suitability was based on several factors, including landcover

4
as defined by Globcover v2.3 (Bontemps, 2011). To do this, we used the link between
habitat suitability and land use to make projections about future potential species
richness based on projections of land-use change. To this end, we determined location-
and land-use specific average values of potential species richness, and applied these
to the projected land-use map. Because our aim was to create projections of potential
species richness based on the PLUC land-use scenarios, the Globcover land cover types
were aligned to PLUC land-use types (see Annex 24 for a detailed description of the
reclassification). This resulted in the following land-use types being included in the
analysis; ‘urban’, ‘water’, ‘abandoned land & bare soil’, ‘forest’, ‘shrubs’, ‘grass’, ‘crops’
and ‘sugarcane’. The land-use category water contains open water (unsuitable for most
terrestrial species) and shorelines. Spatially variable and land-use specific potential
species richness values were then determined by creating a Boolean map of each land-
use type, and multiplying these maps with the potential species richness maps, thereby
creating a land-use specific potential species richness map. This provides a map with
species richness for each land use, only for the locations where this land use occurs.
93

94
Potential LU-/location Species
Neighbourhood
Habitat
species richness
operation
suitability
Chapter 4
richness sp. richness index

Sum Multiply specific pot.
Query
richness
operation Projected sp.
index 2012
300m 5km
Boolean ∆ sp. richness

300m
Boolean LU projection
300m
land use 2012 land use

Boolean
2001
2030
operation
Land use Projected sp.

5km Subtract index 2010 –
Query reference
index 2030 Baseline biodiversity

Boolean
5km operation richness
reference
5km
impacts
operation
300m
Boolean
300m
2030
∆ sp. richness
LU projection Boolean
reference land use

operation 5km
Projected sp. reference-

Subtract index 2030
richness ethanol
index 2030
5km Query
ethanol Ethanol biodiversity

5km
operation
Boolean impacts
2030 ethanol land use
Boolean
LU projection operation
5km
5km
Figure 4.1: Overview of the approach applied in this study. Rectangles represent maps, stacked rectangles represent sets of maps. Text inside the rectangles
5km
describes the type of map, text above and below the arrow describes the analysis step. Text below the rectangles provides the resolution of the map. Pot. =
potential, sp. = species.
16-03-20 12:27
A spatial neighbourhood analysis was carried out to calculate the potential species
richness for each cell in the study area at a 25 km 2 resolution. This was done by
calculating the average species richness in the cell’s neighbourhood for a particular
land-use type. We used four different window sizes in the calculation; 15x15 km,
85x85 km, 395x395 km and finally the whole of Brazil. These window sizes were
applied sequentially; if there was a value for the specific land use type within the
smallest window, this window was applied, if not, a larger-sized window was selected
(see Annex 25 for a detailed description of window sizes). For the majority of cells
(75%), the 15x15 km window was applied because a cell with the same land-use type
was found within this window size. Potential species richness maps for 2012 and
2030 (both for the ethanol and the reference scenario) were created for all species
groups. Potential species richness in 2012 and 2030 for each cell was expressed as
land-use specific window-averaged number of mammal species with suitable habitat
in the direct neighbourhood of each cell, from here on referred to as ‘species richness
4
index’ or SRI. The projected SRI map for 2012 was compared to the 2030 map for the
reference scenario to assess changes SRI due to changes in demand for agricultural and
silvicultural commodities (besides ethanol) during this period. Finally, SRI in 2030 for
the reference scenario was compared to SRI in 2030 for the ethanol scenario to assess
the impact of ethanol-driven sugarcane expansion on biodiversity.
4.3 Results
Potential species richness in 2012

Total potential mammal SRI in 2012 ranged from 0 to 150, depending on the location.
In line with an earlier study based on the same potential species range data (Rondinini,
2011), some hotspots of SRI were identified; SRI was highest in the Amazon region,
particularly in the western part (Figure 4.2), and was slightly lower at the edges of
the Amazon rainforest, in the so-called ‘arc of deforestation’, where forested land
borders mainly pasture area. Total SRI was also relatively high in remnants of the
Atlantic Forest. SRI was relatively low in the Pampas biome and at the southern
edge of the Cerrado region. These areas are dominated by pasture and cropland.
Threatened mammal SRI showed a spatial pattern similar to that of total SRI, but
also had high values in the north-eastern Cerrado region, where more remnant natural
vegetation remains. Endemic SRI was highest in the Amazon and eastern Cerrado.
Range-restricted SRI was highest along the coastal areas of the Atlantic Forest region,
where most forest remnants of Atlantic Forest remain. See Annex 26 for SRI maps for
threatened, endemic and range-restricted species in 2012.
95

Chapter 4
AM
CA
CE
Total SRI 2012
PN
High: 150
Low: 0
AF
PM
Figure 4.2: Total mammal Species Richness Index (SRI) in Brazil in 2012 at 25 km2 resolution. Black
lines show biome boundaries, abbreviations show biome names; AM = Amazon, CA = Caatinga, CE
= Cerrado, PN = Pantanal, AF = Atlantic Forest, PM = Pampas.
We determined ranges and median values of SRI per land-use type. The median value
of total SRI of forest in Brazil was about 3 to 6 times higher compared to the median
values for the other land uses (Figure 4.3). Median total SRI was lowest in sugarcane,
urban land, bare land and cropland. Shrubland, grassland and water (including
shorelines) had slightly higher median values for total SRI. Forest showed the largest
variation in total SRI, this can be due to the fact that it includes different forest types
such as Amazonian rainforest and Atlantic Forest, but also plantation forest. Urban
area showed the smallest variation. As most land-use types showed a wide range in
SRI values, a transition from one land use to another could result either in positive or
negative changes, depending on the local SRI values of each land-use type.
96

60
50
40
30
20
10
4
0
forest cropland grass shrubs sugarcane water urban bare soil
Land use type
Figure 4.3: Species richness index (SRI) per land‐use type in Brazil for 2012. Boxplots show median
values (middle horizontal bar) and first (upper limit) and third quartile (lower limit) values, while
vertical lines indicate maximum and minimum values. Bare soil = abandoned land & bare soil.
Projected changes in species richness index between 2012 and 2030 for the
reference scenario
Between 2012 and 2030 in the reference scenario, estimated changes in total mammal
SRI in Brazil ranged from plus 115 to minus 114 (Figure 4.4). The largest area of
SRI changes occurs in the Cerrado biome (Figure 4.5), including both increases and
decreases in SRI. In the Cerrado, reductions in SRI are mainly caused by crop expansion
at the expense of grassland, shrubland, abandoned land & bare soil and forest (see
Annex 27 for an overview of SRI changes per land-use change), most changes resulting
in a reduction of SRI of over 50% (Figure 4.5). These areas however generally have
low SRI values and the absolute changes in SRI are often small (Annex 27). In some
areas in the Cerrado, crop expansion results in a small increase in SRI, for example
when cropland expands over sugarcane and grassland. In the Amazon region, SRI was
reduced mainly due to the projected conversion from forest to grassland (Annex 27).
Increases in SRI occur when sugarcane and abandoned land & bare soil are converted
into cropland. In the Pantanal and Atlantic Forest biomes, the area with increased
total SRI was slightly larger than the area with decreases. The magnitude of positive
and negative changes was relatively high in the Amazon biome, where 37% of changes
resulted in a loss of total SRI over 50-100%. A number of land-use changes, including
97

Chapter 4
crop expansion over abandoned land & bare soil and expansion of forest over cropland
and grassland, resulted in positive SRI shifts in the Pantanal and Atlantic Forest
biomes, for example, forest expansion over cropland resulted in an average increase
in total SRI of 8.6 in the Atlantic Forest biome (Annex 27). The intensity (in % change)
of changes in SRI were relatively large for range-restricted species compared to the
other indicators, with over 20.000 km2 in the Amazon and over 35.000 km2 in the
Cerrado projected to lose all range-restricted species. The intensity of changes for
endemic and threatened SRI was similar to the magnitude of change for total SRI but
was higher for threatened and endemic species in the Atlantic Forest biome. Compared
to total SRI, the intensity of change was higher for SRI of range-restricted species.
AM
CA
CE
∆ total SRI
2030 reference - 2012 PN
High: 115
Low: 0 AF
0
High: 0
Low: -114
PM
Figure 4.4: Change in total mammal species richness index (SRI) between 2012 and 2030 (reference
scenario). Black lines show biome boundaries, biome names: AM = Amazon, CA = Caatinga, CE =
Cerrado, PN = Pantanal, AF = Atlantic Forest, PM = Pampas. Grey areas indicate areas with no
change in SRI.
98

100.000
50.000
-50.000
Area of SRI change (km²)

-100.000
Tot
Tot
Tot
Tot
Tot
Tot
Rrs
Rrs
Rrs
Rrs
Rrs
Rrs
Thr
Thr
Thr
Thr
Thr
Thr
-150.000
End
End
End
End
End
End
Amazon Caatinga Cerrado Atlantic forest Pantanal Pampas

0-25% 25-50% 50-100% >100% 0--25% -25--50% -50--100% >-100%
2
Figure 4.5: Area (in km ) in which changes in the mammal species richness index (SRI) occur between 2012 and 2030 (reference scenario) per biome for
total SRI (Tot), threatened SRI (Thr), endemic SRI (End) and range-restricted SRI (Rrs). Decreases in SRI are in red, increases are in green, darker colours
represent larger percent change in SRI.
99
16-03-20 12:27
Chapter 4
Projected differences in species richness index in 2030 between the ethanol

and reference scenario
Differences in SRI between the ethanol scenario and the reference scenario were found
predominantly in the Cerrado, Atlantic Forest and Pantanal biomes, and range from
minus 111 to plus 103 (Figure 4.6). The ethanol scenario was projected to have a
total SRI 50% lower than the reference scenario in over 2,500 km2 in the Cerrado, and
over 3,500 km2 in the Atlantic Forest biome (Figure 4.7). In the Atlantic Forest and
Cerrado, lower total SRI in the ethanol scenario was caused by sugarcane replacing
forest, grassland and abandoned land & bare soil. Lower total SRI in the ethanol
scenario compared to the reference in the Pantanal occurred where cropland replaces
shrubland and forest. Positive shifts also occurred in the Atlantic Forest, this is due to
expansion of forest and grassland, as well as expansion of sugarcane over cropland.
Even though relative differences in total SRI between cropland and sugarcane can be
large, absolute SRI differences between crop and sugarcane were small in most regions
(Annex 27). A shift from cropland to sugarcane in the ethanol scenario (compared to
the reference) resulted in higher SRI in the ethanol scenario in most biomes. In the
Pantanal and Amazon biome however, it resulted in a strong decrease in SRI (Annex
27), because the average total SRI value of cropland in these biomes was about twice
as high as in the other biomes (Annex 27). Impacts in the Atlantic Forest and Cerrado
region were mostly caused by direct land-use change, while impacts in the Amazon,
Caatinga, Pantanal and Pampas were mainly the result of indirect land-use change (see
Annex 23 for an overview of SRI changes for direct and indirect land-use change). As a
result, changes in biodiversity due to sugarcane expansion should be seen in context
of changes in biodiversity occurring due to other drivers of land-use change. When
comparing the ethanol and the reference scenario in 2030, direct land-use change
(sugarcane expansion) was responsible for 38% of all land-use transitions, and was
responsible for 23% of decline in total SRI losses (Annex 27). Forest loss (indirect
land-use change) constituted 16% of all land-use transitions, but was responsible
for 27% of all total SRI losses. Between 2012 and 2030, gross sugarcane expansion
constitutes about 50,000 km2 in the ethanol scenario, compared to about 20,000 km2 in
the reference scenario (Annex 23). Summarizing, sugarcane expansion was projected
to result both in relatively small positive and negative shifts in SRI, depending on the
location. Indirect land-use changes however consist of a range of land-use transitions
and has a larger impact. Especially the loss of forest and shrub land resulted in SRI
declines.
100

AM
CA
CE
∆ total SRI
2030 ethanol - reference PN
High: 103
Low: 0 AF 4
0
High: 0
Low: -111
PM
Figure 4.6: Difference in total species richness index (SRI) at 25 km2 resolution between the ethanol
scenario and the reference situation in 2030. Black lines show biome boundaries, abbreviations show
biome names; AM = Amazon, CA = Caatinga, CE = Cerrado, PN = Pantanal, AF = Atlantic Forest, PM
= Pampas.
101

102
15.000
Chapter 4
10.000
5.000
-5.000
-10.000
Area of SRI difference (km²)

-15.000
-20.000
Tot
Tot
Tot
Tot
Tot
Tot
Rrs
Rrs
Rrs
Rrs
Rrs
Rrs
Thr
Thr
Thr
Thr
Thr
Thr
-25.000
End
End
End
End
End
End
Amazon Caatinga Cerrado Atlantic forest Pantanal Pampas

0-25% 25-50% 50-100% >100% 0--25% -25--50% -50--100% >-100%
Figure 4.7: Area in which differences in species richness index (SRI) occur in 2030 between the ethanol scenario and the reference, per biome for total SRI
(Tot), threatened SRI (Thr), endemic SRI (End) and range-restricted SRI (Rrs). Negative changes are in red, positive changes are in green, darker colours
represent larger percent change in SRI.
16-03-20 12:27
4.4 Discussion
The method applied in this study allows for a spatial and quantitative assessment
of biodiversity impacts of land-use change for large groups of species over a large
area, and the comparison of different scenarios provides a way to identify impacts
due to specific drivers of land-use change. The application of this method to assess
biodiversity impacts of increased ethanol demand in Brazil has provided a first
indication of potential biodiversity risks and opportunities of ethanol-driven
sugarcane expansion. We found that in the absence of increased ethanol demand, Brazil
was projected to experience a range of SRI changes due to land-use change caused by
increased demand for other agricultural and silvicultural commodities. Biodiversity
impacts, both in terms of area and intensity of change, were highest in the Cerrado
biome, where cropland is expected to expand, and in the Amazon area, where grassland
is expected to replace forest. In the ethanol scenario, expansion of sugarcane was
4
predominantly projected to result in loss of SRI in the Cerrado and Atlantic Forest
biomes, while crop expansion was expected to result in both positive and negative
SRI impacts in the Pantanal region. SRI impacts due to sugarcane expansion were
generally small, as it mostly replaces cropland and pasture. Impacts of indirect land-
use change showed a large range of potential impacts, highlighting the need to include
indirect land-use changes in the assessment of impacts of increased ethanol demand
on biodiversity. Forest loss led to relatively large losses SRI, impacts were particularly
high when forest was converted to cropland or sugarcane. However, compared to
biodiversity impacts due to other land-use changes, biodiversity impacts of sugarcane
expansion are relatively small. This implies that other drivers of land-use change need
to be addressed as well in order to reduce negative impacts on biodiversity in the
future.
Uncertainties in input maps of habitat suitability and land use of Globcover and PLUC
influence the results of our analysis. Therefore, our projections of potential species
richness should be considered as indicators of trends and areas of interest, rather than
robust projections of future biodiversity. Habitat suitability maps contain uncertainty
due to variable extent of knowledge of habitat suitability per region, habitat type or
species (Rondinini, 2011). Furthermore, this study did not take into account landscape
composition or connectivity, while landscape configuration has been shown to
influence local species richness (Dale, 2000; Dauber, 2003). When compared to point
locality data for a subset of species, species habitat maps were able to accurately
predict about 77% of species occurrences (Rondinini, 2011). Validation of the PLUC
land-use projections for Brazil has shown that at aggregated spatial levels (250 x 250
103

Chapter 4
km2 and for the whole of Brazil), PLUC has relatively low uncertainty (coefficient of
variation, or cv, of 0.91) at allocating direct land-use change, but the precise location
of indirect land-use change remains highly uncertain (cv of 1.61) (Verstegen, van der
Hilst, 2016). Projections of land-use transitions towards sugarcane, cropland and
pasture were shown to be relatively accurate in validation of the PLUC model, while
transitions towards rangeland and planted forest were more uncertain (Verstegen,
Our approach also introduced new uncertainty in the analysis. The habitat suitability
maps used in this study are, amongst other maps, based on land cover as classified by
Globcover 2.3 (Bontemps, 2011). This map is based on remote sensing data (Bicheron,
2008), and its accuracy has been shown to be 67.5%, based on expert validation
(Bontemps, 2011). Some classes, such as bare soil, cropland, closed broadleaved
evergreen forest, and water bodies, were classified more accurately, while urban
areas, sparse vegetation and herbaceous vegetation were found to be more prone to
misclassification (Bontemps, 2011). Aligning land-use classification used in Globcover
with the land-use classification in PLUC may have added to uncertainty. Because the
Globcover map contains mosaic land-use types, some assumptions had to be made
about the proportion of each PLUC land-use type within these categories.
As shown by our results, SRI values per land-use type show a wide range, and therefore
partly overlap. This is due to spatial variation in potential species richness and
the large size of our study area. SRI in cropland differs from one area to the other,
for example due to differences in elevation or proximity to water, but also due to
heterogeneity within the land-use type due to different crop or management types.
For this reason, a specific land use transition could result in an increase or a decrease
in SRI, depending on the location. For example, the replacement of cropland by sugar
cane has resulted in both positive and negative changes in SRI. This highlights the
importance of a spatial assessment. Future research using this method would be aided
by biodiversity data based on more detailed land-use categorisation, and more detailed
projections of land-use change. We were unable to separate natural grassland from
pasture because the Globcover land-use map did not make a distinction between the
two types. The availability of maps of managed pasture could reduce the uncertainty
of the analysis.
Species richness as an indicator for biodiversity status is simple to calculate and easy
to interpret (Lamb, 2009), and has been shown to correlate to ecosystem function
and resilience (Folke, 1996; Lefcheck, 2015; Potter, 2014; Soliveres, 2016; Zavaleta,
104

2010b). However, species richness has also received criticism regarding its ability to
reflect the concept of biodiversity and provide information on conservation priority
of areas (Noss, 1983). An assessment of the number of species does not take into
account species composition (beta-diversity) and therefore provides no information
on functional diversity, endemism, or rarity (Fleishman, 2006). We find that the mean
mammal species richness index varies little between most non-forest land-use types.
We expect however that the species composition varies considerably across these
different land-use types. A change from shrubland to cropland may therefore result
only in a small reduction of species richness index, while potentially all original shrub-
dwelling species are replaced by species tolerant to agricultural lands. This failure to
report information on species composition is significant, because community structure
and diversity are related to the delivery of ecosystem services (Chapin, 2000; Noss,
1983). In this study, we alleviate this shortcoming by including subsets of species
as an indicator, including threatened, endemic and range-restricted species, that
4
provide some information on the conservation priority of species assemblies. Our
study provides projected trends in potential species richness, which require long-term
monitoring for verification. We therefore recommend future research to provide field
measurements of biodiversity in a range of land-use types, including agricultural lands
such as sugarcane fields.
The focus on potential species richness allows us to carry out a spatially explicit
analysis based on high resolution habitat data (Rondinini, 2011) for a large number of
species. The use of potential habitat data to assess potential species richness avoids
problems of sampling effort and observation chances that may skew presence/absence
data between different species or regions. However, the presence of suitable habitat is
no guarantee of the actual presence of the species that may use this habitat (Redford,
2010). Activities such as hunting may affect the presence of species within suitable
habitat (Benítez-López, 2017). While not explicitly included in the analysis, this aspect
was indirectly included in the analysis due to the fact that ‘human disturbance’ was
included as a factor in the habitat suitability maps. The habitat data on which this
analysis was based is only available for mammals. It has been shown that different
taxonomic groups respond differently to (projected) land-use changes (Lawton, 1998).
Therefore, results based on species richness for mammals cannot be used as proxy
for overall biodiversity.
The results presented here therefore provide first indications of potential areas with
high biodiversity loss. However, further research is necessary to assess how other
taxonomic groups may be affected by projected land-use changes. As a next step,
105

Chapter 4
it will be interesting to apply the method to different mitigation scenarios, such as

policy measures that prohibit the transition of forest land, or scenarios that include
strategies for nature-inclusive agriculture and agroforestry. Forests were found to
have a higher species richness index than the other land-use types. Therefore, in
order to reduce the negative impacts of sugarcane expansion on biodiversity, loss of
forest should be avoided. Furthermore, to reduce impacts on endemic species, loss
of other natural land-use types containing high numbers of endemic species, such
as shrubland in the Cerrado biome, should also be prevented. Protection of natural
areas can reduce the encroachment of land-use transitions into natural areas, but
current protections levels may not be sufficient to avoid impacts; for example, only
2% of the Cerrado biome is legally protected. Biodiversity losses on a landscape scale
may furthermore be reduced by efforts to increase biodiversity values of agricultural
land-use types. Biodiversity values of cropland could for instance be increased by
implementing agroforestry practices (Dornburg, 2012), which could provide suitable
habitat for a number of species while also maintaining economically viable yields
(Steffan-Dewenter, 2007). More intensive agriculture and livestock production could
also reduce the extent of projected indirect land-use changes, but also reduce the
species richness values of agricultural land. These trade-offs are at the heart of the land
sparing versus land-sharing debate. More detailed data on species habitat suitability
and spatial distribution of land management would allow for the assessment of these
types of mitigation measures. Research has shown that a combination of mitigation
strategies (including land conservation policies, yield improvements and shifts to
second generation ethanol production) could reduce the loss of natural vegetation
up to 96% (van der Hilst, Verstegen, 2018). The quantification of the effect of such
mitigation policies and land management scenarios on the biodiversity impact of
sugarcane expansion will be the focus of further research.
4.5 Conclusions
This study provides the first quantification of spatial patterns of ethanol-driven

biodiversity impacts in Brazil. We apply a method developed specifically to assess
spatial trends in biodiversity impacts due to land-use change.
We find that:
• Increased demand for ethanol (from 24×109 litres in 2012 to 54×109 litres in
2030) resulted in large areas of negative changes in potential species richness
in the Cerrado (close to 20,000 km2), Atlantic Forest (over 15,000 km2) and
Pantanal (almost 7,000 km2) biomes.
106

• In the Cerrado and Atlantic Forest, about 14% of these changes resulted in
losses of total SRI of over 50%. These impacts were mainly due to the direct
effects of sugarcane expansion, while in the Pantanal this was due to indirect
land-use change.
• These impacts should be viewed in the context of land-use change resulting
from increased demand for other agricultural products, which, in the absence
of increased ethanol demand, resulted in substantially larger areas with
projected biodiversity losses and gains, particularly in the Cerrado (over
200,000 km2, of which over 120,000 km2 negative changes) and the Amazon
(close to 100,000 km2, of which over 70,000 km2 negative changes) biomes.
• Loss of forest and shrubland resulted in the largest losses of potential species
richness.
Loss of natural vegetation due to sugarcane expansion can be strongly reduced by

4
mitigation strategies such as reducing the loss of natural vegetation (e.g. through
forest protection), increasing biodiversity values of agricultural land (e.g. through
agroforestry) or reducing the extent of land-use change (e.g. through agricultural
intensification). Future research should focus on assessing potential mitigation options
to avoid biodiversity impacts of sugarcane expansion. The identification of locations of
potential biodiversity impacts provided in this study is instrumental for policy makers
and ethanol producers aiming to reduce or mitigate biodiversity impacts.
107

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
5
Hydrological
and hydrologyimpacts
impactsofof
ethanol-driven
sugarcane expansion in Brazil
increased biomass Environmental
demand
Anna Duden
A.S. Duden, P.A. Verweij, Y.V. Kraak, L.P.H van Beek, N. Wanders, D.J. Karssenberg, E.H.
Sutanudjaja, F. van der Hilst
This chapter has been submitted for publication .
Anna Duden

Chapter 5
Abstract
Ethanol production in Brazil is projected to double between 2012 and 2030 in order
to meet increased global demand, resulting in the expansion of sugarcane cultivation.
Sugarcane expansion drives both direct and indirect land-use changes, and subsequent
changes in hydrology may exacerbate problems of (local) water scarcity. This study
assesses the impacts of projected ethanol-driven sugarcane expansion on agricultural
and hydrological drought in Brazil. Drought due to sugarcane expansion is modelled
using a spatial terrestrial hydrological model (PCR-GLOBWB) with spatiotemporally
variable land-use change and climate change scenarios as input. We compare an
ethanol scenario with increased ethanol demand to a reference situation in which
ethanol demand does not increase.
The results show that, on average, 29% of the Centre West Cerrado region is projected
to experience agricultural drought between 2012 and 2030, and the drought deficit
in this region is projected to be 7% higher in the ethanol scenario compared to the
reference. The differences between the ethanol and the reference scenario are small
when averaged over macro-regions, but can be considerable at a local scale. Differences
in agricultural and hydrological drought between the ethanol and reference scenario
are most notable in the Centre West Cerrado and Southeast regions. Locally,
considerable changes may also occur in other regions, including the Northeast Coast
and Northern Amazon region.
Because the South East and Centre West Cerrado regions are responsible for a large
proportion of agricultural production, increased agricultural drought may result in
significant economic losses, while increased hydrological drought could exacerbate
existing problems of water supply to large metropolitan areas in these regions. The
identification of areas at risk of increased droughts can be important information for
policy makers to take precautionary measures to avoid negative hydrological impacts
of increased ethanol demand.
110

Hydrological impacts of ethanol-driven sugarcane expansion in Brazil
5.1 Introduction
Agriculture is responsible for over 70% of global water withdrawals (Molden, 2007).
Growing demand for food, feed, and fibre is expected to increase the existing stress
on freshwater resources (Gerbens-Leenes, 2009), exacerbated by a rapid increase in
the production of biofuels (Watkins, 2015). One of the key sustainability concerns
related to biofuel production is its potential impact on water resources (Berndes,
2002; Guarenghi, 2016). Brazil has strongly increased its biofuel production over the
last decade (IEA, 2018), consisting largely of sugarcane-based ethanol (IEA, 2018). To
satisfy both domestic and international demand, Brazilian sugarcane production has
increased from 4 Mha in 1990 to 10 Mha in 2015, making Brazil the largest sugarcane
producer in the world (FAO, 2019). Sugarcane expansion has taken place mainly at
the expense of pasture and cropland (Adami, 2012; Guarenghi, 2016; Pereira, 2013).
Sugarcane production is currently concentrated in south-eastern Brazil, which has
sufficient rainfall for rainfed sugarcane production (Scarpare, Hernandes, Ruiz-Corrêa,
Kolln, 2016). However, future expansion of sugarcane into dryer areas may result in 5
hydrological impacts.
Global demand for bioethanol is expected to grow from 95.6ˑ109 litres in 2012 up to
168ˑ109 litres in 2030, of which 53ˑ109 L is expected to be produced in Brazil (van der
Hilst, Verstegen, 2018). Brazilian sugarcane area is projected to increase from 10.1 Mha
in 2012 to 13.6 Mha in 2030 (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst,
2016). Sugarcane expansion is projected to occur predominantly in the Centre West
Cerrado (CWC) and Southeast (SE) regions of Brazil (Scarpare, Hernandes, Ruiz-Corrêa,
Kolln, 2016; Scarpare, Hernandes, Ruiz-Corrêa, Picoli, 2016; van der Hilst, Verstegen, 2018;
Verstegen, van der Hilst, 2016). Besides direct land-use change (sugarcane expansion
over other land uses), indirect land-use change (ILUC) is also expected to occur
(Adami, 2012; Lapola, 2010; van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016).
Although the precise location is highly uncertain, ILUC is projected to occur mainly
in the Center West Cerrado, North Eastern Cerrado and the Northern Amazon region
(van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016).
Land-use change can have a strong influence on hydrological processes (Bosmans,

2017), e.g. by contributing to flooding and drought. Drought has been defined as ‘a
deficit of water relative to normal conditions’ (Sheffield, 2012). Brazil has abundant
freshwater resources (Martinelli, 2008) and overall water resources volume should
be sufficient to accommodate current and future biofuel demand (Berndes, 2002).
However, water availability is unevenly distributed (Flach, 2016; Hernandes, 2013;
111

Chapter 5
Moreira, 2007), and several sub-basins in Brazil’s south-eastern and central regions
are facing water scarcity problems (Scarpare, Hernandes, Ruiz-Corrêa, Kolln, 2016). The
Cerrado biome, into which sugarcane is expected to expand, has a longer dry spells and
soils with a lower water storing capacity than the current sugarcane cultivation areas
(Cabral, 2012; Marin, 2011; Scarpare, Hernandes, Ruiz-Corrêa, Kolln, 2016; Scarpare,
Hernandes, Ruiz-Corrêa, Picoli, 2016). Therefore, concerns have been raised about the
potential impacts of sugarcane expansion in Brazil on droughts (de Cerqueira Leite,
2009).
Drought can be categorized into meteorological, agricultural and hydrological drought

(Pedro-Monzonís, 2015). Meteorological drought is defined as a continuous shortage of
rainfall (Pedro-Monzonís, 2015). Meteorological drought can cause agricultural drought
when soil moisture deficits lead to hydric stress in plants (Pedro-Monzonís, 2015).
Land-use change affects soil moisture, e.g. through changes in evapotranspiration
and the interception of precipitation. Agricultural drought can cause water stress for
natural systems and may reduce agricultural yield significantly (Hisdal, 2004). The
2014 drought resulted in economic losses of about US$ 5.25 billion in the Brazilian
agricultural sector (Nobre, 2016). Hydrological drought occurs when soil moisture
deficit and/or increased evapotranspiration results in reduced river discharge (Hisdal,
2004; Pedro-Monzonís, 2015), which may severely impact downstream communities
and natural areas (Hisdal, 2004). For example, the 2014 drought jeopardized the water
supply of 28 million people in the south-east region of Brazil (Melo, 2016).
Because the hydrological impacts of sugarcane expansion are expected to vary over
space and time, it is important to assess hydrological impacts of sugarcane expansion
in a spatiotemporally explicit manner. A number of spatially explicit studies have
assessed the hydrological impacts of sugarcane expansion, and other land-use
changes in Brazil. The majority of these studies focus on hydrological impacts within
a single river basin (Costa, 2003; Pereira, 2013; Scarpare, Hernandes, Ruiz-Corrêa, Picoli,
2016) or a particular region in Brazil (Georgescu, 2013; Hernandes, 2013; Loarie, 2011;
Oliveira, 2014). For example, an increase in agricultural area at the expense of natural
vegetation resulted in an increased discharge of the Tocantíns river basin (Costa, 2003).
On the other hand, expansion of sugarcane, predominantly over pasture, resulted in
a decrease runoff in the Rio Grande basin (Pereira, 2013). These findings indicate that
land-use change due to sugarcane expansion may have different impacts on droughts
in different areas. Furthermore, ethanol-driven land-use change is projected to extend
beyond boundaries of regions or river basins (van der Hilst, Verstegen, 2018), and
impacts on hydrological processes thus transcend the catchment level.
112

The aim of this study is to assess the impacts of increased ethanol demand in Brazil
on agricultural and hydrological drought in a spatiotemporally explicit manner.
We include all watersheds in Brazil to assess the hydrological effects of both direct
and indirect land-use change due to sugarcane expansion. Hydrological impacts are
assessed based on spatiotemporal land-use projections and climate change scenarios
using an adapted version of the terrestrial hydrological model PCR-GLOBWB (van Beek,
2011; Wada, 2014; Wada, van Beek, 2011; Wada, Van Beek, 2011).
5.2 Methods
General approach
We assessed the impact of increased demand for Brazilian ethanol on agricultural and
hydrological drought using the spatial terrestrial hydrological PCR-GLOBWB model
(van Beek, 2011; Wada, 2014; Wada, Van Beek, 2011). The study area constitutes the
hydrological catchment areas situated entirely or partly in Brazil (Annex 32). The model 5
calculates water flows and storage, including soil moisture and discharge, based on
input of meteorological and land-use data (Figure 5.1). Land-use types included in the
model are defined by a set of land-use specific hydrological parameters. Soil moisture
and discharge modelled in PCR-GLOBWB are used to identify occurrence of droughts.
We compared indicators of agricultural and hydrological drought for an ethanol scenario,
which assumes increasing ethanol demand, to a hypothetical reference scenario, in which
ethanol demand remains static. Both meteorological and land-use projections are based
on the Shared Socio-economic Pathway 2 (van Vuuren, 2014), a middle-of-the-road
pathway assuming no strong shifts from historical patterns (Fricko, 2017).
Soil moisture and discharge output maps were used to calculate indicators of
agricultural and hydrological drought respectively using the threshold method
(Wanders & Van Lanen, 2015; Yevjevich, 1967). Droughts between 2012 and 2030 were
defined in terms of relative values of soil moisture or discharge; those values below
a spatially variable threshold derived from a historical reference were defined as
drought. Drought events were quantified by calculating:
• Area in drought
• Drought duration
• Drought deficit
These indicators provide information on different aspects of drought: spatial
prevalence (area in drought), temporal prevalence (drought duration) and an
integrated measure of intensity and duration of drought (drought deficit).
113

Chapter 5
PCRGLOB
Meteorological LU specific Land-use Meteorological
Input
Static land
observations parameters projections projections
use 2012
1958-2005 (5 km)
(~10 km, 2012-2030 2012-2030
(~50 km, daily) monthly) (5 km, annual) (~50 km, daily)
PCRGLOB
Historical Projected
model
hydrological hydrological
flows & storage flows & storage
(~10 km, daily) (~10 km, daily)
PCRGLOB
Output
Historical soil Historical Projected soil Projected

moisture discharge moisture discharge
(~10 km, monthly) (~10 km, monthly) (~10 km, monthly) (~10 km, monthly)
calculation
Threshold
Agricultural Hydrological
drought threshold drought threshold
calculation
Prpjected
Drought
Projected
agricultural hydrological
drought drought
(~10 km, monthly) (~10 km, monthly)
indicators
Area in Area in
Drought
Agricultural Agricultural Hydrological Hydrological

agricultural drought hydrological drought drought
drought
drought duration drought duration deficit
deficit
(% of total (% of total
(% of year) (dimensionless) (% of year) (dimensionless)
area) area)
Figure 5.1: Overview of the PCR-GLOBWB modelling process, with input (blue), modelling (green),
output (red), threshold calculation (orange) and drought calculations based on output and threshold
values (purple). For spatial data, spatial and temporal resolution is provided between brackets. A
spatial resolution of 5 arcmin, which equals about 10 km, is given as ~10 km. For drought indicators,
the unit is provided between brackets.
PCR-GLOBWB model
PCR-GLOBWB is a spatial terrestrial hydrology model which calculates water storage
in the soil and groundwater, as well as vertical water flows between soil, groundwater
and atmosphere at 5’ resolution and daily timesteps, based on input data including
land-use and meteorological data. The model includes both natural water flows and
anthropogenic water consumption. We included the following land-use types in PCR-
GLOBWB: urban, forest, grass- and shrubland, pasture, rice paddy, irrigated crops,
rainfed crops, irrigated sugarcane and rainfed sugarcane. The PCR-GLOBWB model
was adapted to include irrigated and rainfed sugarcane as separate land-use types,
114

which required adaptation of the land-use parameterization (see below). The model
was used to simulate water flows for a historical period (1958-2005) to establish a
historical baseline of soil moisture and discharge. The model was then run for land-
use projections and climate change scenarios between 2012 and 2030, resulting
in spatiotemporally explicit PCR-GLOBWB output for soil moisture and discharge.
Although PCR-GLOBWB calculates water flows and storage (including soil moisture
and discharge) with daily timesteps, output was reported as monthly averages due
to data storage and processing limitations. The performance of PCR-GLOBWB in
modelling river discharge was validated by e.g. (Sutanudjaja, 2018; van Beek, 2011).
We repeated the validation because of adapted land-use parameterisation used in this
study, and found similar results (for results, see Annex 33).
Land-use change scenarios

Land-use projections for the period 2012 and 2030 were taken from previous studies
(van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). We used an ethanol
scenario, which assumes Brazilian ethanol production to increase by 26ˑ109 litres (96% 5
increase) to meet current and planned ethanol mandates worldwide, resulting in an
expansion of sugarcane area of 3.5ˑ106 ha (35% increase) between 2012 and 2030. The
ethanol scenario was compared to a hypothetical reference scenario, in which demand
for ethanol was kept static up to 2030 at the 2013 level of 27ˑ109 litres. Sugarcane yields
are assumed to increase according to extrapolation of historical trends ( Jonker, 2015;
van der Hilst, Verstegen, 2018). The scenarios also assumed increasing production of
crops (82%), livestock (24%), and wood products (20%) (van der Hilst, Verstegen, 2018).
Land-use projections included the land-use types urban, crops, sugarcane, planted
pasture, grass- & shrubland, rangeland, natural forest, planted forest, bare soil, water
and abandoned land, and consist of annual land-use maps at a 5 km resolution. Land
use outside Brazil was kept static throughout the modelling period, because no land-
use projections were available for this area. The allocation method used to create the
initial land-use map for Brazil (Verstegen, van der Hilst, 2016) (Annex 34) was applied
to classify land use for the study area outside of Brazil, based on the Globcover v2.2
land-use map (Bicheron, 2008).
Land-use parameterisation
Projection land-use types (5 km resolution) were reclassified into PCR-GLOBWB land-
use types (Annex 35) and resampled to 5’ resolution (~10 km). Since PCR-GLOBWB
includes sub-grid variability of land use, the resulting PCR-GLOBWB cells can
contain fractions of multiple land-use types. This resulted in PCR-GLOBWB grid-cells
containing fractions of multiple land-use types. PCR-GLOBWB requires hydrological
115

Chapter 5
parameters related to root and vegetation structure for each land-use type (Van
Beek, 2009) (Annex 36). These land-use specific parameters vary both spatially and
temporally, and consist of interception capacity, groundcover, root depth and root
fraction. These parameters were assigned to each land-use fraction in each cell. For all
non-agricultural land-use types, we applied PCR-GLOBWB default parameter values
(van Beek, 2011) (Annex 36). Additionally, we derived parameters for the land-use
types irrigated sugarcane, rainfed sugarcane, irrigated crops and rainfed crops using
the method applied to calculate PCR-GLOBWB default values (van Beek, 2011) (see
Annex 36).
Climate scenarios
Future incidents of drought are expected to be affected by climate change (Marx, 2018;
Wanders & Van Lanen, 2015; Wanders & Wada, 2015). We used climate projections from
Climate scenarios
the Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) (Warszawski, 2014)
Future incidents of drought are expected to be affected by climate change (Marx,
to define future climatic conditions, applying projections from five General Circulation
2018; Wanders & Van Lanen, 2015; Wanders & Wada, 2015). We used climate
Models (GCMs) (Taylor, 2012) (HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, GFDL-
projections from the Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP)
(Warszawski, 2014) to define future climatic conditions, applying projections from
ESM2M and NorESM1-M) for the underlying Representative Concentration Pathway
RCP6.0 forfive General Circulation Models (GCMs) (Taylor, 2012) (HadGEM2-ES, IPSL-CM5A-
atmospheric CO2-equivalent concentrations (Moss, 2010; van Vuuren, 2011).
LR, MIROC-ESM-CHEM, GFDL-ESM2M and NorESM1-M) for the underlying
The soil moisture and discharge
Representative values per
Concentration grid cellRCP6.0
Pathway were averaged over the 5COGCMs
for atmospheric 2-equivalent
to control concentrations climate2010;
for variation in (Moss, changevan Vuuren, 2011). The soil moisture and discharge
projections.
values per grid cell were averaged over the 5 GCMs to control for variation in climate
Calculationchange projections.
of drought indicators

We applied a threshold method (Yevjevich, 1967) to PCR-GLOBWB output of soil
Calculation of drought indicators
We
moisture (m3/m applied
3
) and a threshold
dischargemethod
(m3/s) to (Yevjevich, 1967) to PCR-GLOBWB
define agricultural output of soil
and hydrological
moisture (m 3/m3) and discharge (m3/s) to define agricultural and hydrological
drought. This threshold method defines drought according to a threshold based on a
drought. This threshold method defines drought according to a threshold based on
historical reference.
a historical The 10th percentile
reference. The 10ofth the historical
percentile of (1958-2005) monthly
the historical averagemonthly
(1958-2005)
model output for soil moisture and discharge was chosen as the threshold level, in
average model output for soil moisture and discharge was chosen as the threshold
level, in
line with earlier line with
drought earlier
studies drought
(Demuth, 1997;studies (Demuth,
Prudhomme, 1997;
2014; Prudhomme,
Tallaksen, 1997). 2014;
Tallaksen, 1997). The threshold was determined for each month of the year and for
The threshold was determined for each month of the year and for each cell. Thus, a
each cell. Thus, a location- and time-specific relative measure of drought was
location- and time-specific relative measure of drought was derived.
derived.

Location
Location specific specific and
historical historical and soil
projected projected
moisture soil
andmoisture
dischargeand discharge
timeseries timeseries
were
were standardized using equation 1, drought events were defined using equation 2.
standardized using indicators
Drought equation 1,area
drought events were
in drought, defined
drought using equation
duration 2. Drought
and drought deficit were
indicatorscalculated using equations 3-6 according to (Wanders & Van Lanen, 2015; Wanders,
area in drought, drought duration and drought deficit were calculated using
equationsWada, 2015).
3-6 according to (Wanders & Van Lanen, 2015; Wanders, Wada, 2015).

! ",$ %!($)
stZ(t,n) = (1)
stZ(t,n) = (!(",$)

(1)

1 +,- ./0(/, 1) < ./010(/, 1)
D(t,n) = (2)
0 +,- ./0(/, 1) ≥ ./010(/, 1)

116 6
5 ",$
AID(t) = 789 * 100 (3)
:

;<
=8>< 5 ",$
Dduri,n = *100 (4)
?@
Location
Location specific
specific historical
historical and
and projected
projected soil
soil moisture
moisture and
and discharge
discharge timeseries
timeseries
Location specific historical and projected
were standardized using equation 1, drought events were defined using equation 2. soil moisture and discharge timeseries
Drought
Drought indicators
indicators area
area in
in drought,
drought, drought
drought duration
duration and
and drought
drought deficit
deficit were
were
Drought
Drought indicators
indicators area
area in
in in drought,
drought, drought
drought duration
duration and
and drought
drought deficit
deficit were
were
Drought indicators area drought,
calculated using equations 3-6 according to (Wanders & Van Lanen, 2015; Wanders,
calculated using equations 3-6 according to (Wanders & Van Lanen, 2015; Wanders, drought duration and drought deficit were
Wada, 2015).
Wada, 2015).
Wada, 2015).
Wada, 2015).
Wada, 2015).

stZ(t,n) =
!! ",$
",$ %!($)
%!($)
(1)
stZ(t,n) = !!! (!(",$)
stZ(t,n) =
",$
",$
",$
! ",$
%!($)
%!($)
%!($)
%!($)

Hydrological

impacts of

ethanol-driven

sugarcane

expansion

(1)
in(1)
Brazil

stZ(t,n) =
stZ(t,n) = (!(",$) (!(",$)
(!(",$) (1)(1)
(!(",$)
(!(",$)
1 +,- ./0(/,
1 +,- ./0(/, 1) 1) < < ./010(/,
./010(/, 1) 1)
D(t,n) = 1 +,- ./0(/,
D(t,n) = 0 +,- ./0(/,
1 +,- ./0(/, 1)
1)1) <
≥ ./010(/,
< ./010(/,
./010(/, 1)
1) (2)
(2)
D(t,n) = 1 +,- ./0(/,
0 +,- ./0(/, 1)
1) ≥ < ./010(/,
./010(/, 1)1)
1) (2)
D(t,n)
D(t,n) =
D(t,n) = = 0 +,- ./0(/, 1) ≥ ./010(/, 1) (2) (2)
(2)
0 +,- ./0(/,
0 +,- ./0(/, 1) 1) ≥ ≥./010(/,
./010(/, 1) 1)

66 5 ",$
6
789 5 ",$
",$
AID(t) =
AID(t) =
AID(t) =
6
6 65 ",$
789 5
789 * 100
* 100
5 ",$ * 100
(3)
(3)
(3)
AID(t)
AID(t) = = 789
789 :: 5 ",$
789 * 100 (3) (3)
AID(t) = :
: * 100 (3)

;<
: :

=8><
;< 5 ",$
;< 5
5 ",$
",$

Ddur
Dduri,n
Ddur i,n =
= =8><
=
;<
;< ;< 5
=8><
=8>< ",$ *100
*100 (4)
(4) (4)
i,n
Dduri,ni,n = = ?@ 5 ",$
5 *100
",$
*100 (4)
*100 (4)
i,n =8>< ?@
Ddur i,n = ?@ (4)
=8><
?@
?@ ?@
010
010 /, /,/, 11 −1 − 0(/,
− 0(/, 1) +,- B
0(/, 1) +,- B
1) +,- B /, /, 1
/, 11 = = 1
= 11
Ddef(t,n) =
Ddef(t,n)
Ddef(t,n) = = 010 010 010/, 1 /, 1− −0(/, 0(/, 1) +,- B
1) +,- B /, 1 == 1 (5)
(5) (5)
Ddef(t,n) =
Ddef(t,n) =
Ddef(t,n) = 0 +,- B
0 +,- B
0 +,- B /,
/, 1/, 1=
/, 11 = 0 1
= 00

(5)
(5)
(5)
0 +,- B
0 +,- B /, 1 = 0 /, 1 = 0

Ddef ii(n) =
Ddef (n)
(n) = = "FGE
DE
DE BHI+(/, 1)
BHI+(/, 1) (6)
(6) (6)

DE
DE
Ddef iii(n) = DE DE BHI+(/, 1)
"FGE (6)
Ddef
Ddef i (n) =
i(n) = "FGE
"FGE
"FGE BHI+(/, 1)
"FGE BHI+(/, 1)
(6)
(6)

where Z(t,n) is hydrological parameter soil moisture (m
where Z(t,n) is hydrological parameter soil moisture (m333/m 33/m33) or discharge (m33/s) at
/m3333) or discharge (m
) or discharge (m /s) at
where Z(t,n) is hydrological parameter soil moisture (m /m 3
)the
or discharge 33/s) at
(m 3
/s)
time
time t t
t (in
(in months)
months) for cell n,
for cell n, 0(1)
0(1) is is the
the average
average value value /m3/m ) or discharge (m
over 3) or discharge (m
over the timeseries
timeseries
3/s) at
3/s) at
for
for
time (in months) for cell n, 0(1) is the average value over the timeseries for
time t
time t t (in
at time (in months)
(in months)
months) for for cell
for cell n,
cell n, 0(1)
n, 0(1) isis the
is the average
the average value
average value over
value overover the the timeseries
the timeseries
timeseries for for
for
101
hydrological
101 parameter Z for cell n, σZ(t,n) is the standard deviation of parameter
101
Z, 101
stZ(t,n) is the standardised hydrological parameter Z parameter and Z10(t,n) is the

10th percentile threshold. D(t,n) a Boolean variable indicating whether a grid cell (n)
is experiencing a drought event (1) or not (0) at time t, which is a particular month in 5
a particular year. AID is Area In Drought (% of the total region), N is the total number
of cells in the region. Dduri,n is the drought duration (% of the year) of event i at cell n,
Si the first time step (specific month of a specific year) of a drought event i and Li the
last time step of the event. Ddef(t,n) is the standardised drought deficit of at time t of
cell n, expressed as the distance (in standard deviations) from the average discharge
or soil moisture at time t for cell n. Ddef i(n) is the sum of monthly drought deficits at
cell n during the drought event.
We mapped the PCR-GLOBWB output for historical soil moisture and discharge at 5’
resolution and averaged the monthly values for the period 1958-2005. Monthly values
of drought area, duration and deficit for the ethanol scenario and the reference were
averaged for the period 2012-2030 and per region to provide a measure of regional
agricultural and hydrological drought. Regions were defined as clusters of states, as
in earlier research (Verstegen, van der Hilst, 2016). We also calculate the difference
in drought indicators between the scenarios by subtracting values of the reference
scenario from those of the ethanol scenario, we provided maps of the difference in
drought deficit between the scenarios at 5’ resolution.
117

Chapter 5
5.3 Results
Historical hydrological properties

The assessment of historical hydrological conditions in the study area for the reference
period from 1958 to 2005 shows that annual average soil moisture is relatively low
in the Northeast Cerrado (NER) and Northeast Coast (NEC) regions (Figure 5.2),
consistent with areas of relatively low rainfall. Average annual soil moisture is also
relatively low Outside of Brazil (OB region), west of the Centre West Cerrado (CWC)
region in Bolivia and Paraguay. The dryer patches in the CWC, Southeast (SE) and
South (S) overlap with the presence of irrigated crops (including sugarcane). The
SE, CWC and OB regions show a strong seasonality in soil moisture, with a dry spell
from September to November and a wet season from February to May (Annex 37).
Seasonality is less pronounced in the Northern Amazon (NA) and almost absent in the S
region. Of the main sugarcane growing areas, the CWC shows a stronger seasonality in
soil moisture than the SE region. Discharge is highest for the Amazon river, particularly
in the downstream areas in the NA region (Figure 5.2).
Impact of increasing ethanol demand on droughts

Agricultural drought in terms of area, duration and deficit was projected to be highest
in the CWC region in both the ethanol scenario and the reference situation (Figure
5.3). In both scenarios, on average, 28% of the region experience agricultural drought
(averaged over the years 2012 to 2030 and over the grid cells in the region), which
lasts for about 3 months (25% of the year, Annex 38). Hydrological drought is most
widespread in the NA region in both scenarios, where on average, 26% of the region
experiences drought, while hydrological droughts last longest and are most intense in
the NEC region. The NEC, which is a relatively dry area to begin with (Figure 5.2), has
on average the smallest relative increases in agricultural drought area and duration.
When comparing the ethanol and the reference scenario, the difference in area in
agricultural and hydrological drought between the scenarios increases over time, and
becomes more seasonal, particularly in the CWC and the SE regions (Annex 37). The
CWC region experiences the largest impact of sugarcane expansion on agricultural
drought, the regional average of drought deficit was almost 7% higher in this region in
the ethanol scenario (Annex 38). The ethanol scenario results in lower regional average
drought deficits compared to the reference in the NEC region (Annex 38). Sugarcane
expansion had a relatively small impact on regional averages of hydrological drought;
differences between the ethanol and reference scenario were less than 1% (Annex 38).
Locally, however, larger impacts of increased ethanol demand can be seen.
118

Average soil moisture Average discharge

(m3/m3) 1958-2005 (m3/s) 1958-2005
High: 1 High: 200,000
Low: 0 Low: 0
isture Average discharge

2005 (m3/s) 1958-2005
High: 200,000
Low: 0
Figure 5.2: Annual soil moisture in m3/m3 (left) and discharge in m3/s (right), averaged over the
historical period (1958-2005). Abbreviations show region names; OB = outside of Brazil, NA = North-
ern Amazon, CWC = Centre West Cerrado, NER = North East Cerrado, NEC = North East Coast,
SE = South East, S = South.
Throughout the study area, there is substantial variation in the difference in

agricultural and hydrological drought deficit between the ethanol and reference
situation (Figure 5.4 and 5.5). Higher agricultural and hydrological drought deficits
(more negative Δ drought deficit in Figure 5.4 and 5.5) in the ethanol scenario occur
119

Chapter 5
in the SE and part of the CWC region that borders the SE region, and the western and
central area of the CWC. In the SE region and the part of the CWC region that borders
the SE region, higher drought deficits in the ethanol scenario are mainly due to the
projected expansion of rainfed and irrigated sugarcane at the expense of rainfed and
irrigated cropland and grass and shrubland. In the western and central area of the
CWC, increased drought deficit in the ethanol scenario is mainly due to projected
indirect land-use change where irrigated cropland replaces grass and shrubland. In the
ethanol scenario, the area of irrigated sugarcane in the CWC increased from 3,400 km2
in 2012 to over 11,000 km2 in 2030, compared to 6,700 km2 in the reference scenario
in 2030 (Annex 39). Lower drought deficits in the ethanol scenario compared to the
reference occur in parts of the NEC region (Figure 5.5). This is mainly due to indirect
land-use change projected for the ethanol scenario in this region, consisting of the
expansion of planted forest. Sugarcane is irrigated more, both in terms of irrigated
area and volume of water used, in the areas in which sugarcane expands Water use for
sugarcane irrigation projected for 2030 is almost 50% higher in the ethanol scenario
than in the reference situation.
120

Reference Ethanol
35
Area in agricultural drought Area in agricultural drought

30 Agricultural drought duration Agricultural drought duration
Area in hydrological drought Area in hydrological drought
Hydrological drought duration Hydrological drought duration
25
20
15
10
Agricultural & hydrological drought

area/duration (% of total region/year)
5
S SE CWC NER NEC NA OB

0
Figure 5.3: Average area (in blue and red) and duration (in green and orange) of drought for the reference situation (solid fill) and the ethanol scenario (shaded
Region
bars) per region in percentage of total region area and percentage of the year respectively. Monthly values of area and duration of drought were averaged
for the period 2012-2030 and per region. Abbreviations represent region names; S = South, SE = South East, CWC = Centre West Cerrado, NER = North East
Cerrado, NEC = North East Coast, NA = Northern Amazon, OB = outside of Brazil.
121
5
16-03-20 12:28
Chapter 5
∆ agricultural drought deficit

3,500
>0
0
<0
-2,500
Figure 5.4: The difference in standardized agricultural drought deficit (dimensionless) averaged
for the period 2012-2030 at 5’ resolution. A negative value means that there was a higher drought
deficit in the ethanol scenario, while a positive value means there was a lower drought deficit in
the ethanol scenario compared to the reference situation. Abbreviations represent region names;
OB = outside of Brazil, NA = Northern Amazon, CWC = Centre West Cerrado, NER = North East
Cerrado, NEC = North East Coast, SE = South East, S = South.
∆ hydrological drought deficit

500
>0
0
<0
-500
Figure 5.5: The difference in standardized hydrological drought deficit (dimensionless) averaged
for the period 2012-2030 at 5’ resolution. A negative value means that there was a higher drought
deficit in the ethanol scenario, while a positive value means there was a lower drought deficit in
the ethanol scenario compared to the reference situation. Abbreviations represent region names;
OB = outside of Brazil, NA = Northern Amazon, CWC = Centre West Cerrado, NER = North East
Cerrado, NEC = North East Coast, SE = South East, S = South.
122

5.4 Discussion
We assessed the impact of ethanol-driven sugarcane expansion in Brazil on

agricultural and hydrological drought, for all watersheds in Brazil. Our results show
that throughout the study area, agricultural and hydrological drought both increase
with increased demand for ethanol. Differences between the ethanol and reference
scenario are mainly localised, within the Centre West Cerrado and Southeast regions
showing the highest risk of increased agricultural and hydrological drought when
ethanol demand increases. Locally, considerable changes also occur in other regions,
including the Northeast Coast and Northern Amazon region. Sugarcane is irrigated
more, both in terms of irrigated area and volume of water used, in the areas in which
sugarcane expands. This is in line with earlier studies (Dallemand, 2015; Scarpare,
Hernandes, Ruiz-Corrêa, Picoli, 2016). Our results contradict earlier studies that project
that sugarcane expansion will not affect soil moisture content (Georgescu, 2013) nor
have severe impacts on water stress (Hernandes, 2013), which may be due to the
spatially aggregated level of these studies. This highlights the importance of a spatial 5
assessment at high resolution, as performed in this study.
The performance of the PCR-GLOBWB model has been validated previously by

comparing modelling results to observational data (Sutanudjaja, 2018; van Beek, 2011).
Validation showed that the PCR-GLOBWB model was able to accurately reproduce
discharge patterns including seasonal changes and inter-annual variation (Sutanudjaja,
2018), and PCR-GLOBWB successfully captures trends and seasonality for most large
river basins (Tangdamrongsub, 2016). However, PCR-GLOBWB does not perform
optimally in modelling total water storage of the Amazon basin due to uncertainties
in meteorological data and modelled groundwater response time (Sutanudjaja, 2018).
A number of assumptions and data limitations may have influenced the results. We
assumed static land use outside of Brazil, for which projections were not available.
Furthermore, we assumed a static demand for water for domestic and industrial
use for both historical and projected model runs. However, domestic and industrial
water use is expected to increase during the modelling period, since water withdrawal
for domestic use increased by 9% between 2006 and 2010, and water withdrawal
for industrial purposes increased by 26% in this period (Agencia Nacional de Agua,
2013). This could further exacerbate hydrological drought. Climate change will have
a significant influence on the occurrence and intensity of droughts change (Marx,
2018; Wanders & Van Lanen, 2015; Wanders & Wada, 2015). To reduce the uncertainty
of future climate change, we calculated our results based on five models of climate
123

Chapter 5
change, and averaged our results over these scenarios. Therefore, our drought values
may be considered conservative, vis-a-vis scenarios of stronger climate change.
We made use of a threshold method to identify drought events. This method does not
take into account the fact that some regions may be more adapted to droughts than
others (Wanders, Wada, 2015). In areas with variable soil moisture, native vegetation
may be better adapted to survive periods of droughts than in wetter areas. Farmers
may also adapt their agricultural practices to drought, through increased irrigation
(which is captured by the model) or by changing crops (which is not captured by
the model) (Maneta, 2009). Therefore, for two regions experiencing the same relative
drought increase, the severity of the impacts may differ.
The projected strong increase in production of Brazilian ethanol may have multiple
potential benefits; provision of renewable fuel, reduced greenhouse gas emissions,
enhanced energy security and contribution to rural development. However, sugarcane
expansion may also result in negative impacts such as increased drought, which
depend on the location of new sugarcane growing area, the previous land-use type
and indirect effects. Agricultural and hydrological drought can result in large social,
economic and ecological impacts, in particular in the Southeast and Centre West
Cerrado regions, where population density and agricultural production are high, and
where unique ecosystems provide a wide range of ecosystem services. The information
on areas at risk of drought due to increased ethanol demand can be used as input
for land-use planning, to minimize negative impacts of sugarcane expansion. Future
research is needed to quantify how projected drought may result in economic, social
and ecological impacts, and how these impacts may be avoided.
5.5 Conclusion
We assessed the impact of potential ethanol-driven sugarcane expansion in Brazil

on agricultural and hydrological drought. Expansion of sugarcane between 2012
and 2030 projected to occur predominantly in the Southeast and the Centre Cerrado
region, where it replaces grass- and shrubland and rainfed cropland. In the Centre
West Cerrado, where sugarcane expands, seasonality in soil moisture and discharge
is strong. Water use for sugarcane irrigation projected for 2030 is almost 50% higher
in the ethanol scenario than in the reference situation. Sugarcane expansion between
2012 and 2030 is expected to result in increased and more intense agricultural
droughts, particularly in the Centre West Cerrado region (7% increase in drought
124

deficit). The differences between the ethanol and the reference scenario are small when
averaged over entire regions, but are considerable at a local scale. This underlines the
importance of a spatially explicit approach for the assessment of drought impacts of
land-use change. Our results show that locally, ethanol-driven sugarcane expansion
is expected to lead to increased problems of water scarcity, particularly in terms of
agricultural drought. Land-use planning should take these risks into account.
125

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
6
Environmental
Synthesis and conclusions
Anna Duden
Anna Duden

Chapter 6
6.1 Research context
Increased bioenergy use is considered an important strategy in transitioning towards

more sustainable energy supply systems (IPCC, 2019) and can offer a number of benefits,
including diversification of (energy) markets, a reduction of the dependency on fossil
fuels and local socio-economic development through income and job generation
(Creutzig, 2015; Dale, 2010). Increased demand for bioenergy occurs alongside other
global trends resulting from population and economic growth, such as increased
demand for arable land and fresh water resources (IPCC, 2019). Projected increases in
bioenergy feedstock production can result in land-use change, which can be both direct
and indirect (Wicke, 2012). A number of risks of increased production of bioenergy
have been identified, related to potential environmental and socio-economic impacts
(Creutzig, 2015). A net reduction of greenhouse gas emissions (GHG) is the key priority
for bioenergy deployment, bioenergy systems may also have other environmental
impacts, for example when land-use change results in impacts on biodiversity and
hydrology (Creutzig, 2015; Díaz, 2019; Junginger, 2019). Freshwater availability is
already limited in some areas, and existing pressure on freshwater resources is
expected to increase due to the growing demand for food, feed, and fibre (Gerbens-
Leenes, 2009; Molden, 2007), as well as for biofuels (Watkins, 2015) and exacerbated by
climate change (IPCC, 2019). About a million species of plants and animals are currently
considered threatened with extinction (Díaz, 2019), and species extinction rates are
currently up to 1,000 times higher than natural background extinction rates, and are
projected to increase to 10,000 higher rates in the future (De Vos, 2015). Land-use
change is considered to have the largest impact on terrestrial ecosystems, through
habitat loss and degradation (Díaz, 2019).
Ex-ante modelling of potential land-use changes and subsequent environmental

impacts due to increased bioenergy demand provides information that allows for
the avoidance of negative impacts, and promotion of positive effects of increased
bioenergy demand. The effects of indirect land-use change can be included in the
quantification of the environmental impacts of increased bioenergy demand by
defining a study area at a geographical scale that extends beyond the bioenergy
sourcing or production area. Because these environmental impacts are strongly
related to location specific conditions, including spatially explicit projections of land-
use changes in the assessment of environmental impacts of bioenergy is essential
(Souza, 2015; van der Hilst, Hoefnagels, 2018). An integrated approach that includes
the strong interlinkages of bioenergy markets with other markets is required to
assess the impacts of bioenergy demand (Dale, Kline, 2017). In order to quantify the
128

environmental impacts of bioenergy, the impacts that can be attributed to bioenergy

should be separated from impacts of developments in other sectors. There are few
quantitative studies that are spatially explicit, separate bioenergy impacts from
impacts of developments in other sectors, and are carried out at a spatial scale that
allows inclusion of most indirect effects of market interactions and land-use change.
A number of studies assesses impacts of increased bioenergy demand in a spatially
aggregated way (Dornburg, 2012; Galik & Abt, 2015; Gerssen-Gondelach, 2017; Miner,
2014). Existing spatially explicit studies are predominantly performed at a local scale,
for example at state level or regional level (Costanza, 2017; Georgescu, 2013; Hernandes,
2013; Loarie, 2011; Oliveira, 2014; Tarr, 2017; van der Hilst, 2012), at a feedstock sourcing
area (Evans, 2013) or river basin area (Costa, 2003; Hernandes, 2018; Pereira, 2013;
Scarpare, Hernandes, Ruiz-Corrêa, Picoli, 2016), and may therefore exclude indirect
impacts of bioenergy demand. Other studies are carried out for large regions, but
with limited spatial resolution (Louette, 2010; Prestemon, 2002). Spatio-temporal and
quantitative approaches to assess biodiversity and hydrological impacts of land-use
change are desirable because environmental impacts of increased bioenergy demand
are location and time dependent. In this thesis, such approaches are developed and
demonstrated for two different case studies. 6
6.2 Aim and research questions
This thesis aims to quantify environmental impacts of increased bioenergy demand in

a spatially explicit manner. The following research questions are addressed:
I. How can environmental impacts of increased bioenergy demand be

quantified ex-ante?
II. What are the impacts of bioenergy driven land-use changes on biodiversity
and hydrology?
III. How do the impacts of increased bioenergy demand compare to impacts due
to the developments in demand for other agricultural and forestry products?
This thesis focuses on a regional to national level and projections for the near and
mid-term future (2030). The research questions are addressed in chapters 2 to 5. Table
6.1 provides an overview of the research questions, case studies and environmental
impacts addressed in each chapter.
129

Chapter 6
Table 6.1: Overview of the thesis chapters, and the topics, research questions, case studies and
environmental impacts they address.
Research
Case Environmental
Chapter Chapter title question
study impact
I II III
Modelling the impacts of wood
2 pellet demand on forest dynamics in x x US x
southeastern United States
Impact of increased wood pellet
3 demand on biodiversity in the x x x US Biodiversity
southeastern US
Biodiversity impacts of increased
4 x x Brazil Biodiversity
ethanol production in Brazil
Hydrological impacts of ethanol-
5 x x Brazil Hydrology
driven sugarcane expansion in Brazil
6.3 Summary of the results
Chapter 2 quantifies the effect of increasing wood pellet demand on land-use dynamics
in the southeastern US in a spatially explicit manner, while taking into account demand
for other wood products as well as development of other land uses. The modelling
approach presented in this chapter consisted of an economic model and a land use
allocation model, and was used to allocate projected changes in land use in a spatially
explicit way based on scenarios of future demand for wood pellets and other wood
products in southeastern US.
The results showed that an additional demand for wood pellets (an increase from
0.5 Mt in 2010 to 12.1 Mt in 2030) and other wood products can have a considerable
impact on timberland area in 2030. In the absence of additional demand for wood
pellets, natural timberland area was projected to decline by over 2,000 km2 (high
housing demand) to almost 15,000 km2 (low housing demand) by 2030, mainly
through urbanization and pine plantation establishment. However, increased wood
pellet demand provides an additional incentive for land-owners to retain their land
in timberland. Scenarios of increased wood pellet demand resulted in projections of
avoided loss of natural timberland of almost 2,000 km2 (high housing demand) to
7,500 km2 (low housing demand), which equals 0.4 and 1.5% of the current natural
timberland area respectively. Additionally, increased wood pellet demand led to
projections of expansion of pine plantation by approximately 8,000 km2 (high housing
demand) to 20,000 km2 (low housing demand) compared to stable pellet demand
130

(which amounts to an increase of 4.7 and11.2% in pine plantation area respectively).

Pine plantations were projected to be established often at the expense of natural
timberland; i.e. mainly natural pine and mixed forest. Projected establishment of new
pine plantations occurred mainly in the coastal parts of Virginia, North Carolina, South
Carolina, Georgia, Alabama and Mississippi. Shifts from natural timberland to pine
plantation occurred predominantly in the Atlantic coastal region, while regeneration
of natural forest occurred throughout the southeastern US (with concentrations in
the eastern parts of Georgia, Arkansas and Texas). The innovative modelling approach
presented in this chapter can also be applied to other landscapes or products.
Chapter 3 builds on the results of Chapter 2 and assesses the impact of increased
wood pellet demand on biodiversity in the southeastern US in a spatially explicit
manner, by assessing changes in projected potential species richness for all terrestrial
vertebrate species reported for the region. Increased demand for wood pellets was
projected to cause both positive and negative biodiversity impacts. These impacts were
location specific and different taxonomic groups showed different responses. Projected
differences in total species richness index (SRI) due to increased wood pellet demand
ranged from -11 to 5. Increased wood pellet demand resulted in a mean increase in 6
SRI of 0.4 (high housing demand scenario) to 0.7 (low housing demand scenario), and
a decrease of SRI of -1.0 (high housing demand scenario) to -1.4 (low housing demand
scenario). The negative changes in SRI due to increased wood pellet demand were more
pronounced than the positive impacts. We found that the potential species richness
was relatively high in the Coastal Plains, particularly for the land-use type lowland
hardwood forest. Negative shifts in potential species richness were projected for areas
in Florida, coastal Virginia and North Carolina, and parts of the Gulf Coast. Positive
shifts in potential species richness were projected for parts of Oklahoma and Arkansas.
The area of negative SRI changes was larger for reptiles than for birds, mammals and
amphibians.
Changes in potential species richness due to increased wood pellet demand

were considerably smaller compared to the changes due to other drivers, such as
urbanization and increased timber demand. Biodiversity impacts due to wood pellet
demand should therefore be considered in the context of other drivers of land-use
change and biodiversity loss. The spatially explicit approach presented in this study
provides a way to combine spatial datasets of land-use projections and biodiversity in
order to project potential impacts of increased bioenergy demand on species richness,
and the approach can be applied to different regions and drivers of land-use change,
131

Chapter 6
for example to show how projected demand for an internationally traded commodity
may lead to impacts on biodiversity in the procurement region.
Chapter 4 applied the approach presented in Chapter 3 to quantify the biodiversity

impacts of increased ethanol production from sugarcane in Brazil. The method
presented in Chapter 3 was used to spatially explicitly assess the impact of increased
demand for ethanol on potential total, threatened, endemic and range-restricted
mammal species richness in Brazil. Increased demand for ethanol in 2030 was
projected to result decreased potential species richness for about 19,000 km2 in the
Cerrado, 17,000 km2 in the Atlantic Forest and 7,000 km2 in the Pantanal. In the Cerrado
and the Atlantic Forest, biodiversity impacts of sugarcane expansion were mainly due
to direct land-use change, whereas in the Pantanal they were largely due to indirect
land-use change driven by expansion of crops other than sugarcane.
Biodiversity impacts due to sugarcane expansion were generally small, as sugarcane

was mostly projected to replace cropland and pasture. Impacts of indirect land-
use change showed a wide range of potential impacts (both positive and negative),
highlighting the need to include indirect land-use changes in the assessment of impacts
of increased ethanol demand on biodiversity. The biodiversity impact of increased
ethanol demand was projected to be smaller than the impact of other drivers of land-
use change. This implies that other drivers of land-use change need to be addressed as
well in order to reduce negative impacts on biodiversity in the future. Loss of forest and
shrubland resulted in the largest decreases in potential species richness. Therefore,
in order to reduce the negative impacts of sugarcane expansion on biodiversity, loss
of forest and natural shrubland, either through direct or indirect land-use change,
should be avoided. The identification of locations of potential biodiversity impacts of
increased ethanol demand in Brazil in this study can be instrumental for policy makers
and ethanol producers aiming to avoid or mitigate biodiversity impacts.
Chapter 5 provided an assessment of the hydrological impacts of ethanol-driven

sugarcane expansion in Brazil, by quantifying changes in agricultural and hydrological
drought. An adapted version of the terrestrial hydrological model PCR-GLOBWB was
used to assess the effects of both direct and indirect land-use change due to sugarcane
expansion. The study area consisted of all watersheds that are found completely or
partially within Brazil. Expansion of sugarcane between 2012 and 2030 was projected
to occur predominantly in the Southeast and the Centre-West Cerrado region, where
it replaced grass- and shrubland and rainfed cropland. In the Centre West Cerrado,
seasonality in soil moisture and discharge is strong. Sugarcane is irrigated more, both
132

in terms of irrigated area and volume of water used, in the areas in which sugarcane
expands. Projected sugarcane expansion between 2012 and 2030 resulted in increased
agricultural droughts, particularly in the Centre West Cerrado region, where on
average 29% of the region was projected to experience agricultural drought between
2012 and 2030, and the drought deficit in this region was projected to be 7% higher.
Sugarcane expansion had a relatively small impact on regional averages of hydrological

drought (<1%). The impacts of increased ethanol demand were small when averaged
over macro-regions, but were often considerable at a local scale, with areas within
the Centre West Cerrado and Southeast regions showing the highest risk of increased
agricultural and hydrological drought. This underlines the importance of a spatially
explicit approach for the assessment of drought impacts of land-use change. Our
results showed that locally, ethanol-driven sugarcane expansion is expected to
lead to increased problems of water scarcity, particularly in terms of agricultural
drought. Agricultural and hydrological drought can result in large social, economic
and ecological impacts, in particular in the Southeast and Centre West Cerrado regions,
where population density and agricultural production are high, and where unique
ecosystems provide a wide range of ecosystem services. The information on areas 6
at risk of drought due to increased ethanol demand can be used as input for land-use
planning, to minimize negative impacts of sugarcane expansion.
6.4 Main findings and conclusions
The environmental impacts of increased bioenergy demand were addressed for two
case studies; wood pellet production in the southeastern US and ethanol production
in Brazil. The results of these studies, presented in Chapters 2-5, provide information
to answer the research questions identified in this thesis.
RQ1: How can environmental impacts of increased bioenergy demand be

quantified ex-ante?
This thesis provides an ex-ante spatially explicit approach to quantify environmental
impacts of increased bioenergy demand, including indirect effects, both in terms of
economic interactions and in terms of land-use changes, and applies this approach
to two case studies. The innovative components of this approach consist of the
combination of the spatial allocation of bioenergy-driven land-use changes (based on
combined economic and land-use modelling) with a spatially explicit environmental
impact assessment. It enabled the quantification of hydrological and biodiversity
133

Chapter 6
impacts of increased ethanol demand in Brazil, and of biodiversity impacts of increased

wood pellet demand in the southeastern US. Spatial variability in impacts has been
determined, and high risk areas were identified. The approach is applicable to other
regions, and for other drivers of land-use or types of environmental impacts.
For the case study on increased ethanol demand in Brazil, spatially explicit land-
use change projections were already developed in previous studies (van der Hilst,
Verstegen, 2018; Verstegen, van der Hilst, 2016). Such projections did not yet exist for
increased wood pellet demand in the southeastern US. These were developed in this
thesis using an approach that combines an economic model for the wood markets of
the southeastern US with a land-use change model (Chapter 2). The economic model
(SRTS) (R. C. Abt, 2000; Galik, 2009b) allowed for the translation of increased wood
pellet demand into changes in forest area based on relative changes in forest rents,
whilst taking into account developments in demand for wood feedstocks from other
wood markets in the southeastern US. Developments in demand for other land-use
types, e.g. agriculture and urban area, were also projected. Changes in land use
requirements were then spatially allocated in the southeast US using the land-use
change model (PLUC) (Van der Hilst, 2012; Verstegen, 2012), based on land suitability
for different uses and competition between different land-use types. In this way, land-
use change projections were developed that include interaction of bioenergy demand
with developments in markets for other agricultural and silvicultural products, and
with other drivers of land-use change. These projections ran from 2010 to 2030 with
an annual time step. While the SRTS is specific for the southeastern US, the land-use
change model can be combined with other economic models and be applied to other
regions and markets.
The projected land-use changes following increased bioenergy demand were used
in this thesis as input for the analysis of the impact on hydrology and biodiversity.
The assessment of hydrological impacts of increased ethanol demand in Brazil made
use of the hydrological model PCR-GLOBWB (van Beek, 2011; Wada, 2014; Wada, Van
Beek, 2011), which was adapted to the Brazilian case study. PCR-GLOBWB is a global
spatial terrestrial hydrology model which calculates water storage and water flows
between soil, groundwater and atmosphere at 5’ resolution with daily timesteps. The
PCR-GLOBWB model uses land-use projections, climate scenarios and vegetation-
specific parameters as input. The model was adapted to include parameterization for
both irrigated and rainfed sugarcane as separate land-use types. This allowed for an
assessment of the impact of the ethanol induced land-use changes on both agricultural
and hydrological drought. The model was applied to all watersheds that are partially or
134

fully in Brazil, in order to account for all down-stream hydrological effects of land-use
change, and provided maps of soil moisture and discharge as output. The occurrence,
duration and intensity of drought between 2012-2030 was compared to a historical
reference period between 1958-2005.
For biodiversity, no spatially explicit model existed that allowed an ex-ante impact
assessment of increased wood pellet demand, including indirect effects. Therefore,
a new approach was developed to assess biodiversity impacts of bioenergy demand.
This approach combines species habitat maps with spatial land-use change projections,
using a neighbourhood analysis to make projections of changes in potential species
richness. The spatial neighbourhood analysis uses the link between habitat suitability
and land use to make projections about future potential species richness based on
projections of land-use change. Location- and land-use specific average values of
potential species richness were calculated and applied to the projected land-use map.
The approach allows for a spatial and quantitative assessment of biodiversity impacts
of land-use change for large groups of species over a large area, and the comparison of
different scenarios provides a way to identify impacts due to specific drivers of land-
use change. The approach is applied to two case studies, using land-use projections 6
up to 2030 with an annual time step. Biodiversity impacts were determined by
comparing the potential species richness in the start year of the projections (2010
in the southeastern US, 2012 in Brazil) to the potential species richness in 2030. The
approach can be adapted to other case studies, in other regions and for different land-
use types.
The results presented in this thesis are subject to a number of uncertainties and
limitations. For the southeastern US case study, the economic model (SRTS) is based
on the assumption that land or forest owners respond to market prices, while it has
been shown that other factors (e.g. property size, age and income of the land owner)
also play a role in forest management and ownership decisions in the southeastern US
(Aguilar, 2014; Butler, 2017; Joshi, 2009; Lubowski, 2008). Demand for non-forest land-
use types in the southeastern US case study was extrapolated from historical land-use
trends to project future developments in land use. Similarly, suitability factors for land
use allocation used in PLUC were based on historical land-use change patterns. The
models were calibrated based on historical trends, however, this may not properly
reflect potential changes in future drivers of land-use change (Verstegen, Karssenberg,
2016). Furthermore, it has been noted that the MAGNET model used to develop land-
use projections for Brazil has a limited capacity to deal with uncertainties in currency
volatility and policy uncertainties, e.g. with respect to forest protection (van der
135

Chapter 6
Hilst, Verstegen, 2018). The spatial allocation of projected land use also introduces
additional uncertainties, especially regarding indirect land-use changes (Verstegen,
van der Hilst, 2016). Additionally, there was no feedback mechanism between the
economic model and the land-use model, while a more robust integration of the two
models could improve the strength of results. The combination of the SRTS model (in
the southeastern US case) or the MAGNET model (in the Brazil case) with the PLUC
model and the PCR-GLOBWB model (for the assessment of hydrology) or the spatial
neighbourhood assessment (for the assessment of biodiversity) and their uncertainties
results in error propagation throughout the modelling chain. Nevertheless, the
demonstrated approach is still a significant step forward in understanding, ex-ante
potential impacts on biodiversity and water use in a spatially explicit way
Data limitations also influenced the results. The studies presented in Chapters 3 and
4 used the potential species richness index as the indicator of biodiversity. Despite
multiple benefits, the use of species richness alone as an indicator of biodiversity is
often considered to be insufficient because it does not provide information on species’
functional roles, contributions to ecosystem processes, and on species composition.
Other aspects of biodiversity are also vital for biodiversity conservation and ecosystem
functioning and the effect of land-use changes on species composition should be
assessed. This shortcoming was partly alleviated by including species richness of
different taxonomic groups (Chapter 3) and subsets of species, including threatened,
endemic, and range-restricted species (Chapter 3 and 4). Spatial information on
management intensity of land use would improve the impact assessments for
hydrology and biodiversity. Biodiversity impacts studied in this thesis did not include
a temporal component. For example, if a forest was established, it was immediately
assumed to have the same potential species richness index value as an existing forest
in that location. Data on changes in biodiversity over time in regenerating and newly
established forests would aid the temporally explicit analysis of biodiversity impacts
of projected land-use changes. The threshold method (i.e. droughts were defined in
terms of relative values of soil moisture or discharge; those values below a spatially
variable threshold derived from a historical reference were defined as drought) used
to quantify droughts does not take into account the effect of possible adaptations
(of vegetation as well as land owners) on the impact of drought. PCR-GLOBWB does
not perform optimally in modelling total water storage of the Amazon basin due
to uncertainties in meteorological data and modelled groundwater response time
(Sutanudjaja, 2018). For these reasons, the projections of potential environmental
impacts should be considered as indications of expected trends and areas of interest,
rather than absolute projections of future developments in biodiversity or hydrology.
136

Validation of the projected trends in potential species richness and hydrology would
require long-term monitoring. If such data and validation becomes available, the
demonstrated methodology will achieve higher reliability. Furthermore, depending
on data availability, the approach presented in this thesis could be used to asses other
environmental impacts. For example, if the carbon storage of different types of land
use and land management is known, the approach can be used determine changes in
carbon stocks over time or differences in carbon storage between different scenarios
of land-use change.
In conclusion, the approach presented in this thesis allows for a spatially explicit
assessment of environmental impacts of increased demand of bioenergy, and provides
a methodological framework to combine spatial land-use change projections with
the modelling of environmental impacts which can be applied at regional or national
scale. By combining an economic model with a land-use model, the approach offers the
capability to include indirect market and land-use effects in impacts of bioenergy. By
combining land-use projections with spatial environmental modelling, it is possible
to assess environmental impacts of both direct and indirect land-use changes ex-ante.
This approach can be applied to other regions, other drivers of land-use change, other 6
land use scenarios and different types of environmental impacts, with or without
policy measures to mitigate negative impacts. The successful application may however
be limited by the availability and quality of spatial data for environmental indicators.
RQ2 What are the impacts of bioenergy driven land-use changes on

biodiversity and hydrology?
Land-use changes due to increased wood pellet demand were projected to occur not
only in wood pellet sourcing areas, but throughout the entire southeastern US. Projected
establishment of new pine plantations occurs mainly in the coastal parts of Virginia,
North Carolina, South Carolina, Georgia, Alabama and Mississippi, while projected
forest regeneration occurs throughout the southeastern US, this regeneration is more
concentrated in the eastern parts of Georgia, Arkansas and Texas. This is in line with
findings of other studies (Drummond, 2010; Prestemon, 2002). Increased wood pellet
demand resulted in a mean increase in SRI of 0.4 (high housing demand scenario) to
0.7 (low housing demand scenario), and a decrease of SRI of -1.0 (high housing demand
scenario) to -1.4 (low housing demand scenario). The negative changes in SRI due to
increased wood pellet demand were more pronounced than the positive impacts.
When considering the results obtained from the case study on Brazil, sugarcane
expansion (direct land-use change) was projected to occur predominantly in the
137

Chapter 6
Cerrado, while indirect land-use changes were projected to occur throughout the
country. Sugarcane expansion was projected to result in small positive and negative
biodiversity impacts, depending on the location, because it replaces mainly cropland.
Indirect land-use changes had a larger biodiversity impact. Forest loss (due to indirect
land-use change) constituted 16% of all land-use transitions and was responsible for
27% of reductions in potential mammal richness in Brazil. Although indirect land-use
changes were included in the analysis of hydrological impacts of increased ethanol
demand in Brazil, the results presented in Chapter 5 do not separate between direct
and indirect impacts. This is because these impacts are the result of interplay of
various geographical scales, and therefore a combined result of direct and indirect
land use change.
For both the case study in the United States and the case study in Brazil, increased
demand for bioenergy was projected to cause both positive and negative biodiversity
impacts which varied widely over the landscape. These impacts were location specific
and differed per species group. For example, in the southeastern US, reptiles were found
to be relatively vulnerable to the projected changes in land use, compared to mammals,
birds and amphibians. Furthermore, in some areas, there were contrasting results for
different taxonomic groups; positive shifts were projected for one taxonomic group
while negative shifts were projected for the other. This highlights the importance of
analysing different taxonomic groups in biodiversity impact assessments. The Coastal
Plains in the southeastern US was rich in biodiversity, but it was also the region in
which biodiversity impacts were likely to occur, agreeing with earlier findings (Galik,
2016; Olesen, 2016). In Brazil, habitat availability of range-restricted mammal species
was found to be relatively strongly impacted compared to habitat availability of
threatened species, endemic mammal species. Only mammals were assessed for this
case study. However, it would be good to include other taxonomic groups, as the study
in the southeastern US shows that different taxonomic groups respond differently to
land-use changes.
For both case studies, some specific land-use transitions were found to result in an
increase and in a decrease in biodiversity, depending on the location. For example, the
replacement of cropland by sugarcane resulted in both positive and negative changes
in SRI in Brazil. The same was true for transitions from pasture to natural pine forest
in the southeastern US. This finding emphasizes the added value of a spatially explicit
assessment of biodiversity impacts. Some areas in Brazil and the US showed positive
biodiversity impacts due to increased bioenergy demand. In Brazil these were mainly
found in the Atlantic forest region, which is due to the projected expansion of forest and
138

grassland, as well as the expansion of sugarcane over cropland. In the southeastern US,
positive impacts were found mainly in the Coastal Plains, mainly caused by transitions
to natural forest and pasture. The total area with negative biodiversity impacts in
Brazil was larger (52,225 km2) than the area with positive impacts (28,300 km2). The
use of potential species richness index as calculated per raster cell does not allow for
a cumulative measure of biodiversity gain or loss for the entire study region.
Sugarcane was found to be irrigated more, both in terms of irrigated area and volume
of water used, in the areas in which sugarcane was projected to expand compared to
the traditional sugarcane areas, in line with other studies (Dallemand, 2015; Scarpare,
Hernandes, Ruiz-Corrêa, Picoli, 2016). Throughout the study area, both agricultural
and hydrological drought increased with increased demand for ethanol, The Centre
West Cerrado and Southeast regions showed the highest risk of increased agricultural
and hydrological drought related to increased ethanol demand. Sugarcane expansion
between 2012 and 2030 was expected to result in increased and more intense
agricultural droughts. For example, drought deficit was projected to increase by 7%
in the Centre West Cerrado region. Only small changes in hydrological drought were
found. The impact of increased ethanol demand was small when averaged over macro- 6
regions, but was often found to be considerable at a local scale. Earlier studies found
that sugarcane expansion will not affect soil moisture content (Georgescu, 2013) or
impact water stress (Hernandes, 2013), which may be due to the spatially aggregated
level of these studies. This underlines the importance of a spatially explicit approach
for the assessment of drought impacts of land-use change.
In conclusion, the results show that it is not straightforward to determine increased

bioenergy production is good or bad news for biodiversity – it depends on the
location and on the type of indicator studied (e.g. taxonomic group, threatened,
endemic, or range-restricted species). Trade-offs between environmental impacts
may exist (e.g. between hydrological and biodiversity impacts), and may also exist
within environmental impacts, depending on the indicator (e.g. different taxonomic
groups show different responses to projected land-use changes). Nevertheless, we
can conclude that increased demand for bioenergy does not only have negative
consequences for biodiversity, as in some areas it was projected to result in increased
potential species richness. The same is true for hydrological impacts of increased
ethanol demand, which in some locations resulted in reduced droughts while other
areas in increased droughts. The scenarios of increased bioenergy demand used in this
thesis are, however, modest, and more extreme scenarios of future bioenergy demand
may lead to a (non-linear) increase in projected impacts. The approach presented in
139

Chapter 6
this thesis could also be used to assess the environmental impacts of other land-use
and land management strategies. For example, it could be used to assess the potential
environmental impacts of scenarios of stronger improvements in agricultural and
livestock efficiencies, which could reduce the extent of land-use change and lower
negative impacts. Another example are strategies including agroforestry and mixed
cropping.
RQ3 How do the impacts of increased bioenergy demand compare to impacts

due to the developments in demand for other agricultural & forestry
products?
Bioenergy markets are not isolated systems, they interact with markets for other
agricultural and silvicultural products. For example, an increase in demand for
timber in the southeastern US will lead to an increase in the availability of sawdust
as a feedstock for wood pellets, while sugarcane grown in Brazil can be used to
produce either sugar or ethanol. Furthermore, the use of land for bioenergy feedstock
production competes with alternative uses, related to both nature conservation and
the production of food, feed and fibre. As a result, the impacts of increased bioenergy
demand should be considered in a context of changing demand for other products.
In the Brazilian case study, the scenarios that were assessed included, apart from
developments in ethanol demand, assumptions on increased production of crops,
livestock and wood products. These reference trends of agricultural and silvicultural
production were based on the Shared Socio‐economic reference Pathway SSP2 (a
‘Middle of the Road’ scenario), which projects socio-economic developments up to
2100 in line with historical trends. The biodiversity impact of increased ethanol
demand was projected to be considerably smaller than the impact of other drivers
of land-use change. The biodiversity impacts of the reference trends in agricultural
and silvicultural products resulted in impacts in all regions in Brazil, but the biggest
impacts (in terms of area affected) were found in the Cerrado region, followed by
the Amazon region. In absence of increased ethanol demand, the developments
due to these reference trends resulted in substantially larger areas with projected
biodiversity losses and gains, particularly in the Cerrado (over 200,000 km2, of which
over 120,000 km2 negative changes) and the Amazon (close to 100,000 km2, of which
over 70,000 km2 negative changes) biomes. In comparison, the additional biodiversity
impacts of sugarcane expansion were found to be relatively small; areas with projected
biodiversity losses and gains occurred predominantly in the Atlantic Forest (about
28,000 km2, of which over 17,000 km2 negative changes) and the Cerrado (over 20,000
km2, of which over 19,000 km2 negative changes) biomes.
140

In the case study in the United States, land-use change projections with and without
increased wood pellet demand were modelled. The wood pellet sector is strongly
influenced by developments in other larger wood-using sectors, such as the saw timber
and pulp and paper markets. A harvested forest stand typically yields timber, pulp
and paper, and bioenergy. This is a fundamental difference to the sugarcane system in
Brazil (which only has one type of primary output which is used to make either ethanol
or sugar). Therefore, the scenarios used in this thesis included projections for wood
pellet markets in conjunction with pulp and paper markets and the domestic housing
market in the US. Projections showed that in the absence of additional demand for
wood pellets, natural timberland area was projected to decline significantly, mainly
through urbanization and pine plantation establishment. Wood pellet demand had a
larger influence on pine plantation establishment in the absence of a strong housing
market; pine plantation area in 2030 increased by 19,359 km2 due to increased wood
pellet demand (low housing demand) compared to an increase of 8,120 km2 (high
housing demand). With increased wood pellet demand, more natural forest area was
projected to be retained and more pine plantations were projected to be established.
Compared to the low wood pellet demand scenario, 1,813 km2 (high housing demand)
to 7,436 km2 (low housing demand) less natural forest area was lost in 2030. Changes 6
in forest area due to increased wood pellet demand are relatively small compared
to changes due to increased timber demand; there is close to 7,000 km2 less natural
forest when demand for housing is low, and close to 2,000 km2 less natural forest
when demand for wood pellets is low (compared to the high housing and high wood
pellet demand scenario). However, even if the housing market is relatively strong,
pellet demand can still have an added influence on pine plantation establishment. In
both scenarios, natural forest is lost in the coastal parts of Virginia, North Carolina,
South Carolina, Georgia, Alabama and Mississippi, where the largest increases in pine
plantation area are expected. The Coastal Plains in the southeastern US are a hotspot
of potential biodiversity impacts of increased bioenergy demand. In particular the
lowland hardwood forests in this region are high in biodiversity, and are therefore
high-risk areas for biodiversity impacts. To avoid these impacts, land use and land
management changes in these areas should be avoided.
When compared to the changes due to other drivers, including urbanization and
increased demand for timber, the biodiversity impacts due to increased demand
for wood pellets were small. Increased wood pellet demand resulted in SRI changes
ranging from -11 to 5, while other drivers in the reference scenario resulted in changes
ranging from -21 to 37. No strong additional biodiversity impacts (∆SRI >1.5) of
increased wood pellet demand were found apart from a small number of areas where
141

Chapter 6
larger changes in potential species richness were projected, including the coastal area
of Virginia and the southern tip of Florida. However, the impacts of increased wood
pellet demand were larger when the demand from the housing market was assumed
to be low; low housing demand resulted in SRI changes ranging from -23 to 33, high
housing demand in SRI changes ranging from -11 to 5. This underlines that the impact
of wood pellet demand on biodiversity also depends on developments in other sectors.
As mentioned above, the projected developments of non-bioenergy markets were

subject to uncertainties. Furthermore, the scenarios used in this thesis are based
on Shared Socio-economic Pathway 2 (van Vuuren, 2014), which is a middle of the
road pathway of global socio-economic development. Scenarios based on more
extreme developments (with regard to both bioenergy and in general) may result in
stronger impacts. Similarly, land-use strategies focusing either on land sharing (e.g.
agroforestry) or land sparing (e.g. agricultural intensification) may also yield very
different impacts, but have not been modelled in this thesis.
In spite of these limitations, this thesis shows that compared to the assumed
development of bioenergy, projected developments in other sectors, such as expansion
of other crops or urbanisation, have a bigger impact on hydrology and biodiversity.
This implies that other drivers of land-use change will need to be addressed as well in
order to reduce negative impacts on hydrology and biodiversity in the future. In future
research, a holistic and integrated approach, such as demonstrated in this thesis, is
therefore required to further quantify these impacts. This approach would include
the interlinkages of bioenergy market with other markets (e.g. by using an economic
model), as well as the direct and indirect interplay between different land uses (e.g.
through a land use model), and quantify environmental impacts in a spatially explicit
way.
6.5 Recommendations for future research
This thesis has provided useful insights into the potential future environmental
impacts of increased bioenergy use. More specifically, a new approach has been
developed for assessing the environmental impacts of increased bioenergy demand in
a spatially explicit manner. This approach was applied to two case studies to quantify
hydrological and biodiversity impacts of increased bioenergy demand and subsequent
direct and indirect land-use changes. This allowed for the identification of areas at
risk of biodiversity loss or increased drought in the future.
142

A number of key research topics still remain uncovered, and should be addressed
in future research efforts. The reduction of GHG emissions due to the production
and use of bioenergy is the main motivation to increase bioenergy use. However,
uncertainty remains regarding the carbon impact of increased bioenergy use. The
spatially explicit impacts of increased bioenergy demand on carbon storage in the
landscape are largely unknown in the case studies in Brazil and the southeastern US,
especially related to indirect land-use changes. The approach presented in this thesis
could also be used to quantify carbon impacts. Future research may also address
bioenergy-driven hydrological impacts in the southeastern US. A second topic that
needs further attention are the potential synergies or trade-offs between different
types of environmental impacts. Future research could assess the trade-offs between
different impacts of land-use change, including impacts on hydrology and biodiversity,
as well as carbon storage and socio-economic impacts. This can identify pathways
with overall more positive or negative impacts. In other pathways, trade-offs between
different types of environmental impacts may indicate that there are no win-win
situations for increased bioenergy production. This would mean that prioritising the
reduction of one type of environmental impact over the other may be required, which
can be highly area specific. 6
Species richness alone is not sufficient to quantify biodiversity. Other aspects of
biodiversity, such as species composition, are also vital for biodiversity conservation
and ecosystem functioning. The effect of land-use changes on species composition
should also be quantified to provide insights on biodiversity impacts. Furthermore,
the temporal aspects of environmental impacts of bioenergy-driven land-use change
could be assessed in more detail. For example, the studies presented in Chapters 3
and 4 did not address the time-lag of species re-establishment in an area following
land-use change. Data on the development of biodiversity in regenerating forests or
shrublands would be helpful to be able to account for these time-lags. Similarly, data
on hydrological properties and species richness of land under different management
types would allow for a more detailed assessment of environmental impacts due to
changes in land management.
The approach presented in Chapters 2, 3, 4 and 5 could also be used to quantify

the effect of different mitigation strategies, for example by limiting the expansion
of bioenergy production in certain natural land use types, or by promoting e.g.
agroforestry practices. Land-use planning scenarios directed towards either land-
sparing or land-sharing strategies could be compared to assess their impact. The
approach presented in this thesis could be applied to include more extreme projections,
143

Chapter 6
and over longer time horizons (e.g. up to 2050) to assess the range of potential impacts
under these scenarios. In the current scenarios, only moderate increases in wood
pellet and ethanol demand were modelled. However, long-term scenarios described
in Energy Outlook reports (e.g. International Energy Agency 2019) require bioenergy
production to increase to higher levels (also in the case study areas); therefore, the
projected environmental impacts may be larger. An assessment of more extreme
scenarios combined with an assessment of different mitigation scenarios would allow
for a better assessment of the boundaries of sustainable bioenergy production.
Finally, the projected environmental impacts presented in this thesis are the result of
modelling efforts that are subject to various and sometimes propagating uncertainties.
Efforts could be made to quantify and reduce these uncertainties. Further ground
truthing of biodiversity values and hydrological properties of different land-use types
is also required to reduce uncertainty of results and to validate and monitor impacts
on the ground. Validation of the projected trends in potential species richness and
hydrology would require long-term monitoring.
6.6 Policy recommendations
There is a strong spatial variation in potential environmental impacts of increased

bioenergy demand. Impacts in and on natural forests and shrublands with high
conservation value should be avoided. In the southeastern US, lowland hardwood
forests in particular have high potential species richness, and loss of these forests is
expected to result in relatively large negative biodiversity impacts. In Brazil, loss of
forest also had a disproportionately large impact on biodiversity, compared to other
land-use transitions. Impacts on forests and shrublands of high conservation value
could be substantially reduced by increasing the protected status of these areas.
Agricultural intensification could also reduce encroachment into high conservation
areas. Protection of natural ecosystems can limit negative biodiversity impacts, but
due to the growing demand for agricultural products (including bioenergy), efficient
land-use planning strategies will also need to be defined. These strategies should
minimize environmental impacts while allowing for increased agricultural production.
The impact of land sparing and land sharing strategies can be modelled in order to
assess the effects of such strategies. In the southeastern US, increased wood pellet
demand can help to retain more land as forest and stimulate establishment of new
pine plantations, which could result in biodiversity benefits. Local best management
practices should include specific guidelines for species of interest, to maintain or create
suitable habitat in new plantations and forests.
144

Track and trace systems of bioenergy feedstock and products can be implemented to
assess whether feedstocks originate from high-risk areas. These areas can be defined
as high-risk areas because of high risk of conversion, because the area contains high
environmental value (e.g. high species richness), or both. High risk areas identified
in the two case studies were lowland hardwood forests in the Coastal Plains of the
southeastern US, as well as forests and shrublands in the Cerrado and Atlantic Forest
biomes in Brazil. However, track-and-trace systems can only be used to assess direct
impacts of bioenergy production. Top-down sustainability criteria can provide
guidelines for sustainable production or sourcing of bioenergy. However, increased
demand for bioenergy can have indirect effects which can be (partly) beyond the
control of bioenergy sectors, and are difficult to monitor. Indirect impacts of bioenergy
production and strategies to mitigate negative indirect impacts should therefore be
modelled. Finally, due to increased international trade, bioenergy can have impacts
on locations that are beyond control of governments of consuming countries. This is
complicated by the fact that bioenergy is often sourced from privately owned land,
where government regulations may have limited impact. A holistic and integrated
policy framework is therefore required to address these impacts and avoid leakage.
Such a framework would include policy measures at the regional land use level, and/ 6
or policy measures aimed at sectors of multiple products produced.
Whether or not increased bioenergy production results in environmental benefits or

losses depends on where and how biomass feedstock is grown. Policy frameworks
can be designed to target stimulation of positive impacts while avoiding negative
impacts, which requires, amongst others, quantitative projections of environmental
impacts in order to make informed decisions. A basis to do so is provided by the
modelling approach presented in this thesis. It is up to policy makers to answer
the questions on which impacts are (un)acceptable, and which types of impacts are
prioritised. Wood pellet production has expanded rapidly over the last decade. This
raises questions on when the limits of sustainable production are reached, or have
whether they have already been crossed in some regions. Answering this question
requires the formulation of quantitative sustainability thresholds that are specific for
different environmental impacts and different regions. These thresholds could then be
included in sustainability guidelines and government policies. The limits that should be
respected for sustainable biomass sourcing will be specific for different environmental
impacts, and will differ per region. The Coastal Plains of the Southeastern US and
the Cerrado and Atlantic Forest regions of Brazil have been identified as high-risk
areas, while other areas offer more opportunities for increased biomass production,
including potential positive impacts. Future expansion of biomass production in other
145

Chapter 6
areas therefore is recommended to be developed only when the future impacts are
understood and how these can be mitigated in case of negative and stimulated in case
of positive impacts.
Demand for bioenergy is increasing rapidly. The quantification of environmental

impacts of increased bioenergy production is therefore an urgent issue, which requires
high priority of scientists and policy makers. The approach presented in this thesis
allows for a spatial assessment of both direct and indirect impacts of bioenergy
production. The quantitative and spatial information provided in this thesis provides
input for an informed decision about environmental impacts of bioenergy.
146

147

Environmental
Environmental
Anna Duden
Anna Duden

Environmental
Annexes Environmental
Anna Duden
Anna Duden

Annexes
Annex 1: Feedstock demand scenarios for the southeastern US
Feedstock demand scenarios, consisting of demand per feedstock type per year,
were used as input for the SRTS model. Four types of feedstock are included; large
roundwood softwood, large roundwood hardwood, small roundwood softwood
and small roundwood hardwood. Four scenarios are included, which consist of
combinations of a low or a high housing demand and a low or high wood pellet demand.
150

Table A1: Projected demand for large and small roundwood in Mt for the different scenarios between 2010 and 2030. Sources: a Based on (Ince, 2012), b Based
on (Cocchi, 2011; Ince, 2012). Total pellet demand of 11.6 Mt is derived from the sum of the difference between the low pellet and the high pellet scenarios
for both softwood and hardwood.
Feedstock type Large roundwood Small roundwood
Scenario name Low housing High housing Low wood pellet High wood pellet
Demand (Mt) Softwooda Hardwooda Softwooda Hardwooda Softwoodb Hardwoodb Softwooda Hardwooda
2010
2011
110,740 37,665 110,740 37,665 63,756 37,324 63,756 37,324
2012
82,092 27,896 82,092 27,896 75,843 35,409 75,843 35,409
2013
82,913 28,053 88,801 28,053 76,654 35,463 76,654 35,463
2014
84,571 28,210 96,058 28,210 77,475 35,518 77,475 35,518
2015
87,109 28,368 103,908 28,368 78,304 35,573 78,449 35,609
2016
89,722 28,737 108,139 28,737 80,120 36,131 80,410 36,204
2017
92,413 29,111 112,542 29,111 81,978 36,698 82,559 36,843
2018
95,186 29,490 117,124 29,490 83,880 37,274 85,041 37,564
2019
97,090 29,874 121,893 29,874 85,825 37,859 88,147 38,439
2020
99,031 30,263 126,855 30,263 87,815 38,453 92,460 39,614
2021
100,022 30,655 130,552 30,655 89,010 38,969 98,299 41,291
2022
101,022 31,051 134,356 31,051 90,219 39,492 99,509 41,814
2023
101,527 31,453 138,271 31,453 91,446 40,022 100,736 42,344
2024
102,035 31,860 142,300 31,860 92,689 40,559 101,979 42,881
2025
102,545 32,272 146,447 32,272 93,949 41,103 103,239 43,426
2026
103,058 32,691 150,026 32,691 95,151 41,633 104,441 43,955
2027
103,573 33,115 153,692 33,115 96,368 42,169 105,658 44,492
2028
104,091 33,545 157,448 33,545 97,601 42,713 106,890 45,035
2029
104,611 33,981 161,296 33,981 98,849 43,263 108,138 45,585
2030
105,134 34,422 165,238 34,422 100,114 43,820 109,403 46,143
105,660 34,695 166,709 34,695 101,284 44,196 110,574 46,518
151
Annexes
&
16-03-20 12:28
Annexes
Annex 2: TPO adjustment of FIA data
Forest inventory data was collected for the southeastern US by the Forest Inventory
and Analysis (FIA) national program of the USDA Forest Service. FIA field data is
collected with a plot sampling cycle of 5-7 years and therefore does not reflect one
fixed point in time. Biannually collected mill consumption data from the Timber
Product Output (TPO) database of the FIA is therefore used in order to estimate actual
removals in 2010. TPO data is gathered by FIA biannually (in odd years) by sending
questionnaires to mills and estimating round wood use and residues produced. Values
of roundwood utilized in mills per survey unit are used to adjust FIA based removal
data in the demand scenarios, to better represent what happened in the base year.
This is done by calculating the average value of TPO data for the period overlapping
the FIA sampling cycle of the data used in version 28b of SRTS, which is 2005 to 2013.
This average value is then compared to TPO data specifically from 2011. The difference
between the average for the period of 2005 to 2013 and the specific values for 2011 is
used to correct the demand scenarios in 2011. The factors used to correct are shown
in table A2.
This means that for all product types except softwood small roundwood, removals in
2011 were slightly lower than the average of the 2005-2013 period. This is especially
true for both softwood and hardwood large roundwood, the demand for which is
influenced by the housing market and the economic recession. For softwood small
roundwood, removals were higher than average.
Table A2: Correction factors based on Timber Product Output data, applied to Forest Inventory and
Analysis data per product type.
Softwood Softwood Hardwood Hardwood
small large small large
roundwood roundwood roundwood roundwood
TPO correction
0.19 -0.26 -0.05 -0.26
factor
152

Annexes
Annex 3: Land-use projection scenarios for the southeastern US
The tables below show the demand for all land-use types between 2010 and 2030for the
four scenarios used as input in the PLUC model. Table A3a: HhHp scenario, Table A3b:
HhLp scenario, Table A3c: LhHp scenario, Table A3d: LhLp scenario. LU class = land-
use type Federal = federal land, Pl. pine = planted pine, N. pine = natural pine forest,
mixed f. = mixed forest, Ulhw = upland hardwood forest, Llhw = lowland hardwood
forest, Nfveg
&
153

Table A3a: Land-use projections for the HhHp scenario in km2
154
LU type Urban Water Federal Cropland Pasture Pl. pine N. pine Mixed f. Ulhw Llhw Nfveg
Annexes
2010
2011
138,939 71,490 89,735 172,569 198,434 173,256 94,260 60,804 228,136 104,368 43,427
2012
139,233 71,490 89,735 171,947 196,897 172,657 94,016 60,643 227,528 104,249 47,021
2013
139,527 71,490 89,735 171,326 195,360 171,912 93,721 60,450 226,794 104,085 51,017
2014
139,820 71,490 89,735 170,704 193,823 172,943 93,683 60,432 226,808 104,116 51,861
2015
140,114 71,490 89,735 170,083 192,286 174,619 93,672 60,436 226,964 104,183 51,835
2016
140,408 71,490 89,735 169,461 190,750 176,973 93,674 60,455 227,247 104,279 50,945
2017
140,702 71,490 89,735 168,840 189,213 176,981 93,550 60,374 226,949 104,232 53,351
2018
140,995 71,490 89,735 168,218 187,676 177,529 93,471 60,326 226,817 104,224 54,935
2019
141,289 71,490 89,735 167,597 186,139 178,535 93,421 60,300 226,803 104,246 55,862
2020
141,583 71,490 89,735 166,975 184,602 180,035 93,392 60,292 226,898 104,295 56,119
2021
141,877 71,490 89,735 166,354 183,066 182,528 93,382 60,304 227,174 104,385 55,121
2022
143,751 71,490 89,735 165,876 182,796 185,345 93,370 60,318 227,493 104,481 50,762
2023
145,625 71,490 89,735 165,398 182,527 186,272 93,304 60,281 227,435 104,489 48,860
2024
147,499 71,490 89,735 164,920 182,258 188,409 93,270 60,275 227,610 104,554 45,397
2025
149,373 71,490 89,735 164,442 181,988 190,946 93,236 60,273 227,842 104,630 41,460
2026
151,247 71,490 89,735 163,964 181,719 193,370 93,194 60,264 228,041 104,698 37,693
2027
153,122 71,490 89,735 163,486 181,449 195,351 93,142 60,245 228,153 104,745 34,499
2028
154,996 71,490 89,735 163,009 181,180 197,457 93,086 60,224 228,275 104,794 31,171
2029
156,870 71,490 89,735 162,531 180,911 200,459 93,027 60,211 228,524 104,867 26,792
2030
158,744 71,490 89,735 162,053 180,641 203,513 92,958 60,192 228,762 104,936 22,391
160,618 71,490 89,735 161,575 180,372 206,839 92,880 60,170 229,018 105,005 17,714
16-03-20 12:28
Table A3b: Land-use projections for the HhLp scenario in km2
2010
2011
138,939 71,490 89,735 172,569 198,434 173,256 94,260 60,804 228,136 104,368 43,427
2012
139,233 78,839 90,999 171,947 196,897 172,657 94,016 60,643 227,528 104,249 38,408
2013
139,527 78,839 90,999 171,326 195,360 171,912 93,721 60,450 226,794 104,085 42,403
2014
139,820 78,839 90,999 170,704 193,823 172,943 93,683 60,432 226,808 104,116 43,247
2015
140,114 78,839 90,999 170,083 192,286 174,619 93,672 60,436 226,964 104,183 43,221
2016
140,408 78,839 90,999 169,461 190,750 176,933 93,673 60,453 227,240 104,277 42,382
2017
140,702 78,839 90,999 168,840 189,213 176,870 93,543 60,368 226,919 104,225 44,898
2018
140,995 78,839 90,999 168,218 187,676 177,309 93,455 60,314 226,755 104,209 46,646
2019
141,289 78,839 90,999 167,597 186,139 178,059 93,389 60,277 226,676 104,214 47,938
2020
141,583 78,839 90,999 166,975 184,602 179,080 93,339 60,251 226,664 104,236 48,847
2021
141,877 78,839 90,999 166,354 183,066 180,453 93,306 60,238 226,731 104,278 49,276
2022
143,751 78,839 90,999 165,876 182,796 180,642 93,196 60,168 226,487 104,242 48,420
2023
145,625 78,839 90,999 165,398 182,527 181,255 93,116 60,121 226,362 104,234 46,941
2024
147,499 78,839 90,999 164,920 182,258 182,780 93,082 60,109 226,449 104,279 44,202
2025
149,373 78,839 90,999 164,442 181,988 184,675 93,053 60,103 226,595 104,338 41,010
2026
151,247 78,839 90,999 163,964 181,719 186,619 93,020 60,095 226,741 104,397 37,775
2027
153,122 78,839 90,999 163,486 181,449 188,309 92,979 60,080 226,832 104,441 34,879
2028
154,996 78,839 90,999 163,009 181,180 190,172 92,936 60,065 226,945 104,490 31,785
2029
156,870 78,839 90,999 162,531 180,911 192,858 92,897 60,061 227,187 104,567 27,696
2030
158,744 78,839 90,999 162,053 180,641 195,669 92,852 60,053 227,433 104,643 23,490
160,618 78,839 90,999 161,575 180,372 198,719 92,798 60,044 227,698 104,720 19,034
155
Annexes
&
16-03-20 12:28
Table A3c: Land-use projections for the LhHp scenario in km2
156
Annexes
2010
2011
138,939 71,490 89,735 172,569 198,434 173,256 94,260 60,804 228,136 104,368 43,427
2012
139,233 78,839 90,999 171,947 196,897 172,657 94,016 60,643 227,528 104,249 38,408
2013
139,527 78,839 90,999 171,326 195,360 171,912 93,721 60,450 226,794 104,085 42,403
2014
139,820 78,839 90,999 170,704 193,823 170,197 93,132 60,071 225,360 103,722 48,749
2015
140,114 78,839 90,999 170,083 192,286 168,828 92,635 59,749 224,148 103,423 54,312
2016
140,408 78,839 90,999 169,461 190,750 167,819 92,250 59,500 223,212 103,200 58,979
2017
140,702 78,839 90,999 168,840 189,213 167,212 91,994 59,334 222,579 103,061 62,643
2018
140,995 78,839 90,999 168,218 187,676 166,877 91,816 59,217 222,133 102,974 65,671
2019
141,289 78,839 90,999 167,597 186,139 166,990 91,707 59,148 221,884 102,937 67,888
2020
141,583 78,839 90,999 166,975 184,602 167,412 91,626 59,098 221,736 102,923 69,622
2021
141,877 78,839 90,999 166,354 183,066 169,267 91,631 59,113 221,950 103,004 69,317
2022
143,751 78,839 90,999 165,876 182,796 172,602 91,687 59,175 222,494 103,138 64,058
2023
145,625 78,839 90,999 165,398 182,527 172,469 91,548 59,084 222,151 103,077 63,698
2024
147,499 78,839 90,999 164,920 182,258 174,154 91,536 59,087 222,305 103,140 60,678
2025
149,373 78,839 90,999 164,442 181,988 176,491 91,537 59,104 222,579 103,231 56,831
2026
151,247 78,839 90,999 163,964 181,719 179,601 91,548 59,136 222,988 103,345 52,029
2027
153,122 78,839 90,999 163,486 181,449 183,211 91,568 59,178 223,498 103,463 46,602
2028
154,996 78,839 90,999 163,009 181,180 187,111 91,590 59,224 224,060 103,580 40,828
2029
156,870 78,839 90,999 162,531 180,911 190,928 91,584 59,253 224,549 103,687 35,267
2030
158,744 78,839 90,999 162,053 180,641 194,879 91,566 59,276 225,036 103,788 29,596
160,618 78,839 90,999 161,575 180,372 199,023 91,540 59,296 225,539 103,885 23,729
16-03-20 12:28
Table A3d: Land-use projections for the LhLp scenario in km2
2010
2011
138,939 71,490 89,735 172,569 198,434 173,256 94,260 60,804 228,136 104,368 43,427
2012
139,233 78,839 90,999 171,947 196,897 172,657 94,016 60,643 227,528 104,249 38,408
2013
139,527 78,839 90,999 171,326 195,360 171,912 93,721 60,450 226,794 104,085 42,403
2014
139,820 78,839 90,999 170,704 193,823 170,197 93,132 60,071 225,360 103,722 48,749
2015
140,114 78,839 90,999 170,083 192,286 168,828 92,635 59,749 224,148 103,423 54,312
2016
140,408 78,839 90,999 169,461 190,750 167,736 92,223 59,483 223,149 103,182 59,186
2017
140,702 78,839 90,999 168,840 189,213 167,055 91,944 59,301 222,461 103,028 63,035
2018
140,995 78,839 90,999 168,218 187,676 166,617 91,741 59,169 221,953 102,924 66,285
2019
141,289 78,839 90,999 167,597 186,139 166,283 91,563 59,052 221,508 102,836 69,310
2020
141,583 78,839 90,999 166,975 184,602 165,868 91,366 58,924 221,015 102,735 72,510
2021
141,877 78,839 90,999 166,354 183,066 165,630 91,209 58,821 220,624 102,661 75,337
2022
143,751 78,839 90,999 165,876 182,796 164,903 90,913 58,629 219,898 102,494 76,318
2023
145,625 78,839 90,999 165,398 182,527 164,281 90,650 58,459 219,251 102,348 77,039
2024
147,499 78,839 90,999 164,920 182,258 164,682 90,570 58,410 219,104 102,333 75,801
2025
149,373 78,839 90,999 164,442 181,988 165,557 90,532 58,390 219,097 102,355 73,844
2026
151,247 78,839 90,999 163,964 181,719 167,009 90,526 58,395 219,229 102,413 71,075
2027
153,122 78,839 90,999 163,486 181,449 169,153 90,540 58,419 219,503 102,506 67,398
2028
154,996 78,839 90,999 163,009 181,180 171,644 90,558 58,447 219,832 102,610 63,303
2029
156,870 78,839 90,999 162,531 180,911 174,125 90,567 58,471 220,142 102,708 59,253
2030
158,744 78,839 90,999 162,053 180,641 176,590 90,568 58,489 220,434 102,802 55,257
160,618 78,839 90,999 161,575 180,372 179,664 90,574 58,516 220,823 102,911 50,525
157
Annexes
&
16-03-20 12:28
Annexes
Annex 4: Comparison between land-use classifications for the

southeastern US
The forest land use category in NRI (USDA, 2013) was split into the 5 SRTS forest
categories by using the proportion of SRTS forest categories within the NRI total
forest area in 2010. Federal lands are defined as lands owned or administered by the
Federal government, and include national parks, national wildlife refuges, military
reservations, federal prisons, and public-domain land of at least 2,5 km2 in area. All
input maps were made into Boolean maps per land use category. The input maps had
an original resolution of 30 m. By resampling the Boolean maps to a resolution of 2
km, the mean value was calculated for all pixels included in the new, larger cells. This
created a map with values between 0 and 1, depicting the chance of the cell to contain
a specific land use based on the original 30 by 30 m pixels. The resulting maps will
from here on be referred to as land use probability maps. A python script was used to
assign the land use types to map cells using the area per state from NRI and expected
locations from the land use probability maps, using PCRaster software and following
the methodology developed by Verstegen et al. (Verstegen, Karssenberg, 2016).
Table A4: Crosswalk between statistical and spatial land-use classification
Land-use Source
NLCD categories GAP categories
type map
Water NLCDa Open water X
Developed, open space, Developed, low
Urban NLCDa intensity, Developed, medium intensity, X
Developed, high intensity
Federal lands USGSb X X
Cropland NLCDa Cultivated Crops X
Pasture NLCDa Pasture/Hay X
Non-forest Barren land, Shrub/Scrub, Grassland/
NLCDa X
vegetation herbaceous, Emergent Herbaceous wetlands
Evergreen Plantation or
Planted pine GAPc X Managed Pine, Managed
Tree Plantation
Natural pine NLCDa Evergreen forest X
Mixed forest NLCDa Mixed forest X
Upland
NLCDa Deciduous forest X
hardwood
Lowland
NLCDa Woody wetlands X
hardwood
a
Source: (Homer, 2015)
b
Source: (USGS, 2014)
c
Source: (US Geological Survey Gap Analysis Program (GAP), 2011)
158

Annexes
Annex 5: PLUC model input parameters
Allocation order
For the allocation of land cover types to a location, an allocation order was used. The
allocation order was determined by 1) a relative confidence of observation (e.g. in
remote sensing it is easier to distinguish urban area from other land uses than it is
to distinguish different forest types), 2) visual observation of land use patterns in
resulting maps and, 3) a minimization of the absolute difference between the input
statistics and the output areas. Non-forest vegetation was a passive land use, and water
and federal lands were no-go areas. The allocation order was as follows:
1. Urban
2. Cropland
3. Pine plantation
4. Lowland hardwood
5. Upland hardwood
6. Natural pine
7. Mixed forest
8. Pasture
&
Suitability factors
A number of suitability factors and their weights were used as input in the PLUC model.
These factors are listed in Table A5.
159

Table A5: Suitability factors per dynamic land use type and their weights as used as input in PLUC. D= direction. W= weight. CO= cut-off point, see also
160
(Verstegen, 2012). N= number of pixels in historical data that transitioned into land use type.
Annexes
Suitability Urban Cropland Pasture Planted pine Natural pine Mixed forest Upland Lowland
factors forest hardwood hardwood
N=1,821 N=459 N=396 X N=6,279 N=189 N=1,329 N=811
Allocation Allocation Allocation Allocation Allocation Allocation Allocation Allocation
order =1 order =2 order =8 order =3 order =6 order =7 order =5 order =4
D W CO D W CO D W CO D W CO D W CO D W CO D W CO D W CO
Elevation - 0.09 327 - 0.04 258 - 0.04 258 - 0.12 319 - 0.02 117
(m)
Slope (∘) - 0.03 0.597 - 0.04 0.390 - 0.04 0.590 - 0.04 0.590 + 0.10 0.663 - 0.02 1.193
Rainfall + 0.01 1,489 - 0.05 1,365 + 0.09 1,492 + 0.09 1,492 - 0.14 1,462
(mm)
Yield (kg + 0.03 7,959 - 0.22 7,990 + 0.05 8,018 + 0.05 8,018 - 0.12 7,853
DW/ha)
Growing - 0.10 + 0.19 + 0.19 + 0.18 + 0.09 + 0.01
season
(days)
c c c
Soil type 0.15 0.62 * 0.15 * 0.15 0.22
Distance to - 0.03 41 + 0.79 75
main roads
(km)
Distance to - 0.79 4 + 0.05 20 + 0.05 20 - 0.03 19
urban area
(km)
Distance to + 0.03 6 + 0.03 6 + 0.05 4 - 0.05
river (km)
Distance to - 0.07 83 - 0.04 77 - 0.04 77 - 0.05 33
wood mills
(km)
16-03-20 12:28
Table A5: Continued.
Suitability Urban Cropland Pasture Planted pine Natural pine Mixed forest Upland Lowland
factors forest hardwood hardwood
N=1,821 N=459 N=396 X N=6,279 N=189 N=1,329 N=811
Allocation Allocation Allocation Allocation Allocation Allocation Allocation Allocation
order =1 order =2 order =8 order =3 order =6 order =7 order =5 order =4
D W CO D W CO D W CO D W CO D W CO D W CO D W CO D W CO
Distance to - 0.10 280 - 0.04 389 - 0.04 389 - 0.16 445
main ports
(km)
Density self a - 0.01 + 0.15 + 0.08 + 0.08 + 0.20 + 0.04
Density of + 0.10 - 0.05 - 0.05 - 0.05 - 0.08 - 0.07
urban areab
Density of - 0.21 - 0.14 + 0.15 + 0.15 - 0.20 - 0.13 + 0.13
R2
forest areab
0.644 0.324 0.100 0.272 0.401 0.275 0.343
a
b
Number of pixels of the investigated land use with in a square of 3 by 3 pixels (6 by 6 km2).
c
Number of pixels of urban area or natural timberland within a square of 10 by 10 pixels (20 by 20 km2)
Soil types are treated as separate variables in the regression, and each of 10 soil types has its own relative weight and direction
161
Annexes
&
16-03-20 12:28
Annexes
Annex 6: Potential suitability factors for the southeastern US
The table below shows a list of potential suitability factors that were included in a
regression analysis to identify correlation between historical land use change (2001
to 2011 based on NLCD land cover maps), with spatial variables. It also shows the
sources that identified the factors as potentially linked to the spatial distribution of
land use in the project area. The outcome of this regression analysis was used to select
variables to include in the US version of the PLUC model. The list of potential suitability
factors was based on a literature review focused on the study region, as well as earlier
PLUC studies in other regions. The potential suitability factors are listed in Table A6.
Table A6: Potential suitability factors as included in the regression analysis, and the sources in which
a potential link with the distribution of one of the land uses was described.
Suitability factor Source

(Brown, 2005),(Lubowski, 2008),(Diogo, 2014),(van der Hilst, 2014)
Distance to city
,(Verstegen, 2012)
Distance to road (Napton, 2010),(Evans, 2013),(Verstegen, 2012)
Distance to railroads (Napton, 2010),(van der Hilst, 2014)
Distance to stream (Taverna, 2005),(Diogo, 2014),(Verstegen, 2012)
Distance to mill (Alig, 2004)
Distance to coast (Napton, 2010)
Soil (Napton, 2010),(Taverna, 2005),(Evans, 2013)
Rainfall (Napton, 2010)
Length of growing season (Napton, 2010),(Verstegen, 2014)
Slope (Napton, 2010),(Taverna, 2005),(Evans, 2013),(Verstegen, 2014)
Elevation (Evans, 2013)
(Lubowski, 2008),(Napton, 2010),(Diogo, 2014),(Verstegen,
Potential agricultural yield
2012),(Verstegen, 2014)
Distance to self (Diogo, 2014),(van der Hilst, 2014),(Verstegen, 2012)
Distance to port (Diogo, 2014),(van der Hilst, 2014)
Density of self (Verstegen, 2012)
Density of urban area
Density of forest area
162

Annexes
Annex 7: Spatial levels of the validation of PLUC results for the

southeastern US
The validation of the PLUC model for the southeastern US was performed at multiple
spatial levels. These levels are mapped in Figure A7; at ecoregion level (A7a), state
level (A7b), the USDA Forest Service Forest Inventory and Assessment (FIA) survey
unit level (A7c), and the county level (A7d).
a) b)
c) d) &
Figure A7: Maps at different spatial scale levels used in the validation of PLUC results, including
ecoregions (a), states (b), survey units (c) and counties (d).
163

Annexes
Annex 8: Land-use transitions for different scenarios in the

southeastern US
The PLUC model for the southeastern US projected land-use changes between 2010
and 2030. These are shown in Figure A8. The land-use transitions are categorised
based on the land use in 2030, for the HhLp scenario (A8a), the LhHp scenario (A8b),
and the LhLp scenario (A8c).
a)
No transition
To urban
To pasture
To pine plantation
b)
No transition
To urban
To pasture
To pine plantation
Figure
c) A8: Transitions in land use between 2010 and 2030 for a) the HhLp scenario, b) the LhHp
scenario, and c) the LhLp scenario. Colours show the land use type in 2030.
164
AnnaDuden_BNW.indd 164 Land-use transitions 16-03-20 12:28

To pasture
To pine plantation
Annexes
c)
No transition
To urban
To pasture
To pine plantation
Figure A8: Continued.
&
165

Annexes
Annex 9: Land-use transitions involving natural timberland in

the southeastern US
and 2030. Figure A9 shows the land-use transitions involving natural timberland for
the HhHp scenario (A9a) and the HhLp scenario (A9b).
a)

No transition
Reforestation
Natural timberland
b)

No transition
Reforestation
Natural timberland
Figure A9: Land use transitions between 2010 and 2030 involving natural timberland for a) the
HhHp scenario and b) the HhLp scenario.
166

Annexes
Annex 10: Land-use transitions involving pine plantation in the

southeastern US
and 2030. Figure A9 shows the land-use transitions involving pine plantation for the
HhHp scenario (A10a) and the HhLp scenario (A10b).
a)

No transition
Pine plantation
&
b)

No transition
Pine plantation
Figure A10: Land use transitions between 2010 and 2030 involving pine plantations for a) the HhHp
scenario and b) the HhLp scenario.
167

Annexes
Annex 11: Results sensitivity analysis for HhLp, LhLp and LhLp
scenarios
A sensitivity analysis was carried out for the SRTS model for the following input
variables; the cut-off point between small and large roundwood, elasticity values for
demand and supply, annual increase in agricultural rent and the percentage of large
roundwood that ends up in the small roundwood pool. Figure A11 shows the sensitivity
of total timberland area for the HhLp scenario (A11a), the LhHp (A11b) scenario and
the LhLp scenario (A11c).
a) b)
Agricultural rent increase Inventory % price
1,15 elasticities
1,1 1,15
1,05 1,1
1,05
1
1
0,95
0,95
0,9 0,9
0,85 0,85
c) d) % large rw to small rw
Feedstock cut-off point 1,15
1,1
1,15
1,1 1,05
1,05 1
1 0,95
0,95 0,9
0,9 0,85
0,85 low default high
9 11 13 % to small roundwood value
Cut-off point product definition
(inches) PP NP MX UH LH TT
Figure A11: Sensitivity of total timberland area in the HhLp (upper panel) scenario, the LhLp sce-
nario (middle panel) and the LhLp scenario (lower panel) to a) cut-off point between small and large
roundwood, b) elasticity values for demand and supply c) annual increase in agricultural rent d) the
percentage of large roundwood ending up in the small roundwood pool. For low, default and high
values for elasticities, agricultural rent and % to small roundwood value, see Table 3. The abbre-
viation rw stands for roundwood. Abbreviations in the legend: pp = pine plantation, np= natural
pine, mx= mixed pine, uh= upland hardwood, lh= lowland hardwood, tt= total.
168

Annexes
a) b)
Agricultural rent increase Inventory & price
Area change (indexed) 1,15 elasticities
1,1 1,15

1,05 1,1
1,05
1
1
0,95
0,95
0,9 0,9
0,85 0,85

1,1
1,15
1,1 1,05
1,05 1
1 0,95
0,95 0,9
0,9 0,85
&
a) b)
Agricultural rent increase Inventory & price
1,15 elasticities
1,1 1,15
1,1
1,05
1,05
1
1
0,95 0,95
0,9 0,9
0,85 0,85

1,1
1,15
1,1 1,05
1,05 1
1 0,95
0,95 0,9
0,9 0,85
Figure A11: Continued.
169

Annexes
Annex 12: Projections of pine plantation expansion in the

southeastern US
An overview of previous studies on pine plantation expansion. Table A10 provides a

list of studies and their outcomes for total timberland and pine plantation areas. It
also specifies the geographical and temporal scope of the study, as well as the main
variables in the scenarios.
Table A12: Projections of timberland area increase for total timberland and pine plantation of
previous studies in the southeastern US.
Timberland
area Time Geographical Main scenario
Forest type Study source
increase period extent rationale
max min
Price sensitivity,
Total 1995- (Prestemon,
-2% -15% US Southa pine plantation
timberland 2040 2002)
productivity
Price sensitivity,
Pine 1995- (Prestemon,
67% 25% US Southa pine plantation
plantations 2040 2002)
productivity
Total 2010- Cornerstone
-3% -10% US Southa (Wear, 2013b)
timberland 2060 futures
Pine 2010- Cornerstone
59% 18% US Southa (Wear, 2013b)
plantations 2060 futures
Bioenergy
Total 2007-
-5% -8% US Southeastb demand, forest (K. L. Abt, 2012)
timberland 2037
management
Bioenergy
Pine 2007-
3% -1% US Southeastb demand, forest (K. L. Abt, 2012)
plantations 2037
management
Total 2005- (R. C. Abt,
3% -3% US Southa Recession impacts
timberland 2030 2009b)
Pine 2005- (R. C. Abt,
22% 0% US Southa Recession impacts
plantations 2030 2009b)
Total 2010- US Coastal
-10% -10% Bioenergy demand (K. L. Abt, 2014)
timberland 2040 Southc
Pine 2010- US Coastal
3% -11% Bioenergy demand (K. L. Abt, 2014)
plantations 2040 Southc
a
KT, TX, TN, AL, AK, FL, GA, LA, VA, NC, SC, OK, MS
b
GA, FL, AL
c
some counties within FL, GA, VA, TX, NC, SC, MS, LA, AL
170

Annexes
A13 Threatened and endemic species in the southeastern US
The Latin names of all species included in the analysis as ‘threatened’ and/or endemic
are provided below in Table A13a (threatened species) and Table A13b (endemic
species). When species consisted of subspecies with separate threatened status, they
were included as separate species.
Table A13a: Threatened species in the study area per taxonomic group.
Mammals Birds Reptiles Amphibians

Canis lupus rufus Aimophila aestivalis Caretta caretta Ambystoma bishopi
Corynorhinus rafinesquii Amazona viridigenalis Chelonia mydas Ambystoma cingulatum
Corynorhinus Ammodramus Clemmys
Amphiuma pholeter
townsendii ingens caudacutus guttata
Corynorhinus Ammodramus Crocodylus
Anaxyrus houstonensis
townsendii virginianus savannarum floridanus acutus
Dermochelys
Eumops floridanus Anthus spragueii Aneides aeneus
coriacea
Aphelocoma Drymarchon Cryptobranchus
Eumops glaucinus
coerulescens couperi alleganiensis alleganiensis
Glaucomys sabrinus Eretmochelys Cryptobranchus
Aratinga mitrata
coloratus imbricata alleganiensis bishopi
Glaucomys sabrinus Charadrius Glyptemys
Desmognathus abditus &
fuscus alexandrinus nivosus insculpta
Lasionycteris Glyptemys
Charadrius melodus Desmognathus aeneus
noctivagans muhlenbergii
Gopherus
Lasiurus borealis Clangula hyemalis Desmognathus folkertsi
polyphemus
Graptemys
Lasiurus cinereus Dendroica cerulea Desmognathus imitator
barbouri
Microtus pennsylvanicus Graptemys
Euphagus carolinus Desmognathus santeetlah
dukecampbelli ernsti
Graptemys
Myotis grisescens Grus americana Desmognathus wrighti
flavimaculata
Graptemys
Myotis lucifugus Laterallus jamaicensis Eurycea aquatica
gibbonsi
Graptemys
Myotis septentrionalis Melanitta fusca Eurycea junaluska
nigrinoda
Oceanodroma Graptemys
Myotis sodalis Eurycea tynerensis
leucorhoa oculifera
Patagioenas Heterodon
Neofiber alleni Gyrinophilus gulolineatus
leucocephala gloydi
Neotoma floridana Heterodon
Picoides borealis Gyrinophilus palleucus
smalli simus
171

Annexes
Table A13a: Continued.

Lampropeltis
Neotoma magister Podiceps auritus Haideotriton wallacei
extenuata
Odocoileus virginianus Lepidochelys
Lithobates capito
clavium kempii
Oryzomys palustris Macrochelys
Lithobates okaloosae
natator temminckii
Peromyscus polionotus
Nerodia harteri Lithobates sevosus
allophrys
Peromyscus polionotus Ophisaurus
Necturus alabamensis
ammobates compressus
Peromyscus polionotus Ophisaurus
Necturus lewisi
niveiventris mimicus
Peromyscus polionotus Pituophis Notophthalmus
peninsularis ruthveni perstriatus
Peromyscus polionotus Plestiodon
Phaeognathus hubrichti
phasma reynoldsi
Peromyscus polionotus Pseudemys
Plethodon amplus
trissyllepsis alabamensis
Rhineura
Pipistrellus subflavus Plethodon aureolus
floridana
Sceloporus
Podomys floridanus Plethodon caddoensis
woodi
Sistrurus
Puma concolor coryi Plethodon chattahoochee
catenatus
Sternotherus
Rusa unicolor Plethodon cheoah
depressus
Sciurus niger cinereus Tantilla oolitica Plethodon fourchensis
Spilogale putorius Plethodon hubrichti
Sylvilagus palustris
Plethodon kiamichi
hefneri
Trichechus manatus Plethodon kisatchie
Ursus americanus
Plethodon meridianus
luteolus
Plethodon ouachitae
Plethodon petraeus
Plethodon punctatus
Plethodon savannah
Plethodon sequoyah
Plethodon shenandoah
Plethodon sherando
Plethodon shermani
Plethodon virginia
Plethodon websteri
Plethodon welleri
172

Annexes
Table A13b: Endemic species in the study area per taxonomic group.

Ammodramus
Canis lupus rufus Apalone ferox Acris gryllus
savannarum floridanus
Aphelocoma Apalone mutica
Eumops floridanus Ambystoma bishopi
coerulescens calvata
Crotalus Ambystoma
Geomys pinetis Picoides borealis
adamanteus cingulatum
Microtus pennsylvanicus
Drymarchon couperi Ambystoma mabeei
dukecampbelli
Glyptemys
Neofiber alleni Amphiuma means
muhlenbergii
Neotoma floridana Gopherus
Amphiuma pholeter
smalli polyphemus
Odocoileus virginianus
Graptemys barbouri Anaxyrus quercicus
clavium
Odocoileus virginianus
Graptemys ernsti Anaxyrus terrestris
nigribarbis
Oryzomys palustris Graptemys Desmognathus
natator flavimaculata abditus
Peromyscus polionotus Desmognathus
Graptemys gibbonsi
allophrys aeneus
Peromyscus polionotus Graptemys Desmognathus
ammobates nigrinoda apalachicolae &
Graptemys oculifera
leucocephalus brimleyorum
Graptemys pulchra
niveiventris carolinensis
Heterodon simus
peninsularis folkertsi
Kinosternon baurii
phasma imitator
Peromyscus polionotus Lampropeltis Desmognathus
trissyllepsis extenuata marmoratus
Desmognathus
Podomys floridanus Nerodia floridana
ocoee
Desmognathus
Puma concolor coryi Nerodia taxispilota
orestes
Scalopus aquaticus Ophisaurus Desmognathus
anastasae compressus santeetlah
Desmognathus
Sciurus niger avicennia Ophisaurus mimicus
wrighti
Sylvilagus palustris
Ophisaurus ventralis Eurycea aquatica
hefneri
Pituophis
Ursus americanus Eurycea
melanoleucus
floridanus chamberlaini
lodingi
173

Annexes
Table A13b: Continued.

Pituophis
Ursus americanus
melanoleucus Eurycea junaluska
luteolus
mugitus
Plestiodon egregius Eurycea wilderae
Gyrinophilus
Plestiodon reynoldsi
gulolineatus
Pseudemys Gyrinophilus
alabamensis palleucus
Haideotriton
Pseudemys nelsoni
wallacei
Pseudemys
Hyla andersonii
peninsularis
Pseudemys
Hyla femoralis
suwanniensis
Regina alleni Lithobates capito
Lithobates
Rhadinaea flavilata
heckscheri
Lithobates
Rhineura floridana
okaloosae
Sceloporus woodi Lithobates sevosus
Lithobates
Seminatrix pygaea
virgatipes
Sternotherus Necturus
depressus alabamensis
Storeria victa Necturus lewisi
Tantilla oolitica Necturus punctatus
Notophthalmus
Tantilla relicta
perstriatus
Terrapene carolina Phaeognathus
bauri hubrichti
Plethodon amplus
Plethodon aureolus
Plethodon
caddoensis
Plethodon
chattahoochee
Plethodon cheoah
Plethodon
chlorobryonis
Plethodon
fourchensis
Plethodon grobmani
Plethodon hubrichti
174

Annexes
Table A13b: Continued.

Plethodon jordani
Plethodon kiamichi
Plethodon kisatchie
Plethodon
meridianus
Plethodon metcalfi
Plethodon ocmulgee
Plethodon ouachitae
Plethodon petraeus
Plethodon savannah
Plethodon sequoyah
Plethodon
shenandoah
Plethodon sherando
Plethodon shermani
Plethodon teyahalee
Plethodon
variolatus
Plethodon virginia
Plethodon websteri
Plethodon welleri
&
Plethodon
yonahlossee
Pseudacris brimleyi
Pseudacris nigrita
Pseudacris ocularis
Pseudacris ornata
Pseudobranchus
axanthus
Pseudobranchus
striatus
Siren lacertina
Stereochilus
marginatus
175

Annexes
A14 Reclassification of GAP to PLUC land-use classification for the

southeastern US
The reclassification of GAP land-use classification to PLUC land uses was made in
order to determine the species richness values (based on GAP land use in 2001) per
land-use type in the land-use projections (based on PLUC output for 2010 and 2030).
The reclassification was relatively straightforward for the anthropogenic land-use
types classified as urban, cropland, pasture and pine plantation in PLUC, as well as
for water. The land-use type classified in PLUC as urban was made up out of 4 land-
use types of ‘Developed’ in the GAP land-use classification. The PLUC land-use type
‘cropland’ consists of 2 land-use types in the GAP classification; ‘Cultivated Cropland’
and ‘Deciduous Plantation’. The latter was considered as cropland because a check
of remotely sensed images showed the locations labelled as deciduous plantations
likely contained mostly fruit trees. The reclassification of pasture was a one-on-one
translation from ‘Pasture/Hay’ to ‘pasture’. Three land-use types of open water in the
GAP land-use classification were combined into a single water class. Pine plantation
was classified as including the GAP land-use types ‘Evergreen Plantation or Managed
Pine’ and ‘Managed Tree Plantation’. The difference between these land-use types is
a result of classification of different regional GAP projects, but they include the same
land use.
All other (natural) land-use types were split into those considered ‘forests’ and
those that were not considered to be forests, based on descriptions by NatureServe
(NatureServe, 2017) on which the GAP land-use map is based. Those land-use types
that were classified as containing forest and/or containing woodland were considered
‘forest’, while all land-use types that did not were considered ‘non-forest vegetation’.
Those land-use types distinguished as ‘forest’ were then further classified based on
NatureServe classification as 1) containing needleleaf trees, 2) containing broadleaf
trees, 3) being upland, and 4) having saturated soil. The forest types were determined
as depicted in Figure A14. Not all land-use types had data for all the points described
above. For all those land-use types for which classification was not straightforward, we
based the classification on additional detailed study of the description on NatureServe.
The resulting crosswalk between land-use types of GAP and PLUC is provided in Table
A14.
176

Annexes
Contains ‘forest’
no Land use is
and/or
‘other’
‘woodland’?
yes
Contains no yes Land use is ‘upland

Is ‘upland’?
needleleaf hardwood’
trees? no
Contains Land use is ‘lowland

yes yes
saturated soil? hardwood’
Contains no Land use is ‘natural

broadleaf trees? pine’
yes Land use is ‘mixed

forest’
Figure A14: Process steps of determining forest types in the reclassification of GAP land-use types
into PLUC land-use types.
The land-use projections also contain an area of federal lands, that were excluded from
land-use change because the forest in these areas does not respond to wood market
pressures. The area of open water also did not change during land-use projections.
We did not calculate biodiversity values for these land-use types.
&
Table A14: Reclassification crosswalk between Gap Analysis Program land-use types and PLUC
land-use types.
GAP land-use type PLUC land-use type

Developed, Open Space Urban
Developed, Low Intensity Urban
Developed, Medium Intensity Urban
Developed, High Intensity Urban
Quarries, Mines, Gravel Pits and Oil Wells Non-forest vegetation
Cultivated Cropland Cropland
Pasture/Hay Pasture
Open Water (Fresh) Water
Open Water (Brackish/Salt) Water
Open Water (Aquaculture) Water
Barren Land Non-forest vegetation
Atlantic Coastal Plain Southern Beach Non-forest vegetation
Unconsolidated Shore Non-forest vegetation
Louisiana Beach Non-forest vegetation
Florida Panhandle Beach Vegetation Non-forest vegetation
Atlantic Coastal Plain Sea Island Beach Non-forest vegetation
Southwest Florida Beach Non-forest vegetation
Central Interior Calcareous Cliff and Talus Non-forest vegetation
177

Annexes

North-Central Appalachian Circumneutral Cliff and Talus Non-forest vegetation
Southern Piedmont Cliff Non-forest vegetation
Southern Interior Acid Cliff Non-forest vegetation
Southern Appalachian Montane Cliff Non-forest vegetation
East Gulf Coastal Plain Dry Chalk Bluff Non-forest vegetation
Southern Appalachian Rocky Summit Non-forest vegetation
Southern Piedmont Granite Flatrock Non-forest vegetation
Central and Southern Appalachian Northern Hardwood Forest Upland hardwood
East Gulf Coastal Plain Limestone Forest Natural pine
East Gulf Coastal Plain Northern Loess Plain Oak-Hickory Upland -
Upland hardwood
Hardwood Modifier
Northeastern Interior Dry Oak Forest-Hardwood Modifier Upland hardwood
East Gulf Coastal Plain Northern Loess Plain Oak-Hickory Upland -
Natural pine
Juniper Modifier
Northeastern Interior Dry Oak Forest - Mixed Modifier Mixed forest
Ridge and Valley Calcareous Valley Bottom Glade and Woodland Upland hardwood
East Gulf Coastal Plain Black Belt Calcareous Prairie and Woodland
- Herbaceous Modifier
Southern and Central Appalachian Oak Forest - Xeric Upland hardwood
Northeastern Interior Dry-Mesic Oak Forest Mixed forest
Ozark-Ouachita Dry-Mesic Oak Forest Upland hardwood
Southern Interior Low Plateau Dry-Mesic Oak Forest Upland hardwood
East Gulf Coastal Plain Northern Dry Upland Hardwood Forest Upland hardwood
Crosstimbers Oak Forest and Woodland Upland hardwood
Ouachita Montane Oak Forest Upland hardwood
Southern and Central Appalachian Oak Forest Upland hardwood
Allegheny-Cumberland Dry Oak Forest and Woodland - Hardwood Upland hardwood
Central and Southern Appalachian Montane Oak Forest Upland hardwood
East Gulf Coastal Plain Northern Loess Bluff Forest Upland hardwood
East Gulf Coastal Plain Southern Loess Bluff Forest Mixed forest
Southern Coastal Plain Dry Upland Hardwood Forest Upland hardwood
South Florida Hardwood Hammock Upland hardwood
Atlantic Coastal Plain Dry and Dry-Mesic Oak Forest Upland hardwood
Southwest Florida Coastal Strand and Maritime Hammock Upland hardwood
Southeast Florida Coastal Strand and Maritime Hammock Upland hardwood
West Gulf Coastal Plain Chenier and Upper Texas Coastal Fringe
Upland hardwood
Forest and Woodland
Mississippi River Alluvial Plain Dry-Mesic Loess Slope Forest Upland hardwood
East-Central Texas Plains Post Oak Savanna and Woodland Upland hardwood
East-Central Texas Plains Riparian Forest Lowland hardwood
East-Central Texas Plains Floodplain Forest Lowland hardwood
South Florida Pine Rockland Natural pine
178

Annexes

Southern Coastal Plain Oak Dome and Hammock Upland hardwood
Ozark-Ouachita Dry Oak Woodland Upland hardwood
East Gulf Coastal Plain Maritime Forest Mixed forest
Lower Mississippi River Dune Woodland and Forest Upland hardwood
Mississippi Delta Maritime Forest Upland hardwood
Southern Piedmont Mesic Forest Upland hardwood
Southern Crowley’s Ridge Mesic Loess Slope Forest Upland hardwood
West Gulf Coastal Plain Mesic Hardwood Forest Upland hardwood
East Gulf Coastal Plain Northern Mesic Hardwood Forest Upland hardwood
South-Central Interior / Upper Coastal Plain Flatwoods Upland hardwood
Ozark-Ouachita Mesic Hardwood Forest Upland hardwood
East Gulf Coastal Plain Southern Mesic Slope Forest Upland hardwood
Atlantic Coastal Plain Central Maritime Forest Upland hardwood
Atlantic Coastal Plain Northern Maritime Forest Mixed forest
Atlantic Coastal Plain Southern Maritime Forest Mixed forest
East Gulf Coastal Plain Interior Shortleaf Pine-Oak Forest - Hardwood
Upland hardwood
Modifier
Southern Piedmont Dry Oak-(Pine) Forest - Hardwood Modifier Upland hardwood
East Gulf Coastal Plain Black Belt Calcareous Prairie and Woodland
Upland hardwood
- Woodland Modifier
Northeastern Interior Dry Oak Forest - Virginia/Pitch Pine Modifier Natural pine &
Southern Piedmont Dry Oak-(Pine) Forest - Loblolly Pine Modifier Natural pine
East Gulf Coastal Plain Northern Dry Upland Hardwood Forest -
Natural pine
Offsite Pine Modifier
Allegheny-Cumberland Dry Oak Forest and Woodland - Pine Modifier Natural pine
Southern Ridge and Valley Dry Calcareous Forest - Pine modifier Natural pine
East Gulf Coastal Plain Interior Shortleaf Pine-Oak Forest - Mixed
Mixed forest
Modifier
Southern Piedmont Dry Oak-(Pine) Forest - Mixed Modifier Mixed forest
Northern Atlantic Coastal Plain Dry Hardwood Forest Upland hardwood
West Gulf Coastal Plain Upland Longleaf Pine Forest and Woodland Natural pine
Southeastern Interior Longleaf Pine Woodland Natural pine
Ozark-Ouachita Shortleaf Pine-Oak Forest and Woodland Mixed forest
Central Appalachian Oak and Pine Forest Mixed forest
Appalachian Hemlock-Hardwood Forest Mixed forest
West Gulf Coastal Plain Pine-Hardwood Forest Mixed forest
Southern Ridge and Valley Dry Calcareous Forest Upland hardwood
West Gulf Coastal Plain Sandhill Oak and Shortleaf Pine Forest and
Mixed forest
Woodland
Northern Crowley’s Ridge Sand Forest Mixed forest
Southern and Central Appalachian Cove Forest Upland hardwood
South-Central Interior Mesophytic Forest Upland hardwood
Atlantic Coastal Plain Mesic Hardwood and Mixed Forest Upland hardwood
179

Annexes

East Gulf Coastal Plain Interior Upland Longleaf Pine Woodland -
Upland hardwood
Offsite Hardwood Modifier
Florida Longleaf Pine Sandhill - Open Understory Modifier Natural pine
Florida Longleaf Pine Sandhill - Scrub/Shrub Understory Modifier Natural pine
Atlantic Coastal Plain Fall-Line Sandhills Longleaf Pine Woodland -
Natural pine
Loblolly Modifier
Atlantic Coastal Plain Fall-line Sandhills Longleaf Pine Woodland -
Natural pine
Open Understory Modifier
Natural pine
Scrub/Shrub Understory Modifier
Natural pine
Loblolly Modifier
Natural pine
Open Understory Modifier
Natural pine
Scrub/Shrub Modifier
Atlantic Coastal Plain Upland Longleaf Pine Woodland Natural pine
Southern Appalachian Montane Pine Forest and Woodland Natural pine
Southern Appalachian Low Mountain Pine Forest Natural pine
East-Central Texas Plains Pine Forest and Woodland Natural pine
Ozark-Ouachita Shortleaf Pine-Bluestem Woodland Natural pine
Upland hardwood
Offsite Hardwood Modifier
Central and Southern Appalachian Spruce-Fir Forest Natural pine
Florida Peninsula Inland Scrub Non-forest vegetation
Tamaulipan Mixed Deciduous Thornscrub Non-forest vegetation
Tamaulipan Mesquite Upland Scrub Non-forest vegetation
Tamaulipan Calcareous Thornscrub Non-forest vegetation
Nashville Basin Limestone Glade Non-forest vegetation
Central Interior Highlands Calcareous Glade and Barrens Non-forest vegetation
West Gulf Coastal Plain Catahoula Barrens Non-forest vegetation
Atlantic Coastal Plain Xeric River Dune Non-forest vegetation
Southern Appalachian Grass and Shrub Bald - Shrub Modifier Non-forest vegetation
Southern Appalachian Grass and Shrub Bald - Herbaceous Modifier Non-forest vegetation
Western Great Plains Mesquite Woodland and Shrubland Non-forest vegetation
Eastern Highland Rim Prairie and Barrens - Dry Modifier Non-forest vegetation
Arkansas Valley Prairie and Woodland Non-forest vegetation
Southern Blackland Tallgrass Prairie Non-forest vegetation
Southeastern Great Plains Tallgrass Prairie Non-forest vegetation
Florida Dry Prairie Non-forest vegetation
West Gulf Coastal Plain Northern Calcareous Prairie Non-forest vegetation
West Gulf Coastal Plain Southern Calcareous Prairie Non-forest vegetation
180

Annexes

Atlantic Coastal Plain Southern Dune and Maritime Grassland Non-forest vegetation
Southwest Florida Dune and Coastal Grassland Non-forest vegetation
Texas-Louisiana Coastal Prairie Non-forest vegetation
Northern Atlantic Coastal Plain Dune and Swale Non-forest vegetation
Disturbed/Successional - Shrub Regeneration Non-forest vegetation
Disturbed/Successional - Grass/Forb Regeneration Non-forest vegetation
Harvested Forest - Shrub Regeneration Non-forest vegetation
Harvested Forest - Grass/Forb Regeneration Non-forest vegetation
Deciduous Plantation Cropland
Evergreen Plantation or Managed Pine Pine Plantation
Managed Tree Plantation Pine Plantation
Recently Burned Grassland Non-forest vegetation
Introduced Riparian and Wetland Vegetation Non-forest vegetation
Modified/Managed Southern Tall Grassland Non-forest vegetation
Disturbed, Non-specific Non-forest vegetation
Atlantic Coastal Plain Central Salt and Brackish Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Indian River Lagoon Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Embayed Region Tidal Salt and Brackish Marsh Non-forest vegetation
Mississippi Sound Salt and Brackish Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Northern Tidal Salt Marsh Non-forest vegetation
Florida Big Bend Salt-Brackish Tidal Marsh Non-forest vegetation &
Southern Coastal Plain Nonriverine Basin Swamp - Okefenokee
Lowland hardwood
Taxodium Modifier
Southern Coastal Plain Nonriverine Basin Swamp - Okefenokee Bay/
Lowland hardwood
Gum Modifier
Southern Coastal Plain Nonriverine Basin Swamp - Okefenokee Pine
Natural pine
Modifier
Nupea Modifier
Clethra Modifier
Atlantic Coastal Plain Peatland Pocosin Non-forest vegetation
Southern Coastal Plain Seepage Swamp and Baygall Lowland hardwood
West Gulf Coastal Plain Seepage Swamp and Baygall Lowland hardwood
East Gulf Coastal Plain Tidal Wooded Swamp Lowland hardwood
Atlantic Coastal Plain Streamhead Seepage Swamp - Pocosin - and
Mixed forest
Baygall
Atlantic Coastal Plain Southern Tidal Wooded Swamp Lowland hardwood
South Florida Everglades Sawgrass Marsh Non-forest vegetation
Gulf and Atlantic Coastal Plain Tidal Marsh Systems Non-forest vegetation
Floridian Highlands Freshwater Marsh Non-forest vegetation
Atlantic Coastal Plain Embayed Region Tidal Freshwater Marsh Non-forest vegetation
181

Annexes

Central Florida Herbaceous Seep Non-forest vegetation
Florida Big Bend Fresh-Oligohaline Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Northern Fresh and Oligohaline Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Northern Tidal Wooded Swamp Lowland hardwood
South Florida Mangrove Swamp Non-forest vegetation
West Gulf Coastal Plain Near-Coast Large River Swamp Lowland hardwood
South Florida Bayhead Swamp Lowland hardwood
Southern Coastal Plain Nonriverine Basin Swamp Lowland hardwood
Northern Atlantic Coastal Plain Basin Swamp and Wet Hardwood
Lowland hardwood
Forest
Southern Coastal Plain Herbaceous Seepage Bog Non-forest vegetation
Atlantic Coastal Plain Central Fresh-Oligohaline Tidal Marsh Non-forest vegetation
Atlantic Coastal Plain Nonriverine Swamp and Wet Hardwood Forest
Lowland hardwood
- Taxodium/Nyssa Modifier
Atlantic Coastal Plain Nonriverine Swamp and Wet Hardwood Forest
Lowland hardwood
- Oak Dominated Modifier
Atlantic Coastal Plain Clay-Based Carolina Bay Forested Wetland Lowland hardwood
South Florida Dwarf Cypress Savanna Non-forest vegetation
South Florida Wet Marl Prairie Non-forest vegetation
East Gulf Coastal Plain Savanna and Wet Prairie Non-forest vegetation
Texas Saline Coastal Prairie Non-forest vegetation
Eastern Great Plains Wet Meadow, Prairie, and Marsh Non-forest vegetation
Lower Mississippi River Bottomland Depressions - Forest Modifier Lowland hardwood
South Florida Cypress Dome Lowland hardwood
Southern Coastal Plain Nonriverine Cypress Dome Lowland hardwood
Central Florida Herbaceous Pondshore Non-forest vegetation
East Gulf Coastal Plain Southern Depression Pondshore Non-forest vegetation
Southern Piedmont/Ridge and Valley Upland Depression Swamp Upland hardwood
Atlantic Coastal Plain Depression Pondshore Non-forest vegetation
Mississippi River Bottomland Depression Lowland hardwood
South Florida Freshwater Slough and Gator Hole Non-forest vegetation
Atlantic Coastal Plain Blackwater Stream Floodplain Forest - Forest
Lowland hardwood
Modifier
Central Appalachian Riparian - Forest Modifier Upland hardwood
East Gulf Coastal Plain Large River Floodplain Forest - Forest
Lowland hardwood
Modifier
South-Central Interior Large Floodplain - Forest Modifier Lowland hardwood
Southern Piedmont Large Floodplain Forest - Forest Modifier Lowland hardwood
East Gulf Coastal Plain Large River Floodplain Forest - Herbaceous
Modifier
Western Great Plains Floodplain Systems Non-forest vegetation
Eastern Great Plains Floodplain Systems Non-forest vegetation
182

Annexes

Central Interior and Appalachian Floodplain Systems Upland hardwood
Mississippi River Riparian Forest Lowland hardwood
Cumberland Riverscour Non-forest vegetation
Mississippi River Low Floodplain (Bottomland) Forest Lowland hardwood
Southern Coastal Plain Hydric Hammock Lowland hardwood
West Gulf Coastal Plain Small Stream and River Forest Lowland hardwood
West Gulf Coastal Plain Large River Floodplain Forest Lowland hardwood
Southern Piedmont Small Floodplain and Riparian Forest Lowland hardwood
Atlantic Coastal Plain Small Brownwater River Floodplain Forest Lowland hardwood
Atlantic Coastal Plain Small Blackwater River Floodplain Forest Lowland hardwood
Red River Large Floodplain Forest Lowland hardwood
Atlantic Coastal Plain Brownwater Stream Floodplain Forest Natural pine
South-Central Interior Small Stream and Riparian Upland hardwood
East Gulf Coastal Plain Small Stream and River Floodplain Forest Lowland hardwood
Southern Coastal Plain Blackwater River Floodplain Forest Lowland hardwood
Texas-Louisiana Coastal Prairie Slough Non-forest vegetation
Mississippi River Floodplain and Riparian Forest Lowland hardwood
South-Central Interior Large Floodplain Lowland hardwood
Ozark-Ouachita Riparian Upland hardwood
East Gulf Coastal Plain Jackson Plain Dry Flatwoods - Open
Understory Modifier &
East Gulf Coastal Plain Near-Coast Pine Flatwoods - Offsite Hardwood
Upland hardwood
Modifier
East Gulf Coastal Plain Near-Coast Pine Flatwoods - Open Understory
Natural pine
Modifier
East Gulf Coastal Plain Near-Coast Pine Flatwoods - Scrub/Shrub
Natural pine
Understory Modifier
South Florida Pine Flatwoods Natural pine
Central Atlantic Coastal Plain Wet Longleaf Pine Savanna and
Natural pine
Flatwoods
Atlantic Coastal Plain Southern Wet Pine Savanna and Flatwoods Natural pine
West Gulf Coastal Plain Wet Longleaf Pine Savanna and Flatwoods Natural pine
Central Florida Pine Flatwoods Natural pine
East Gulf Coastal Plain Near-Coast Pine Flatwoods Natural pine
East Gulf Coastal Plain Southern Loblolly-Hardwood Flatwoods Mixed forest
South-Central Interior / Upper Coastal Plain Wet Flatwoods Upland hardwood
West Gulf Coastal Plain Pine-Hardwood Flatwoods Mixed forest
Lower Mississippi River Flatwoods Upland hardwood
West Gulf Coastal Plain Nonriverine Wet Hardwood Flatwoods Upland hardwood
Upper Texas Coast Beach Non-forest vegetation
Southeast Florida Beach Non-forest vegetation
South Florida Shell Hash Beach Non-forest vegetation
Texas Coastal Bend Beach Non-forest vegetation
183

Annexes

North-Central Appalachian Acidic Cliff and Talus Non-forest vegetation
Southern Interior Calcareous Cliff Non-forest vegetation
Southern Appalachian Granitic Dome Non-forest vegetation
Southern Appalachian Northern Hardwood Forest Upland hardwood
North-Central Interior Dry-Mesic Oak Forest and Woodland Upland hardwood
North-Central Interior Dry Oak Forest and Woodland Upland hardwood
North-Central Interior Maple-Basswood Forest Upland hardwood
Central and South Texas Coastal Fringe Forest and Woodland Upland hardwood
Edwards Plateau Limestone Savanna and Woodland Upland hardwood
Southern Piedmont Dry Oak-Heath Forest - Mixed Modifier Mixed forest
Appalachian Shale Barrens Non-forest vegetation
Central Interior Highlands Dry Acidic Glade and Barrens Non-forest vegetation
Southern Piedmont Dry Oak-(Pine) Forest Mixed forest
Central Appalachian Pine-Oak Rocky Woodland Mixed forest
Northeastern Interior Pine Barrens Natural pine
Northern Atlantic Coastal Plain Pitch Pine Barrens Natural pine
Southern Piedmont Glade and Barrens Non-forest vegetation
Central Appalachian Alkaline Glade and Woodland Non-forest vegetation
West Gulf Coastal Plain Nepheline Syenite Glade Non-forest vegetation
Central Mixedgrass Prairie Non-forest vegetation
Western Great Plains Sand Prairie Non-forest vegetation
North-Central Interior Sand and Gravel Tallgrass Prairie Non-forest vegetation
East Gulf Coastal Plain Jackson Prairie and Woodland Non-forest vegetation
Tamaulipan Savanna Grassland Non-forest vegetation
East Gulf Coastal Plain Dune and Coastal Grassland Non-forest vegetation
Central and Upper Texas Coast Dune and Coastal Grassland Non-forest vegetation
Southern Appalachian Grass and Shrub Bald Non-forest vegetation
Introduced Upland Vegetation - Treed Upland hardwood
Introduced Upland Vegetation - Annual Grassland Non-forest vegetation
Ruderal Wetland Non-forest vegetation
Atlantic Coastal Plain Clay-Based Carolina Bay Herbaceous Wetland Non-forest vegetation
Central Interior and Appalachian Swamp Systems Lowland hardwood
Atlantic Coastal Plain Large Natural Lakeshore Lowland hardwood
Central Interior and Appalachian Shrub-Herbaceous Wetland
Systems
Atlantic and Gulf Coastal Plain Interdunal Wetland Non-forest vegetation
Southern and Central Appalachian Bog and Fen Non-forest vegetation
184

Annexes
A15. Neighbourhood analysis for the southeastern US
Species richness index was calculated in a spatial analysis using a moving window
approach. The analysis was based on the input maps described in Table A15. The
starting point for the analysis was the 2001 GAP land use map, this is the map on
which GAP habitat suitability is based. After reclassification of the GAP land use types
into PLUC land use types, we created maps for each PLUC land use type separately
(urban, cropland, pasture, non-forest vegetation, pine plantation, natural pine forest,
mixed forest, upland hardwood forest and lowland hardwood forest), see step 1 in
Figure A15. These maps are at a spatial resolution of 30 x 30 m. These maps contain
values of 1 for the ‘target land use’ (cropland in the example in Figure A15), and NoData
for all other land uses. We then multiplied these maps with the different potential
species richness maps (step 2), which also have a resolution of 30 x 30 m, to create
maps of species richness for each land use type separately. These land-use specific
potential species richness maps were then combined with a PLUC land use map (with
a resolution of 2 x 2 km), using a moving window approach in the following way. For
each cell (‘target cell’) in the PLUC land use map (the 2010 map in the example in
Figure A15), we determined the projected land use. We then took the corresponding
land-use specific potential species richness map for that particular land use (created
in step 2 for cropland in the example in A15), and created a 39 x 39 km calculation &
window around the target cell (step 3). We calculated the species richness index for
the target cell by averaging all species richness values within the calculation window
(step 4). The calculated species richness index value was then assigned to the target
cell (step 5), creating species richness index values at a resolution of 2 x 2 km (based
on averaged values at 39 x 39 km resolution). This procedure was repeated for all cells
in the map until a complete map of species richness index was created based on the
PLUC map (PLUC 2010 map in the example in Figure A15).
This process was repeated for all indicators of potential species richness (total,
threatened, endemic, mammal, bird, reptile and amphibian) and for all PLUC output
maps (2010, 2030 HhHp, 2030 LhHp, 2030 HhLp, 2030 LhLp).
185

Annexes
Table A15: Input maps used in the analysis to calculate spatially variable species richness index
values.
Map type Map Source

Total potential species richness
Threatened potential species richness
Potential Endemic potential species richness
species Mammal potential species richness (McKerrow, 2018)
richness Bird potential species richness
Reptile potential species richness
Amphibian potential species richness
2001 GAP land use map (USGS, 2011)
2010 PLUC land use map
Land use 2030 PLUC map for high housing-high pellet scenario (HhHp)
maps 2030 PLUC map for high housing-low pellet scenario (HhLp) (Duden, 2017)
2030 PLUC map for low housing-high pellet scenario (LhHp)
2030 PLUC map for low housing-low pellet scenario (LhLp)
186

a) GAP 2001 land use c) GAP 2001 land use: cropland e) PLUC 2010 land use: cropland g) Window calculation
1)
2) 3) 4) 5)
2) 3)
b) Potential species richness d) Potential species richness in f) Potential species richness in h) Potential species richness in
(based on GAP 2001 land use)
Figure A15: A schematic visual representation of the analysis steps to calculate species richness index values. The calculation uses cropland in the 2010 PLUC
cropland (based on 2001 LU) cropland (based on 2001 LU) cropland (based on 2010 LU)
map as an example. The figure is not made to scale, i.e. cell sizes were different from the cells depicted in the figure and the ratio between different resolutions
and calculation window is not to scale. Figure A15e: black square = ‘target cell’. Figure A15f: small square = target cell, large square is calculation window.
Figure A15g: small square = target cell, large square is calculation window, beige cells = cells with target land use outside of calculation window, orange
cells = cells with target land use inside calculation window and included in the calculation of species richness index.
187
Annexes
&
16-03-20 12:28
Annexes
A16. Potential species richness patterns per land-use type in the

southeastern US
Potential total species richness per land-use type; for urban area (Figure A16a),
cropland (Figure A16b), pasture (Figure A16c), non-forest vegetation (Figure A16d),
pine plantation (Figure A16e), natural pine forest (Figure A16f), mixed forest (Figure
A16g), upland hardwood forest (Figure A16h) and lowland hardwood forest (Figure
A16i). Black lines show state boundaries, with abbreviations for state names.
Figure A16a: Potential total species richness in urban area. White areas do not contain urban area
and are not included.
188

Annexes
Figure A16b: Potential total species richness in cropland. White areas do not contain cropland and
are not included.
&
Figure A16c: Potential total species richness in pasture. White areas do not contain pasture and
are not included.
189

Annexes
Figure A16d: Potential total species richness in non-forest vegetation. White areas do not contain
non-forest vegetation and are not included.
Figure A16e: Potential total species richness in pine plantation. White areas do not contain pine
plantation and are not included.
190

Annexes
Figure A16f: Potential total species richness in natural pine forest. White areas do not contain
natural pine forest and are not included.
&
Figure A16g: Potential total species richness in mixed forest. White areas do not contain mixed
forest and are not included.
191

Annexes
Figure A16h: Potential total species richness in upland hardwood. White areas do not contain upland
hardwood and are not included.
Figure A16i: Potential total species richness in lowland hardwood. White areas do not contain
lowland hardwood and are not included.
192

Annexes
A17. Maps of species richness for the southeastern US
Potential species richness for species subsets; for threatened species (Figure A17a),
for endemic species (Figure A17b), and for the taxonomic groups amphibians (A17c),
birds (A17d), mammals (A17e) and reptiles (Figure A17f). White areas in the maps are
areas of either open water or Federal lands. These land uses did not change over time
and were excluded in the analysis of potential species richness.
&
Figure A17a: Potential threatened species richness in number of species per 30x30m cell for 2010.
Black lines show state boundaries, abbreviations show abbreviated state names.
193

Annexes
Figure A17b: Potential endemic species richness in number of species per 30x30m cell for 2010. Black
lines show state boundaries, abbreviations show abbreviated state names.
Figure A17c: Potential amphibian species richness in number of species per 30x30m cell for 2010.
194

Annexes
Figure A17d: Potential bird species richness in number of species per 30x30m cell for 2010. Black
&
Figure A17e: Potential mammal species richness in number of species per 30x30m cell for 2010.
195

Annexes
Figure A17f: Potential reptile species richness in number of species per 30x30m cell for 2010. Black
196

Annexes
A18. Species richness per land-use type in the southeastern US
Species richness per land-use type (urban, cropland, pasture, planted pine, natural
pine, mixed forest, upland hardwood, lowland hardwood and non-forest vegetation)
and per biodiversity indicator (threatened, endemic, mammals, birds, reptiles, and
amphibian species richness) for 2010.
&
Figure A18: Potential total species richness per land-use type and per species richness indicator
in the southeastern US for 2010 in index values. Boxplots show median values (middle bar) and
first (upper limit) and third quartile (lower limit) values, while bars show maximum and minimum
values. TH = threatened, EN = endemic, MA = mammals, BI = birds, RE= reptiles, AM = amphibians.
197

Annexes
A19. Maps of biodiversity impact due to drivers other than

increasing wood pellet demand in the southeastern US
Changes (in index) between 2010 and 2030 in potential total, threatened and endemic
species richness (Figure A19a) and species richness per taxonomic group (Figure
A19b) due to drivers other than increasing wood pellet demand for the low housing
demand scenario (LhLp).
Figure A19a: Changes (in index values) between 2010 and 2030 in potential total, threatened and
endemic species richness due to drivers other than increasing wood pellet demand for the low
housing scenario.
198

Annexes
&
Figure A19b: Changes (in index values) between 2010 and 2030 in potential species richness per tax-
onomic group due to drivers other than increasing wood pellet demand for the low housing scenario.
199

Annexes
A20. Potential species richness impact of land-use transitions in

the southeastern US
Land-use transitions that lead to negative (Figure A20a) and positive (Figure A20b) changes
in species richness due to drivers other than increasing wood pellet demand, expressed in
the percentage of total area of positive or negative changes in species richness.
High housing - low pellet scenario Low housing - low pellet scenario
ulhw - urban ulhw - urban

pasture - urban pasture - urban
llhw - urban llhw - urban
npine - ppine npine - ppine
pasture - ppine pasture - ppine
mixed - ppine mixed - ppine
pasture - nfveg pasture - nfveg
pasture - npine pasture - npine
mixed - npine mixed - npine
pasture - mixed pasture - mixed
nfveg - pasture nfveg - pasture
other other
Figure A20a: Land-use transitions causing negative changes in species richness between 2010 and 2030
(in % of total area with negative richness changes). Ulhw = upland hardwood, llhw = lowland hardwood,
npine = natural pine forest, ppine = pine plantation, mixed = mixed forest, nfveg = non-forest vegetation.
High housing - low pellet scenario Low housing - low pellet scenario
pasture - npine
pasture - npine
nfveg - mixed nfveg - mixed

npine - ulhw npine - ulhw
npine - llhw npine - llhw
pasture - nfveg pasture - nfveg
cropland - nfveg cropland - nfveg
ppine - nfveg ppine - nfveg
cropland - pasture cropland - pasture
ppine - pasture ppine - pasture
cropland - urban
cropland - urban
other
other
Figure A20b: Land-use transitions causing positive changes in species richness between 2010 and 2030
(in % of total area with positiverichness changes). Ulhw = upland hardwood, llhw = lowland hardwood,
npine = natural pine forest, ppine = pine plantation, mixed = mixed forest, nfveg = non-forest vegetation.
200

Annexes
A21. Biodiversity impact due to increasing wood pellet demand in

the southeastern US
Differences (in index) in potential total, threatened and endemic species richness
(Figure A21a) and species richness per taxonomic group (Figure A21b) by 2030 due
to drivers other than increasing wood pellet demand for the low housing demand
scenarios.
&
Figure A21a: Differences (in index values) in potential total, threatened and endemic species richness
by 2030 due to drivers other than increasing wood pellet demand for the low housing scenario.
201

Annexes
Figure A21b: Differences (in index values) in potential species richness per taxonomic group by
2030 due to increasing wood pellet demand for the low housing scenario.
202

Annexes
A22. Potential species richness impacts of land-use transitions in

the southeastern US
Land-use transitions that lead to negative (Figure A22a) and positive (Figure A22b)
changes in species richness due to increasing demand for wood pellets, expressed in
the percentage of total area of positive or negative changes in species richness.
High housing scenarios Low housing scenarios

npine - ppine npine - ppine
pasture - ppine pasture - ppine
mixed - ppine mixed - ppine
ulhw - ppine ulhw - ppine
nfveg - ppine nfveg - ppine
llhw - ppine llhw - ppine
mixed - npine mixed - npine
pasture - ulhw pasture - ulhw
other other
Figure A22a: Land-use transitions causing negative changes in species richness between 2010 and
2030 (in % of total area with negative richness changes). Ulhw = upland hardwood, llhw = lowland
hardwood, npine = natural pine forest, ppine = pine plantation, mixed = mixed forest, nfveg = non-for-
est vegetation.
&
High housing scenarios Low housing scenarios

ppine - pasture ppine - pasture
npine - ulhw npine - ulhw
pasture - ulhw pasture - ulhw
npine - llhw npine - llhw
mixed - ulhw mixed - ulhw
nfveg - npine nfveg - npine
pasture - llhw pasture - llhw
mixed - llhw mixed - llhw
nfveg - mixed nfveg - mixed
other other
Figure A22b: Land-use transitions causing positive changes in species richness between 2010 and
2030 (in % of total area with positive richness changes). Ulhw = upland hardwood, llhw = lowland
hardwood, npine = natural pine forest, ppine = pine plantation, mixed = mixed forest, nfveg = non-for-
est vegetation.
203

Annexes
A23. Land-use change projections for Brazil
Projected changes in land use for Brazil were taken from previous studies (van der
Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016) investigating the potential land-
use changes following scenarios of demand for ethanol in Brazil. Chapter 4 and 5
compare land use in 2012 to land use in 2030 for a reference and an ethanol scenario.
Land use in 2012 and 2030 (for both scenarios) is provided in Figure A23, as well as the
changes between 2012 and 2030 (for both scenarios) and the difference between the
two scenarios in 2030. Tables A23a and A23b show the area of sugarcane expansion
and other land-use changes between 2012 and 2030 in the reference scenario and in
the ethanol scenario respectively. The area of sugarcane expansion between 2012 and
2030 is larger than the difference in demand for sugarcane cultivation area because
in the and-use projections, some of the original sugarcane area in 2012 is lost and
established elsewhere.
Land use Δ Land use Δ Land use

9.000.000 500.000 40.000
8.000.000 400.000
7.000.000 300.000
20.000
200.000
6.000.000
Area (km2)
Area (km2)
Area (km2)
100.000
5.000.000 0
0
4.000.000
-100.000
3.000.000
-200.000 -20.000
2.000.000 -300.000
1.000.000 -400.000
-40.000
0 -500.000 2030 eth - 2030
2012 2030 ref 2030 eth 2030 ref - 2012 2030 eth - 2012 ref
natural forest grass & shrubs rangeland natural forest grass & shrubs abandoned planted forest
pasture crops sugarcane rangeland pasture sugarcane crops
planted forest urban water crops sugarcane pasture rangeland
bare soil abandoned planted forest abandoned grass & shrubs natural forest
Figure A23: Area (km2) of projected land use and land-use changes (∆ land use) in PLUC projections.
A) land use in 2012 and 2030 for the reference scenario (ref) and 2030 for the ethanol scenario (eth);
B) land-use change between 2012 and 2030 for the reference scenario (ref) and the ethanol scenario
(eth); C) difference in land use in 2030 between the ethanol scenario (eth) and the reference (ref).
Table A23a: Sugarcane expansion and other land-use transitions between 2012 and 2030 in the
reference scenario.
Other land-use Sugarcane Sugarcane expansion (% of

Ecoregion
change (km2) expansion (km2) total land-use change)
Amazon 97,200 75 0.08
Caatinga 55,225 0 0
Cerrado 197,175 12,225 5.84
Atlantic Forest 64,750 7,225 10.04
Pantanal 57,950 0 0
Pampas 14,125 0 0
Total 487,950 19,525 3.86
204

Annexes
Table A23b: Sugarcane expansion and other land-use transitions between 2012 and 2030 in the
ethanol scenario.
Other land-use Sugarcane Sugarcane expansion (% of

Ecoregion
change (km2) expansion (km2) total land-use change)
Amazon 97,975 250 0.25
Caatinga 54,775 50 0.09
Cerrado 196,425 28,950 12.85
Atlantic Forest 59,575 20,600 25.69
Pantanal 67,225 25 0.04
Pampas 13,500 0 0
Total 489,475 49,875 9.25
&
205

Annexes
A24. Reclassification of land cover categories for Brazil from

Globcover to PLUC
We aggregated some land-use categories from the original land-use projections for
Brazil (van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016) in order to align
them better with the Globcover land-use categories on which the habitat suitability
data was based. We defined ‘forest’ as including both ‘plantation forest’ and ‘natural
forest’, ‘grass’ as both ‘rangeland’ and ‘pasture’, ‘shrub’ as ‘grass & shrubs’ and
‘abandoned land & bare soil’ as both ‘bare soil’ and ‘abandoned’.
The Globcover v2.3 land cover map (Bontemps, 2011), on which the habitat suitability
maps of the Global Mammal Assessment (Rondinini, 2011) were based, was reclassified
in order to align it to the PCRaster Land Use Change (PLUC) land-use projections
(van der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016). Globcover land cover
categories contain composite land cover categories, representing a mosaic of several
land covers. The reclassification of Globcover natural land cover types is consistent
with reclassification of the GMA (Rondinini, 2011). The reclassification of Globcover
to PLUC classification consisted of the following steps:
• Globcover land cover categories were translated to PLUC land use categories
using Table A24
• Globcover land cover categories containing multiple land cover types including
cropland:
• If the majority of the land cover was cropland (50% or more), the land use
type was considered to be cropland (see Table A24)
• If the majority of the land cover was natural vegetation (50% or more),
the land use type was considered to be natural vegetation, to be further
reclassified as described below (also see Table A24)
Globcover land cover categories containing multiple natural land cover types were
further divided by overlaying them with a reclassified map of Brazilian biomes,
adapted from the WWF terrestrial biome map (Olson, 2001) and a land cover map for
the Cerrado (MMA, 2019). The Biomes and Cerrado maps were classified to the land
use types ‘forest’, ‘shrubs’ and ‘grass’ as follows:
• Classes are considered forest if Cerrado (described as ‘f loresta’,
‘reflorestamento’ or ‘savanna florestada’) or Biome (description includes
‘forest’ or ‘mangrove’) map describe as forest type
• Classes are considered shrubs if Biome (caatinga, restinga, savanna) and
Cerrado (savanna arborizada, savanna parquet, formaloes pioneiras,
vegetacao secundaria) describe it as shrubland
206

Annexes
• A cell is considered grass if Cerrado (savanna gramineo-lenhosa) and Biome

(Pantanal) map describe as grassland type
The Cerrado land use map was considered dominant over the Biome map, because it
had higher spatial detail. This means that when a cell was considered shrubs in the
Biome map and grass in the Cerrado map, it was reclassified as grass.
Globcover does not separate between sugarcane and other crops, all cropland is
classified as cropland. Sugarcane was categorized in this study as being classified
as both cropland in Globcover (as reclassified using Table A24) and sugarcane in the
2012 PLUC land use map
Cropland is categorized in this study as both cropland in Globcover (as classified using
Table A24) and any land use but sugarcane in the PLUC 2012 map.
&
207

Table A24: Reclassification of Globcover land cover classes to PLUC land use projections. Bare = abandoned land & bare soil
208
Annexes
Forest
Urban
Water
Shrubs
Grass
Crop
Sugarcane
Bare*
Closed to open (>15%) broadleaved evergreen or semi-deciduous forest (>5m) x

Closed to open (>15%) (broadleaved or needleleaved, evergreen or deciduous) shrubland (<5m) x
Mosaic cropland (50-70%) / vegetation (grassland/shrubland/forest) (20-50%) x x
Closed (>40%) broadleaved deciduous forest (>5m) x
Closed to open (>15%) herbaceous vegetation (grassland, savannas or lichens/mosses) x x
Mosaic vegetation (grassland/shrubland/forest) (50-70%) / cropland (20-50%) x x x
Closed to open (>15%) broadleaved forest regularly flooded (semi-permanently or temporarily) - Fresh
x
or brackish water
Rainfed croplands x x
Closed to open (>15%) grassland or woody vegetation on regularly flooded or waterlogged soil - Fresh,
x x x
brackish or saline water
Mosaic forest or shrubland (50-70%) / grassland (20-50%) x x x
Water bodies x
Sparse (<15%) vegetation x
Bare areas x
Open (15-40%) broadleaved deciduous forest/woodland (>5m) x
Mosaic grassland (50-70%) / forest or shrubland (20-50%) x x x
Permanent snow and ice x
Artificial surfaces and associated areas (Urban areas >50%) x
Closed (>40%) broadleaved forest or shrubland permanently flooded - Saline or brackish water x x
16-03-20 12:28
Annexes
A25. Window sizes for spatial neighbourhood analysis for Brazil
The spatial neighbourhood analysis used to calculate species richness index makes use
of four different window sizes to calculate average potential species richness based on
land-use type. These window sizes were 15, 85 and 395 km, and finally a map average
of all of Brazil. Window sizes were chosen so that 75% of all cells falls within the
smallest window, while another 10% fall into the second and third largest windows
(Table A25). Another 5% of cells is assigned the average for all of Brazil. Larger window
sizes were applied particularly in the Cerrado region and the ‘arc of deforestation’,
the area at the edges of the Amazon rainforest (Figure A25). For all land use types
but shrubs and abandoned land & bare soil, species richness index for most cells is
calculated using the 15 km sized window (Table A25). For abandoned land & bare soil
and shrubs, larger window sizes were necessary. This may indicate that the land use
classification for these land-use types in Globcover and PLUC did not align completely.
Table A25: Area (km2) and percentage of cells (%) using the window sizes 15, 85, 395 and for the
whole of Brazil in the spatial neighbourhood analysis, per land-use type and for the total map.
Abandoned = abandoned land & bare soil.
Window size à 15 km 85 km 395 km Whole Brazil

Land-use type Km % Km % Km2 % Km2 %
&
2 2
Urban 4,425 75 700 12 550 9 225 4

Water 139,850 86 1,275 13 1,275 1 0 0
Forest 4,471,650 96 4,525 4 4,525 0 0 0
Shrubs 186,325 12 744,250 16 744,250 49 332,825 22
Grass 377,275 93 600 7 600 0 0 0
Crop 1,144,525 75 35,975 23 35,975 2 0 0
Sugarcane 95,125 96 75 4 75 0 250 0
Abandoned 4,425 2 68,800 10 68,800 35 101,825 52
Total 6,423,600 75 855,600 10 856,050 10 435,125 5
209

Annexes
AM
CE CA
Window size (km) PA
AF
PM
Figure A25. Window size (in km) applied in the calculation of SRI in 2012 in the spatial analysis.
Black lines show biome boundaries, abbreviations show biome names; AM = Amazon, CA = Caatinga,
CE = Cerrado, PN = Pantanal, AF = Atlantic Forest, PM = Pampas.
210

Annexes
A26. Potential species richness in Brazil in 2012
Threatened species richness index in 2012 ranged from 0 to 14 (Figure A26a). Endemic
species richness index ranged from 0 to 9 (Figure A25b). Range-restricted species
richness index ranged from 0 to 9 (Figure A25c).
a) b)
&
c)
Figure A26. Species richness index in 2012 for threatened (a), endemic (b) and range-restrict-
ed (c) species in Brazil. Black lines show biome boundaries, abbreviations show biome names;
AM = Amazon, CA = Caatinga, CE = Cerrado, PN = Pantanal, AF = Atlantic Forest, PM = Pampas.
211

Annexes
A27. Projected land-use change in Brazil resulting in changes in

species richness index
The projected changes in land use are translated into changes in species richness index.
Figure A27 show the range and mean total SRI value per ecoregion and per land-use
type for grassland, cropland and sugarcane. Table A27a and A27b provide summary
statistics of the type of aggregated land-use transitions and their combined species
richness changes. Table A27a shows the added SRI losses of all land-use transitions
that consist of loss of forest, loss of shrubland, loss of grassland, and all remaining
land-use transitions. It also provides the percentage in total SRI losses that each land-
use transition categories represents. Table A27b shows the same information, but for
transitions into sugarcane, cropland, grassland and other land-use types. Table A27c
provides the average SRI change per land-use transition per ecoregion between 2012
and 2030 (in the reference scenario). Table A27d provides the same information for
differences in land-use in 2030 between the reference and the ethanol scenario.
160
140
120
100
80
60
40
20
Ecoregion - land-use type

Figure A27. Species richness index (SRI) per ecoregion and land‐use type in Brazil for 2012. The
ecoregions are described in the figure as follows; PA = Pantanal, CE = Cerrado, AF = Atlantic Forest,
CA = Caatinga, AM = Amazon, PM = Pampas. The land-use types are described in the Figure as fol-
lows; gr = grassland, cr = cropland, sc = sugarcane. Boxplots show median values (middle horizontal
bar) and first (upper limit) and third quartile (lower limit) values, while vertical lines indicate
maximum and minimum values.
212

Annexes
Table A27a: Summary of the area of difference in land use between the ethanol and the reference
scenario and the total SRI losses represented by each category.
Loss of: Area (km2) % area Total SRI loss % SRI loss
Forest 8,450 16.1 -4689.4 27.3
Shrubs 19,800 37.8 -6312.3 36.8
Grass 10,900 20.8 -2660.2 15.5
Other 13,225 25.3 -3493.9 20.4
Total 52,375 100.0 -17155.8 100.0
Table A27b: Summary of the area of difference in land use between the ethanol and the reference
scenario and the total SRI losses represented by each category.
Gains of: Area (km2) % area Total SRI loss % SRI loss
Sugarcane 20,100 38.4 -3950.0 23.0
Crops 17,575 33.6 -6081.3 35.4
Grass 2,800 5.3 -2358.3 13.7
Other 11,900 22.7 -4766.2 27.8
Total 52,375 100.0 -17155.8 100.0
Table A27c: Area (in km2) per land-use transition, provided as land use in 2012 and land use in 2030
for the reference scenario, average change in species richness index (SRI) and standard deviation
(st. d.) of species richness index. &
Land use Land use Area Average
Biome St. d.
2012 reference 2030 (km2) ∆SRI
forest grass 30,200 -66.2 28.6
forest bare soil 25 -45.0 0.0
shrubs grass 12,250 -19.2 16.3
forest cropland 19,425 -15.5 18.6
cropland grass 250 -8.8 9.7
grass bare soil 925 -6.0 6.3
shrubs forest 325 -5.1 7.9
shrubs cropland 4,125 -3.3 14.9
Amazon grass forest 275 -1.0 5.6
sugarcane grass 325 0.9 11.5
sugarcane bare soil 25 3.5 0.0
cropland forest 25 3.6 0.0
grass cropland 1,925 4.4 11.0
bare soil grass 2,925 4.7 2.8
grass sugarcane 75 6.7 8.5
sugarcane cropland 100 14.7 39.1
bare soil cropland 24,075 14.7 30.7
213

Annexes
Table A27c: Continued.

Biome St. d.
forest bare soil 1,925 -7.4 9.9
grass bare soil 1,725 -4.2 9.1
bare soil grass 650 -4.1 5.4
shrubs grass 675 -4.0 5.7
grass forest 125 -1.8 2.6
grass cropland 1,250 -1.1 6.4
Caatinga
bare soil cropland 15,325 -0.4 9.6
shrubs forest 7,000 0.7 5.9
forest grass 350 2.9 7.7
sugarcane cropland 175 3.5 3.3
cropland forest 3,475 8.9 7.4
bare soil forest 125 19.5 8.3
grass sugarcane 425 -10.8 3.7
forest sugarcane 250 -8.1 15.2
bare soil forest 350 -5.9 12.1
bare soil cropland 35,925 -3.4 10.8
shrubs sugarcane 400 -2.3 4.1
Cerrado
shrubs forest 6,025 -0.3 8.4
sugarcane cropland 2,425 0.0 4.5
forest grass 8,175 1.9 12.6
sugarcane bare soil 1,050 2.4 9.0
bare soil sugarcane 150 6.6 10.7
shrubs grass 12,400 8.2 8.8
sugarcane grass 575 15.6 6.2
cropland grass 1,025 17.5 4.7
cropland sugarcane 11,000 42.0 0.7
214

Annexes
Table A27c: Continued.

Biome St. d.
forest cropland 350 -7.6 10.5
forest grass 5,675 -6.4 16.3
bare soil sugarcane 500 -6.4 5.3
shrubs cropland 500 -4.6 6.2
shrubs sugarcane 1,075 -1.4 4.6
shrubs forest 1,300 -0.6 9.2
sugarcane cropland 3,775 -0.5 5.0
Atlantic Forest forest sugarcane 150 0.3 5.3
sugarcane forest 50 0.4 0.2
sugarcane bare soil 7,325 3.7 9.1
sugarcane grass 1,200 5.7 9.2
cropland grass 2,550 8.5 11.1
grass forest 725 11.1 10.9 &
cropland sugarcane 4,525 21.1 2.7
forest grass 350 -10.2 13.2
Pantanal forest cropland 6,700 -3.1 8.7
grass cropland 29,500 0.5 8.3
grass cropland 100 -5.2 1.4
sugarcane cropland 300 -0.1 2.9
Pampas grass forest 1,800 0.3 5.2
shrubs forest 1,500 1.9 5.1
bare soil grass 175 7.2 4.3
215

Annexes
Table A27d: Area (in km2) per land-use transition, provided as land use in 2030 for the reference
scenario and for the ethanol scenario, average change in species richness index (SRI) and standard
deviation (st. d.) of species richness index.
Land-use Land-use Area Average

Biome St. d.
reference 2030 ethanol 2030 (km2) ∆SRI
crop sugarcane 200 -31.3 39.7
forest crops 1,775 -14.1 15.7
shrubs crops 175 -5.2 11.7
grass crops 200 -2.8 11.4
Amazon crop shrubs 150 -2.1 5.8
forest shrubs 25 6.1 0.0
grass shrubs 350 6.1 19.5
crop grass 175 12.7 2.0
crop forest 1,100 16.8 13.8
bare soil crops 950 35.2 17.2
grass forest 1,550 47.7 23.0
grass shrubs 1,150 -18.7 24.9
crop bare soil 1,375 -7.9 14.6
grass crops 2,400 -4.6 15.6
bare soil sugarcane 3,175 -4.6 9.4
forest crops 25 -3.4 0.0
grass sugarcane 1,900 -2.0 3.5
Atlantic Forest shrubs forest 50 -0.9 1.1
crop sugarcane 6,200 1.9 4.6
sugarcane shrubs 100 2.6 1.2
crop shrubs 325 4.0 5.9
sugarcane crops 75 4.9 5.0
grass forest 700 5.8 12.1
crop forest 275 8.6 11.1
crop grass 275 20.5 24.5
shrubs grass 450 21.2 26.6
216

Annexes
Table A27d: Continued.

Biome St. d.
forest crops 825 -2.0 7.8
bare soil crops 100 -1.4 8.2
Caatinga shrubs crops 1,225 -0.4 6.8
crop forest 600 -0.2 8.8
crop bare soil 1,075 0.1 7.9
grass shrubs 125 5.8 7.2
crop sugarcane 25 9.8 0.0
forest crops 2,750 -6.9 10.1 &
crop bare soil 225 -6.6 8.7
grass shrubs 850 -6.3 7.6
shrubs crops 3,625 -3.9 8.4
grass crops 2,975 -2.8 9.6
sugarcane bare soil 50 -0.8 9.4
Cerrado
crop sugarcane 13,400 0.2 5.4
crop shrubs 2,550 0.8 6.5
sugarcane shrubs 25 2.1 0.0
crop forest 1,550 3.8 8.6
crop grass 375 4.1 8.3
sugarcane forest 25 51.1 0.0
217

Annexes
Table A27d: Continued.

Biome St. d.
bare soil crops 325 -5.3 8.7
crop forest 25 -2.5 0.0
Pampas crop sugarcane 25 0.5 0.0
shrubs forest 25 0.7 0.0
grass forest 125 2.3 4.9
crop sugarcane 25 -8.8 0.0
forest crops 1,425 -5.0 8.1
shrubs crops 2,300 -2.8 5.4
crop bare soil 50 0.4 5.6
Pantanal
crop grass 450 0.7 10.7
crop forest 375 1.0 8.2
grass crops 6,450 1.2 7.4
218

Annexes
A28. Species list for Brazil
We selected all mammal species whose potential habitat overlapped partly or entirely
with Brazil, this was the case for 610 species. Threatened mammal species were
defined as Critically Endangered, Endangered or Vulnerable in the IUCN Red List of
Threatened Species (IUCN, 2019) or the Brazilian National List of Threatened Fauna
Species (Lista Nacional Ofical de Espécies da Fauna Ameaçadas de Extinção (ICMBio,
2014)). Endemic species were defined as having 100% of potential habitat within
Brazil, in line with previous research (Jenkins et al., 2015). Restricted range species
were identified as species whose total area of potential habitat was smaller or equal
to 50,000 km2, in accordance with earlier studies (EKEN et al., 2004; Rodrigues et al.,
2004; Stattersfield, Crosby, Long, & Wege, 1998). All species included in the analysis
are listed in Table A28.
Table A28: Species name, reference model number, and their status as threatened, endemic or range-
restricted.
Model
Species name Threatened Endemic Range-restricted
number
16 Abrawayaomys ruschii x
699 Cuniculus paca
726 Akodon azarae &
730 Akodon cursor
738 Deltamys kempi x
742 Akodon lindberghi x x
750 Thaptomys nigrita
755 Akodon sanctipaulensis x x
756 Akodon serrensis
765 Necromys urichi
918 Alouatta ululata x x
922 Alouatta juara
1137 Ametrida centurio
1565 Anoura caudifer
1568 Anoura latidens
2055 Arctocephalus australis
2121 Artibeus amplus
2122 Artibeus anderseni
2124 Artibeus cinereus
2125 Artibeus concolor
2126 Artibeus fimbriatus
2128 Artibeus glaucus
2129 Artibeus gnomus
2130 Enchisthenes hartii
219

Annexes
Model
number
2137 Artibeus obscurus
2138 Artibeus phaeotis
2139 Artibeus planirostris
2276 Ateles belzebuth x
2282 Ateles marginatus x x
2283 Ateles paniscus x
2610 Bassaricyon beddardi
2802 Bibimys labiosus
2827 Blarinomys breviceps
2828 Blastocerus dichotomus x
2859 Necromys lasiurus
2993 Brachyteles arachnoides x x
2994 Brachyteles hypoxanthus x x
3036 Bradypus torquatus x x
3037 Bradypus tridactylus
3038 Bradypus variegatus
3414 Cabassous tatouay
3415 Cabassous unicinctus
3416 Cacajao calvus x x
3417 Cacajao melanocephalus
3549 Callicebus dubius
3555 Callicebus personatus x x
3564 Callimico goeldii x
3570 Callithrix aurita x x
3571 Callithrix flaviceps x x x
3572 Callithrix geoffroyi x
3575 Callithrix kuhlii x
3611 Calomys callosus
3612 Calomys hummelincki
3613 Calomys laucha
3617 Calomys tener
3648 Caluromys lanatus
3649 Caluromys philander
3651 Caluromysiops irrupta x
3903 Carollia brevicauda
3904 Carollia castanea
3905 Carollia perspicillata
3921 Carterodon sulcidens x
4064 Cavia aperea
4065 Cavia fulgida
220

Annexes
Model
number
4066 Cavia magna x
4074 Cebus xanthosternos x x
4112 Centronycteris maximiliani
4248 Cerdocyon thous
4366 Chaetomys subspinosus x x
4368 Chaetophractus vellerosus
4369 Chaetophractus villosus
4664 Chiroderma doriae
4666 Chiroderma salvini
4667 Chiroderma trinitatum
4668 Chiroderma villosum
4671 Chironectes minimus
4685 Chiropotes albinasus x
4772 Choeroniscus godmani
4774 Choeroniscus minor
4777 Choloepus didactylus
4778 Choloepus hoffmanni
4811 Chrotopterus auritus
4819 Chrysocyon brachyurus x
4988 Clyomys bishopi &
4989 Clyomys laticeps
5083 Coendou bicolor
5084 Coendou nycthemera
5085 Coendou prehensilis
5798 Ctenomys boliviensis x
5800 Ctenomys brasiliensis
5815 Ctenomys minutus x x
5820 Ctenomys perrensi x
5829 Ctenomys torquatus
6019 Cyclopes didactylus
6206 Cyttarops alecto
6220 Dactylomys boliviensis
6221 Dactylomys dactylinus
6277 Dasyprocta punctata
6278 Dasyprocta azarae
6280 Dasyprocta cristata
6281 Dasyprocta fuliginosa
6284 Dasyprocta leporina
6286 Dasyprocta prymnolopha
6288 Dasypus hybridus
221

Annexes
Model
number
6289 Dasypus kappleri
6293 Dasypus septemcinctus
6329 Delomys dorsalis
6330 Delomys sublineatus x
6520 Diaemus youngi
6561 Diclidurus albus
6562 Diclidurus ingens
6563 Diclidurus isabellus
6564 Diclidurus scutatus
6608 Dinomys branickii x
6924 Atelocynus microtis x
6926 Pseudalopex vetulus x x
6928 Pseudalopex gymnocercus
6977 Phyllomys blainvillii x x
6978 Phyllomys brasiliensis x x x
6979 Echimys chrysurus
6980 Phyllomys dasythrix x
6981 Toromys grandis
6982 Phyllomys lamarum x
6983 Makalata macrura
6984 Phyllomys nigrispinus x
6985 Callistomys pictus x x
6989 Phyllomys thomasi x x
6990 Phyllomys unicolor x x x
7916 Eptesicus brasiliensis
7922 Eptesicus diminutus
7928 Eptesicus fuscus
8241 Eumops auripendulus
8242 Eumops bonariensis
8243 Eumops dabbenei
8244 Eumops glaucinus
8245 Eumops hansae
8246 Eumops maurus
8306 Euphractus sexcinctus
8418 Euryzygomatomys spinosus
8771 Furipterus horrens x
8823 Galea flavidens x
8825 Galea spixii
9245 Glironia venusta
9273 Glossophaga commissarisi
222

Annexes
Model
number
9275 Glossophaga longirostris
9417 Gracilinanus agilis
9419 Gracilinanus emiliae
9421 Gracilinanus microtarsus
9422 Hyladelphys kalinowskii
9467 Graomys griseoflavus
10200 Histiotus alienus x x
10202 Histiotus montanus
10203 Histiotus velatus
10217 Holochilus brasiliensis
10219 Lundomys molitor x
10220 Holochilus sciureus
10300 Hydrochoerus hydrochaeris
10831 Inia geoffrensis x
10878 Isothrix bistriata
10879 Isothrix pagurus x
10957 Kannabateomys amblyonyx
10988 Kerodon rupestris x
11061 Kunsia tomentosus x
11170 Lagostomus maximus &
11175 Lagothrix lagotricha x
11349 Lasiurus ebenus x x
11351 Lasiurus egregius
11503 Leontopithecus caissara x x
11505 Leontopithecus chrysopygus x x x
11506 Leontopithecus rosalia x x
11509 Leopardus pardalis
11510 Leopardus tigrinus x
11966 Lichonycteris obscura
12078 Lionycteris spurrelli
12263 Lonchophylla bokermanni
12264 Lonchophylla dekeyseri x x x
12267 Lonchophylla mordax
12269 Lonchophylla thomasi
12270 Lonchorhina aurita x
12272 Lonchorhina marinkellei x
12273 Lonchorhina orinocensis x
12274 Lonchothrix emiliae x
12615 Macrophyllum macrophyllum
12814 Marmosa lepida
223

Annexes
Model
number
12822 Marmosops incanus
12824 Marmosops parvidens
13233 Makalata didelphoides
13234 Mesomys hispidus
13237 Mesomys stimulax
13240 Mesophylla macconnelli
13297 Micoureus constantiae
13375 Glyphonycteris behnii x
13376 Lampronycteris brachyotis
13377 Glyphonycteris daviesi
13378 Micronycteris hirsuta
13379 Micronycteris megalotis
13380 Micronycteris minuta
13381 Trinycteris nicefori
13382 Neonycteris pusilla x x
13383 Micronycteris schmidtorum
13384 Glyphonycteris sylvestris
13410 Microsciurus flaviventer
13559 Mimon bennettii
13560 Mimon crenulatum
13637 Cynomops abrasus
13639 Cynomops greenhalli
13640 Molossops mattogrossensis x
13641 Molossops neglectus
13642 Cynomops planirostris
13643 Molossops temminckii
13646 Molossus currentium
13647 Molossus coibensis
13649 Molossus pretiosus
13692 Monodelphis americana
13693 Monodelphis dimidiata
13694 Monodelphis emiliae
13695 Monodelphis iheringi
13696 Monodelphis kunsi
13697 Monodelphis maraxina x
13699 Monodelphis rubida x
13700 Monodelphis scalops
13701 Monodelphis sorex
13702 Monodelphis theresa x
13703 Monodelphis unistriata x
224

Annexes
Model
number
14025 Mustela africana
14085 Myocastor coypus
14100 Myoprocta acouchy
14140 Myotis albescens
14170 Myotis keaysi
14174 Myotis levis
14185 Myotis nigricans
14187 Myotis oxyotus
14195 Myotis riparius
14197 Myotis ruber
14204 Myotis simus
14224 Myrmecophaga tridactyla x
14386 Neacomys guianae
14388 Neacomys spinosus
14474 Nectomys rattus
14475 Nectomys squamipes
14742 Neusticomys oyapocki x
14829 Noctilio albiventris
15131 Oecomys bicolor
15132 Oecomys cleberi x x &
15133 Oecomys concolor
15135 Oecomys mamorae
15136 Oecomys paricola
15138 Oecomys rex
15139 Oecomys roberti
15140 Oecomys rutilus
15141 Oecomys speciosus
15142 Oecomys superans
15143 Oecomys trinitatis
15243 Oligoryzomys chacoensis
15244 Oligoryzomys delticola
15246 Oligoryzomys eliurus
15247 Oligoryzomys flavescens
15248 Oligoryzomys fulvescens
15252 Oligoryzomys microtis
15253 Oligoryzomys nigripes
15309 Leopardus colocolo x
15310 Leopardus geoffroyi x
15602 Euryoryzomys lamia x x x
15605 Euryoryzomys macconnelli
225

Annexes
Model
number
15607 Euryoryzomys nitidus
15608 Hylaeamys oniscus x x
15614 Cerradomys subflavus x
15617 Hylaeamys yunganus
15783 Oxymycterus angularis
15784 Oxymycterus delator x
15785 Oxymycterus hispidus
15787 Brucepattersonius iheringi
15788 Oxymycterus inca
15789 Oxymycterus nasutus x
15791 Oxymycterus roberti x x
15792 Oxymycterus rufus
15803 Ozotoceros bezoarticus x
16707 Peropteryx kappleri
16708 Peropteryx leucoptera
16709 Peropteryx macrotis
16799 Phaenomys ferrugineus x x
17168 Phylloderma stenops
17216 Phyllostomus discolor
17217 Phyllostomus elongatus
17218 Phyllostomus hastatus
17219 Phyllostomus latifolius
17565 Platyrrhinus lineatus
17566 Platyrrhinus aurarius
17567 Platyrrhinus brachycephalus
17570 Platyrrhinus helleri
17571 Platyrrhinus infuscus
17572 Platyrrhinus recifinus
17831 Podoxymys roraimae x x
17978 Pontoporia blainvillei x x
18144 Priodontes maximus x
18272 Trinomys albispinus x
18275 Proechimys brevicauda
18277 Proechimys guyannensis
18279 Proechimys cuvieri
18281 Trinomys dimidiatus x
18282 Proechimys goeldii
18287 Proechimys hoplomyoides
18288 Trinomys iheringi
18289 Proechimys longicaudatus
226

Annexes
Model
number
18292 Trinomys myosuros
18294 Proechimys roberti
18296 Proechimys quadruplicatus
18298 Trinomys setosus x x
18299 Proechimys simonsi
18300 Proechimys steerei
18340 Promops centralis
18341 Promops nasutus
18598 Pseudoryzomys simplex
18705 Pteronotus davyi
18706 Pteronotus gymnonotus
18708 Pteronotus parnellii
18709 Pteronotus personatus
18711 Pteronura brasiliensis x
18945 Pygoderma bilabiatum
19454 Rhagomys rufescens x
19592 Rhinophylla fischerae
19593 Rhinophylla pumilio
19610 Rhipidomys leucodactylus
19611 Rhipidomys macconnelli &
19612 Rhipidomys mastacalis
19613 Rhipidomys nitela
19618 Rhipidomys wetzeli x
19714 Rhynchonycteris naso
19804 Saccopteryx bilineata
19805 Saccopteryx canescens
19806 Saccopteryx gymnura
19807 Saccopteryx leptura
19839 Saimiri vanzolinii x x x
19945 Scapteromys tumidus
19997 Sciurillus pusillus
20003 Sciurus aestuans
20009 Sciurus gilvigularis
20012 Sciurus ignitus
20013 Sciurus igniventris
20018 Sciurus pucheranii
20022 Sciurus spadiceus
20033 Scleronycteris ega
20037 Scolomys ucayalensis x
20210 Sigmodon alstoni
227

Annexes
Model
number
20468 Speothos venaticus x
20599 Sphaeronycteris toxophyllum
20630 Sphiggurus spinosus
20631 Sphiggurus insidiosus
20634 Sphiggurus villosus
20954 Sturnira ludovici
20956 Sturnira magna
20960 Sturnira tildae
21350 Tamandua tetradactyla
21474 Tapirus terrestris x
21694 Thalpomys cerradensis x x
21695 Thalpomys lasiotis x x
21839 Thrichomys apereoides
21867 Thylamys macrurus x x
21877 Thyroptera discifera
21878 Thyroptera lavali x
21879 Thyroptera tricolor
21974 Tolypeutes matacus
21975 Tolypeutes tricinctus x
21983 Tonatia bidens
21984 Lophostoma brasiliense
21985 Lophostoma carrikeri
21987 Lophostoma schulzi
21988 Lophostoma silvicolum
22029 Trachops cirrhosus
22102 Trichechus inunguis x
22782 Uroderma bilobatum
22783 Uroderma magnirostrum
22837 Vampyressa bidens
22838 Vampyressa brocki
22841 Vampyressa pusilla
22842 Vampyrodes caraccioli
22843 Vampyrum spectrum
23076 Wiedomys pyrrhorhinos
23077 Wilfredomys oenax x x
23078 Juliomys pictipes
23321 Zygodontomys brevicauda
29402 Euryoryzomys emmonsae x
29403 Hylaeamys megacephalus
29404 Hylaeamys laticeps x
228

Annexes
Model
number
29405 Euryoryzomys russatus
29418 Oligoryzomys stramineus
29463 Proechimys echinothrix x
29464 Proechimys kulinae
29465 Proechimys pattoni
29466 Proechimys gardneri
29606 Histiotus humboldti
29607 Lasiurus atratus
29619 Mazama americana
29620 Mazama gouazoubira
29621 Mazama nana x
39910 Mico chrysoleucus x x
39911 Mico intermedius x
39912 Mico leucippe x x x
39913 Mico nigriceps x x
39914 Mico marcai x x
39916 Alouatta guariba x
39925 Lagothrix cana x
39927 Lagothrix poeppigii x
39929 Callicebus barbarabrownae x x x &
39930 Callicebus melanochir x x
39943 Callicebus nigrifrons
39945 Saguinus nigricollis
39947 Saguinus fuscicollis
39948 Saguinus imperator
39949 Cebus apella
39950 Cebus albifrons
39954 Callicebus coimbrai x x
39956 Chiropotes satanas x
39957 Alouatta belzebul x
39985 *
40019 Cebus kaapori x
40021 Cebus olivaceus
40027 Lonchorhina inusitata
40028 Micronycteris brosseti
40029 Micronycteris sanborni x x
40489 Didelphis albiventris
40500 Didelphis aurita
40501 Didelphis marsupialis
40503 Lutreolina crassicaudata
229

Annexes
Model
number
40505 Marmosa murina
40508 Marmosops noctivagus
40509 Metachirus nudicaudatus
40510 Micoureus demerarae
40511 Micoureus regina
40513 Monodelphis brevicaudata
40514 Monodelphis domestica
40515 Philander andersoni
40516 Philander opossum
40520 Thylamys velutinus x x
40642 Alouatta macconnelli
40643 Leontopithecus chrysomelas x x
40644 Saguinus bicolor x x x
41023 Mazama bororo x
41298 Sylvilagus brasiliensis
41518 Callithrix jacchus
41519 Callithrix penicillata x
41520 Mico argentatus x
41521 Mico humeralifer x
41523 Saguinus inustus
41524 Saguinus labiatus
41525 Saguinus midas
41526 Saguinus mystax
41527 Cormura brevirostris
41530 Tonatia saurophila
41535 Cebuella pygmaea
41536 Saimiri boliviensis
41537 Saimiri sciureus
41538 Saimiri ustus
41539 Aotus azarae
41540 Aotus nancymaae
41542 Aotus nigriceps
41543 Aotus trivirgatus
41544 Aotus vociferans
41545 Alouatta caraya
41546 Alouatta sara
41547 Ateles chamek x
41548 Callicebus donacophilus
41549 Callicebus pallescens
41551 Callicebus cupreus
230

Annexes
Model
number
41552 Callicebus caligatus x
41555 Callicebus stephennashi x x
41556 Callicebus moloch x
41557 Callicebus cinerascens x
41558 Callicebus brunneus
41559 Callicebus hoffmannsi x
41560 Callicebus baptista x x
41561 Callicebus bernhardi x
41562 Callicebus torquatus
41563 Callicebus lugens
41564 Callicebus lucifer
41565 Callicebus purinus x
41566 Callicebus regulus
41567 Pithecia albicans x x
41568 Pithecia irrorata
41569 Pithecia pithecia
41580 Mico acariensis x x
41582 Mico manicorensis x x
41583 Mico mauesi x x
41584 Callibella humilis x x x &
41630 Conepatus chinga
41633 Conepatus semistriatus
41639 Galictis cuja
41640 Galictis vittata
41665 Otaria byronia
41678 Bassaricyon alleni
41679 Potos flavus
41684 Nasua nasua
41685 Procyon cancrivorus
41778 Tayassu pecari x
42691 Mico emiliae x
42692 Mico saterei x x
42694 Saguinus niger x
42695 Saguinus martinsi x x
42696 Cebus macrocephalus
42697 Cebus robustus x x
43891 Chiropotes chiropotes
43892 Chiropotes utahickae x x
43912 Alouatta discolor x x
43929 Alouatta seniculus
231

Annexes
Model
number
135161 Sooretamys angouya
135429 Saguinus melanoleucus
136195 Rhipidomys emiliae
136197 Akodon montensis
136198 Brucepattersonius paradisus x
136202 Philander mondolfii
136205 Oxymycterus amazonicus x
136220 Rhogeessa hussoni
136221 Trinomys gratiosus x x
136222 Kerodon acrobata x x x
136236 Neusticomys ferreirai x x
136241 Monodelphis umbristriata x x
136245 Thrichomys pachyurus
136253 Cebus flavius x x
136260 Akodon mystax x x x
136274 Phyllomys mantiqueirensis x x x
136276 Cerradomys maracajuensis
136278 Marmosops paulensis x x
136283 Hylaeamys acritus
136284 Mesomys occultus x x
136294 Mico melanurus
136296 Marmosops bishopi
136305 Trinomys mirapitanga x x x
136307 Monodelphis glirina
136314 Rhogeessa io
136321 Xeronycteris vieirai x x
136332 Alouatta nigerrima x
136336 Oligoryzomys moojeni x
136338 Delomys collinus x x
136346 Cebus libidinosus
136350 Centronycteris centralis
136352 Neacomys paracou
136355 Thrichomys inermis x
136357 Rhipidomys cariri x x x
136358 Oecomys catherinae
136366 Cebus cay x
136368 Cerradomys scotti
136371 Brucepattersonius griserufescens x x
136375 Philander frenatus
136400 Phyllomys lundi x x x
232

Annexes
Model
number
136407 Trinomys eliasi x x
136414 Trinomys yonenagae x x x
136419 Cacajao ayresi x x x
136422 Rhipidomys macrurus
136424 Micronycteris microtis
136425 Oligoryzomys rupestris x x x
136436 Juscelinomys guaporensis x
136442 Brucepattersonius soricinus x
136448 Natalus espiritosantensis x
136457 Brucepattersonius misionensis x
136464 Ctenomys flamarioni x x
136468 Microakodontomys transitorius x x x
136483 Akodon paranaensis
136492 Trinomys paratus x x
136494 Sturnira oporaphilum
136501 Philander mcilhennyi
136511 Cerradomys marinhus x x
136518 Sphiggurus roosmalenorum
136529 Hylaeamys perenensis
136543 Trinomys moojeni x x x &
136545 Cryptonanus agricolai x
136563 Juliomys rimofrons x x x
136567 Ctenomys lami x x x
136570 Neacomys minutus
136573 Marmosops pinheiroi
136583 Reithrodon typicus
136592 Didelphis imperfecta
136594 Thyroptera devivoi
136605 Cynomops paranus
136613 Oxymycterus quaestor
136624 Oecomys auyantepui
136640 Cacajao hosomi x
136652 Echimys vieirai x
136653 Thylamys karimii x x
136655 Neacomys musseri
136663 Myoprocta pratti
136671 Vampyressa thyone
136672 Brucepattersonius igniventris x x
136679 Calomys tocantinsi x
136682 Phyllomys kerri x
233

Annexes
Model
number
136689 Calomys expulsus x
136693 Akodon reigi x
136704 Phyllomys medius
136705 Cryptonanus guahybae x x
136708 Mazama nemorivaga
136717 Cebus nigritus
136725 Oxymycterus caparoae x x
136727 Neacomys dubosti
136728 Oligoryzomys fornesi
136738 Sphiggurus melanurus
136740 Pecari maximus
136745 Wiedomys cerradensis x x
136751 Isothrix negrensis x
136756 Nectomys apicalis
136759 Rhipidomys gardneri
136787 Alouatta puruensis
136794 * x x
136801 Phyllomys pattoni x
136804 Mico rondoni x
136809 Eumops trumbulli
136813 Oxymycterus dasytrichus
136825 Eumops patagonicus
136830 Marmosops neblina x
136842 Marmosops ocellatus
136844 Micoureus paraguayanus
136845 Cryptonanus chacoensis
* Species codes (model number) in Table A28 link to species names on the IUCN Red List
website (IUCN, 2019). For two species, species names were no longer accessible on the
website. This could be due to the fact that phylogenetic classification of the species changed;
they were for example split or merged, and their species code changed as a result. For these
two species therefore, we do not have species names but have used their range data in our
analysis.
234

Annexes
A29. Changes in species richness index in Brazil between 2012

and 2030 in the reference scenario
We compared species richness index in 2012 and in 2030 between for the reference
scenario to assess the species richness index impact of land-use change due to drivers
other than increased ethanol demand. Here we show the difference for threatened
species (A29a), endemic species (A29b) and range-restricted species (A29c).
AM AM
CA CA
CE CE
∆ threatened SRI ∆ endemic SRI
2030 reference - 2012
PN 2030 reference - 2012 PN
High: 9 High: 7
Low: 0 AF Low: 0 AF
0 0
High: 0
Low: -9
High: 0
Low: -6 &
PM
a) b) PM
AM
CA
CE
∆ range-restricted SRI
2030 reference - 2012 PN
High: 6
Low: 0 AF
High: 0
Low: -7
PM
c)
Figure A29. Difference in species richness index in 2012 and 2030 for the reference scenario for a)
threatened mammal species, b) endemic mammal species, and c) range-restricted mammal species.
Black lines show biome boundaries, abbreviations show biome names; AM = Amazon, CA = Caatinga,
CE = Cerrado, PN = Pantanal, AF = Atlantic Forest, PM = Pampas.
235

Annexes
A30. Difference in species richness index in Brazil in 2030

between the reference and the ethanol scenario
We compared species richness index in 2030 between the ethanol and the reference
scenario to assess the difference in species richness index due to increased ethanol
demand. Here we show the difference for threatened species (A30a), endemic species
(A30b) and range-restricted species (A30c).
AM AM
CA CA
CE CE
∆ threatened SRI ∆ endemic SRI
2030 reference - ethanol
PN 2030 reference - ethanol PN
High: 8 High: 4
Low: 0 AF Low: 0 AF
0 0
High: 0 High: 0
Low: -8 Low: -5
PM PM
a) b)
AM
CA
CE
∆ range-restricted SRI
2030 reference - ethanol PN
High: 6
Low: 0 AF
High: 0
Low: -5
PM
c)
Figure A30. Difference in species richness index in 2030 between the ethanol and the reference
scenario for a) threatened mammal species, b) endemic mammal species, and c) range-restrict-
ed mammal species. Black lines show biome boundaries, abbreviations show biome names;
236

Annexes
A31. Intensity of changes in species richness index due to direct

and indirect land-use change in Brazil
Following Verstegen et al. (Verstegen, van der Hilst, 2016), direct land-use change
(dLUC) between the ethanol and the reference scenario in 2030 was defined as a
change from rangeland, planted forest, crops, planted pasture, natural forest, grass
& shrubs, bare soil or abandoned land to sugarcane. Indirect land-use change (iLUC)
was defined as a change from natural forest, grass & shrubs, bare soil or abandoned
land to planted pasture, crops, rangeland or planted forest. Figure A31 provides a map
with the locations of dLUC and iLUC. Table A31 provides the area of direct and indirect
land use change per impact intensity category and per biome.
AM
CA &
CE
PN
Land-use changes AF
Reference-ethanol scenario
no LUC
dLUC
iLUC
PM
Figure A31. Location of direct (dLUC) and indirect (iLUC) land use changes between the reference
and ethanol scenario in 2030. Black lines show biome boundaries, abbreviations show biome names;
237

Table A31: Area (in km2) of changes in total SRI in different regions and different impact categories caused by direct (dLUC) and by indirect (iLUC) land-use
238
change.
Annexes
0-25% 25-50% 50-100% >100% 0--25% -25--50% -50--100% >-100%
Region
dLUC iLUC dLUC iLUC dLUC iLUC dLUC iLUC dLUC iLUC dLUC iLUC dLUC iLUC dLUC iLUC
Amazon
Caatinga
25 0 50 800 50 800 25 850 25 350 0 125 25 0 0 850
Cerrado
75 0 75 0 0 325 50 1,150 0 450 0 300 0 125 50 25
Atlantic Forest
100 0 725 325 2,800 1,850 5,175 2,400 4,300 1,675 2,175 525 2,350 350 375 50
Pantanal
375 125 1,025 125 2,925 325 6,500 225 3,150 375 1,625 250 1,325 400 200 175
Pampas
0 0 0 25 0 625 25 2,200 0 725 0 200 0 100 0 0
0 0 25 175 0 50 0 100 25 25 0 100 0 200 0 0
16-03-20 12:28
Annexes
A32. Study area of hydrological impacts in Brazil
The study area includes all of Brazil, as well as some areas outside of Brazil, to the west
and south, consisting of parts of Uruguay, Argentina, Paraguay, Bolivia, Peru, Ecuador,
Colombia, Venezuela, Guyana, Suriname and French Guyana. The study area boundaries
were created by identifying all watersheds that are located entirely or partly in Brazil.
For those watersheds that are located only partly in Brazil, we included the area of
the watershed outside of Brazil in the study area. This is shown in Figure A32 below.
The study area is divided in a number of regions as in previous studies (van der Hilst,
Verstegen, 2018; Verstegen, van der Hilst, 2016), based on geographical boundaries. The
size of the land use outside of Brazil that was included in the study was approximately
30% of the land area of Brazil. The study area, regional boundaries and states are
shown in Figure A32 below.
&
Figure A32: Map of the study area, with abbreviations for region names and colours showing
the different states. OB = outside of Brazil, NA = Northern Amazon, CWC = Centre West Cerrado,
NER = North East Cerrado, NEC = North East Coast, SE = South East, S = South.
239

Annexes
A33. Validation of the PCR-GLOBWB model in Brazil
Performance of the PCR-GLOBWB model in simulating river discharge has been

validated in earlier studies (Sutanudjaja, 2018; van Beek, 2011). Because the model
was adapted in this study to include new land cover categories and parameterisation,
the validation was repeated for the adapted version of the model. This was done by
comparing discharge modelled in PCR-GLOBWB to discharge measured in measuring
stations and recorded in the Global Runoff Database Centre (Global Runoff Data Centre
(GRDC), 2009) for the period 1958-2005. We compared simulated and measured river
discharge in the study area by calculating the Kling-Gupta Efficiency (KGE) coefficient
(Gupta, 2009). This coefficient is a measure of bias and differences in amplitude
and timing between the measured and simulated datasets and can range from 1 to
minus infinity. A KGE value of 1 means a perfect agreement of observed GRDC data
and simulated data. We included discharge data from 490 measuring stations in the
study area. In our data set, KGE values range from close to 1 to -130, with 92% of
stations having values between 1 and -1. Values less than -1 are found mostly for
measuring stations in the upstream Amazon area. Our results are consistent with
earlier validation of the PCR-GLOBWB model, and show that PCR-GLOBWB is able to
model discharge accurately in most areas but is less accurate in modelling discharge
in upstream areas with small discharge.
Discharge and KGE

<-1
-1 - -0.5
-0.5 – 0
0 – 0.5
0.5 – 1
Rivers
Figure A33: Kling-Gupta Efficiency coefficients (KGE, in colours) and discharge (in circle size) based
on comparison of PCR-GLOBWB model output and discharge measurements from the GRDC database.
240

Annexes
A34. Land use outside of Brazil for the assessment of hydrological

impacts
The method of Verstegen et al. (Verstegen, van der Hilst, 2016) was applied to determine
land use within the study area but outside of Brazil, consisting of areas within
catchments of which a part lies in Brazil. In line with the input data for the land-use
projections within Brazil, the GlobCover v2.2 land-use map (Bicheron, 2008) (300m
resolution) was used as input for the area and location of land use outside of Brazil.
This was done using aggregated density (5 km resolution) per land use as suitability
maps, and using the same allocation order as was used in the PCRaster Land Use
Change (PLUC) model runs within Brazil. Globcover land-use types were translated
into the land-use types used in PLUC as described in Table A34. For the calculation of
total area per land-use type, a proportion of the composite land-use types of Globcover
2005 was assigned to each of the land-use types used in this study. For example, the
GlobCover category ‘Mosaic vegetation (grassland/shrubland/forest) (50-70%) /
cropland (20-50%)’ was considered to consist of 30% natural forest, 30% grass &
shrub land and 40% crops. For the determination of land suitability, composite land-
use types were considered to be suitable for all included categories. For example, cell
assigned the GlobCover category ‘Mosaic grassland (50-70%) / forest or shrubland (20-
50%)’ were considered to be suitable for grass & shrubs, as well as for natural forest. &
241

Table A34: Reclassification of GlobCover 2005 land-use types into categories used in this study. Numbers in the table signify the fraction of the GlobCover
242
area assigned to the category used in this study.
Annexes
Urban
Water
Natural
forest
Rangeland
Planted
forest
Crops
Grass &
shrubs
Sugarcane
Pasture
Bare soil
Rainfed croplands 0 0 0 0 0 1 0 0 0 0
Mosaic cropland (50-70%) / vegetation (grassland/shrubland/forest) (20-50%) 0 0 0.2 0 0 0.6 0.2 0 0 0
Mosaic vegetation (grassland/shrubland/forest) (50-70%) / cropland (20-50%) 0 0 0.3 0 0 0.4 0.3 0 0 0
Closed to open (>15%) broadleaved evergreen or semi-deciduous forest (>5m) 0 0 1 0 0 0 0 0 0 0
Closed (>40%) broadleaved deciduous forest (>5m) 0 0 1 0 0 0 0 0 0 0
Open (15-40%) broadleaved deciduous forest/woodland (>5m) 0 0 1 0 0 0 0 0 0 0
Mosaic forest or shrubland (50-70%) / grassland (20-50%) 0 0 0.3 0 0 0 0.7 0 0 0
Mosaic grassland (50-70%) / forest or shrubland (20-50%) 0 0 0.2 0 0 0 0.8 0 0 0
Closed to open (>15%) (broadleaved or needleleaved, evergreen or deciduous)
0 0 0 0 0 0 1 0 0 0
shrubland (<5m)
Closed to open (>15%) herbaceous vegetation (grassland, savannas or lichens/mosses) 0 0 0 0 0 0 1 0 0 0
Sparse (<15%) vegetation 0 0 0 0 0 0 0 0 0 1
Closed to open (>15%) broadleaved forest regularly flooded (semi-permanently or
0 0 1 0 0 0 0 0 0 0
temporarily) - Fresh or brackish water
Closed (>40%) broadleaved forest or shrubland permanently flooded - Saline or
0 0 0.5 0 0 0 0.5 0 0 0
brackish water
Closed to open (>15%) grassland or woody vegetation on regularly flooded or
0 0 0 0 0 0 1 0 0 0
waterlogged soil - Fresh, brackish or saline water
Artificial surfaces and associated areas (Urban areas >50%) 1 0 0 0 0 0 0 0 0 0
Bare areas 0 0 0 0 0 0 0 0 0 1
Water bodies 0 1 0 0 0 0 0 0 0 0
Permanent snow and ice 0 0 0 0 0 0 0 0 0 1
16-03-20 12:28
Annexes
A35. Translation of PLUC land-use types into PCR-GLOBWB land-

use types for Brazil
The land-use types used in the land-use projections created in previous studies (van
der Hilst, Verstegen, 2018; Verstegen, van der Hilst, 2016) using the PCRaster Land-
Use Change (PLUC) model were translated into the land-use types in PCR-GLOBWB
as described in Table A35. The land-use types cropland and sugarcane were divided
into rainfed cropland, irrigated non-paddy cropland, irrigated paddy cropland, rainfed
sugarcane and irrigated sugarcane. This was done based on 1) the relative suitability
of the designated rainfed and irrigated crops according to the Global Agro-Ecological
Zones (GAEZ) (Fischer, 2012), and 2) the proximity of irrigated crops of the same
type according to the Monthly Irrigated and Rainfed Crop Area (MIRCA2000)
dataset (Portmann, 2010). Water was included as a land-use type in the PLUC land-
use projections but not in PCR-GLOBWB, because in PCR-GLOBWB the location of
waterbodies is determined internally in the model and water is not considered a land
use.
&
243

Table A35: Reclassification of PLUC land-use types to PCR-GLOBWB land-use types
244
PCR-GLOBWB
Annexes
Grass and Rainfed Irrigated non- Irrigated paddy Rainfed Irrigated
category → Urban Forest Pasture
shrubland cropland paddy cropland cropland sugarcane sugarcane
PLUC category ↓
x
x
Urban
x
Natural forest
x
Planted forest
x
Grass & shrubs
x
Rangeland
x x x
Planted pasture
x x
Cropland*
x
Sugarcane*
x
Bare soil
Abandoned
Water**
* Distinction between rainfed and irrigated (paddy and non-paddy) cropland and sugarcane was made based on MIRCA2000 data
** Water was not reclassified into a PCR-GLOBWB land-use type, the location of waterbodies was determined in the PCR-GLOBWB model.
16-03-20 12:28
Annexes
A36. Land cover parameterisation in PCR-GLOBWB in Brazil:

natural vegetation, cropland and sugarcane
Hydrological parameters in PCR-GLOBWB for agricultural land are crop type

specific, and depend on whether the land is irrigated or not. For natural vegetation,
hydrological parameters are vegetation type specific. For each cell, the crop or
vegetation type was established at resolution that is finer than the model resolution
(respectively 30 arc seconds and 5 arc minutes). The parameterization consists of 2
steps; first crop or vegetation types are allocated at 30 arc second resolution, then the
corresponding hydrological parameters are determined and assigned to the cell. The
crop and vegetation types are then grouped into distinct classes that are assigned the
aggregated properties of the contributing crop and vegetation types. The presence
of these land cover classes is then quantified in terms of their relative abundance in
each cell to account for the sub-grid variability at the model resolution following a
methodology used in previous studies (Bosmans, 2017; Dermody, 2014; Van Beek, 2009).
This method can be combined with reconstructions or projections of land cover types
to evaluate the effects of land cover change on the hydrology.
Data on the presence of crop types at 5 arc minutes globally was taken from the
Monthly Irrigated and Rainfed Crop Area (MIRCA2000) dataset (Portmann, 2010). &
To downscale this spatially to 30 arc seconds, suitability data from the Global Agro-
Ecological Zones (GAEZ) (Fischer, 2012) is used. This suitability is provided at 5 arc
minutes and includes information on climate, soil, slope etc. and is crop specific
and provides the constraint on the eventual global distribution of a crop type. To
disaggregate this to 30 arc seconds, a function of the suitability is multiplied with a
function of proximity and slope gradient; the presence of a crop becomes more likely
in the flatter areas and in the heart of the presently cultivated areas and decreases
outwards, becoming more unlikely in steeper and more remote areas. This combined
information on suitability and proximity defines the likelihood that a crop type is
present within a cell at the model resolution of 5 arc minutes. Thus, all the crop types
can be assigned iteratively to the most suitable cell at 30 arc seconds until the desired
fractional presence at 5 arc minutes is met. Consequently, the presence of a land cover
type will meet the required fraction and its composition will reflect the proximity and
suitability of the crop types that are currently cultivated.
245

Annexes
The ten most abundant rainfed crops in Brazil10, with the exception of sugarcane,
were grouped as rainfed crops, all irrigated crops with the exception of wet rice and
sugarcane were assigned to irrigated non-paddy crops, wet rice to irrigated paddy,
and sugar cane to respectively rainfed sugar cane and irrigated sugar cane, bringing
the total to five agricultural land cover types (see Annex 3). For the distribution of
the non-agricultural land cover classes a similar approach was followed but here the
vegetation classes were based on the GLCC land cover classification (Loveland, 2000).
For the natural vegetation, suitability was based on the distribution of biomes in the
Holdridge life zones (Holdridge, 1967) and proximity. For the land cover types urban
and pasture, only proximity was used. For all land cover classes, information from
PLUC was used to determine their abundance over the period 2012-2030 based on
land-use projections from previous studies (van der Hilst, Verstegen, 2018; Verstegen,
For the parameterization of the crop types, information from the GWCM (Siebert,
2008) was used which is compatible with the MIRCA dataset. To determine the
characteristic phenology of a crop, the crop calendar of the MIRCA dataset was used.
For the parameterization of sugarcane we applied the same methodology, but applied
a user defined Brazil-specific crop calendar (United States Department of Agriculture,
1994) and LAI (Pereira, 2013). For the natural vegetation, we used the parameterization
of Hagemann et al. (Hagemann, 1999) that is linked to the Olson classification of the
GLCC. For the natural vegetation, a climatology of the relative leaf area index was used
that reflects the availability of light, temperature and water over the growing season
(Van Beek, 2009) that can be combined with the LAI bounds per vegetation types as
specified by Hageman et al. (1999). This LAI information provides the basis to derive
the temporal varying parameters as the crop coefficient (Allen, 1998), cover fraction
and interception capacity used by PCR-GLOBWB (see (Van Beek, 2009) for details).
Eventually this leads to a distribution of crop and vegetation types at 30 arc seconds
to which specific parameters such as the crop coefficient are assigned to the cell
(Table A36). In this case, we kept this parameter constant (at the value for 2012) for
expediency. As a result, the land cover parameterization of PCR-GLOBWB is spatially
variable and the parameters per land cover type may vary to reflect the actual
composition in terms of crop or vegetation types. Moreover, the parameterization
is temporally variable, with the parameters varying within the year to reflect the
cropping season or growing conditions. Over the years, the relative abundance of the
10 The following crops were included; wheat, maize, rice, sorghum, soybeans, cassava, pulses, citrus,
cotton and coffee. Together, these crops cover 90% of all non-sugarcane cropland area (FAO, 2019).
246

Annexes
land cover types may vary within a cell to follow the trends and variations in land
cover defined by the land-use projections (van der Hilst, Verstegen, 2018; Verstegen, van
der Hilst, 2016). As a result, projected land-use changes may have different impacts
in different regions. For example, crop expansion over forests may have a different
impact in the Amazon region than in the Atlantic forest region, because both the forest
type and the agricultural crops may differ in both regions, as defined by the crop and
vegetation specific parameters taken from the MIRCA2000 and GLCC databases, and
subsequent differences in hydrological parameters. Furthermore, soil properties and
climatic conditions may differ between the different regions, and thereby influence
the modelling of the flow of water in PCR-GLOBWB.
Table A36: Hydrological parameters used in PCR-GLOBWB with description and unit. M = meter.
Parameter Description Unit

The ratio of the crop specific potential
Crop coefficient evapotranspiration to that of a reference crop dimensionless
under ideal conditions
Ground cover Area of the ground surface covered by vegetation m2/m2
The maximum amount of precipitation that
Interception capacity does not reach the soil, but is intercepted by the m/m2
vegetation and stored on the canopy
Leaf area index

The one-sided green leaf area per unit of ground
surface area
m2/m2 &
The fraction of a PCR-GLOBWB cell that has a
Vegetation fraction fraction
certain land cover type
Schedule of the crop and location specific growing Start month, end
Crop calendar
season month
Ratio of the maximum root depth over the average,
Maximum root depth used to parameterize the water storage capacity of
fraction
fraction the soil for the Improved Arno Scheme (Hagemann,
2003)
Ratio of the minimum root depth over the average,
Minimum root depth used to parameterize the water storage capacity of
fraction
fraction the soil for the Improved Arno Scheme (Hagemann,
2003)
Fractional abundance of the total root content per
Root fraction (layer
soil layer assuming an exponential distribution fraction
1 & 2)
over the root depth.
Depth beyond no significant roots are present
Total root depth m
(Canadell, 1996)
247

Annexes
A37. Seasonality of soil moisture and discharge in Brazil
We assessed the historical seasonality of soil moisture and discharge by calculating

average soil moisture and discharge for each month for the years 1958 to 2005. We
then averaged these values over the regions (Figure A37a).
Averagesoil
Average soilmoisture
moisture 1958-2005
1958-2005 Average discharge 1958-2005
0,45
0,45 2500
2500
0,4
0,4
0,35 2000
2000
0,35
0,3
0,3
1500
1500
0,25
0,25
0,2
0,2 1000
1000
0,15
0,15
0,1
0,1 500
500
0,05
0,05
00 00
Southeast
Southeast NortheastCoast
Northeast Coast Southeast
Southeast Northeast Coast
Northeast Coast
NortheastCerrado
Northeast Cerrado Northern Amazon
Northern Amazon Northeast Cerrado
Northeast Cerrado Northern Amazon
Northern Amazon
CentreWest
Centre WestCerrado
Cerrado South
South Centre West
Centre West Cerrado
Cerrado South
South
OutsideBrazil
Outside Brazil Outside Brazil
Outside Brazil
Figure A37a: Average soil moisture in m3/m3 (left) and average discharge in m3/s (right) per month
between 1958 and 2005 per region.
We subtracted timeseries from 2012 to 2030 of drought indicator ‘area in drought’

for the ethanol scenario from the reference situation to assess the difference in
drought due to increased sugarcane production. Figure A37b shows the time series
of the difference in agricultural area in drought (in %) between the scenarios from
2012 to 2030, averaged per region, while Figure A37c shows the difference in area in
hydrological drought.
248

Annexes
∆ ethanol - reference scenario area in agricultural drought

1,2
∆ area in agricultural drought (%)
0,8
0,6
0,4
0,2
-0,2
-0,4
-0,6
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Southeast Northeast Coast Northeast Cerrado Northern Amazon
Centre West Cerrado South Outside Brazil
Figure A37b: Time series of the difference in area in agricultural drought between the ethanol
and the reference scenario (in %) for the period 2012-2030. Area in drought has been averaged
per region.
∆ ethanol - reference scenario area in hydrological drought

1
∆ area in hydrological drought (%)
0,8
&
0,6
0,4
0,2
-0,2
-0,4
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
Southeast Northeast Coast Northeast Cerrado Northern Amazon

Centre West Cerrado South Outside Brazil
Figure A37c: Time series of the difference in area in hydrologicall drought between the ethanol
and the reference scenario (in %) for the period 2012-2030. Area in drought has been averaged
per region.
249

Annexes
A38. Regional drought area, duration and deficit in Brazil
We quantified drought events between 2012 and 2030 using the indicators area in
drought, drought duration and drought deficit. These indicators were averaged over
the years and then per region to assess regional differences. Regional averages of
agricultural drought, as well as the differences in agricultural drought between the
ethanol scenario and the reference, are provided in Table A38a. Regional averages of
hydrological drought, as well as the differences in hydrological drought between the
ethanol scenario and the reference, are provided in Table A38b.
250

Table A38a: Area in agricultural drought (% of total area of the region), and agricultural drought duration (% of the year) and deficit (dimensionless) averaged
for the period 2012-2030 and then per region, for the reference (ref) and the ethanol scenario (eth). Columns on the right shown the difference between the
ethanol scenario and the reference (%) for the area, duration and deficit of agricultural drought. A negative difference means that drought area, duration
or deficit is higher in the reference than in the ethanol scenario, a positive difference means that drought duration or deficit is higher in the ethanol scenario.
Area in drought (% Drought duration Drought deficit Area in Drought Drought

Agricultural drought
of region) (% of year) (dimensionless) drought duration deficit
Region ref eth ref eth ref eth % difference eth scenario - ref
Southeast 22.29 22.29 19.39 19.40 39.36 39.01 0.01 0.04 -0.89
Northeast Coast 10.57 10.62 12.69 12.65 58.39 54.08 -0.40 -0.34 -7.38
Northeast Cerrado 14.34 14.33 13.55 13.55 29.66 29.60 0.04 0.03 -0.19
Northern Amazon 21.75 21.73 16.80 16.81 44.86 45.44 0.08 0.09 1.28
Centre West Cerrado 28.84 28.40 24.27 24.73 109.64 117.08 1.54 1.92 6.78
South 12.75 12.74 16.24 16.25 45.05 44.93 0.12 0.04 -0.27
Outside Brazil 19.79 19.79 16.66 16.66 33.13 33.12 -0.01 -0.01 -0.01
Table A38b: Area in hydrological drought (% of total area of the region), and hydrological drought duration (% of the year) and deficit (dimensionless) averaged
for the period 2012-2030 and then per region, for the reference (ref) and the ethanol scenario (eth). Columns on the right shown the difference between the
ethanol scenario and the reference (%) for the area, duration and deficit of hydrological drought. A negative difference means that drought area, duration
or deficit is higher in the reference than in the ethanol scenario, a positive difference means that drought duration or deficit is higher in the ethanol scenario.
Area in drought Drought duration Drought deficit Area in Drought Drought

Hydrological drought
(% of region) (% of the year) (dimensionless) drought duration deficit
Region ref eth ref eth ref eth % difference eth scenario - ref
Southeast 10.97 11.88 10.78 10.84 12.04 12.16 0.81 0.59 0.97
Northeast Coast 6.46 6.49 19.03 18.99 52.34 51.87 -0.47 -0.19 -0.91
Northeast Cerrado 11.71 11.72 11.40 11.40 19.98 20.06 -0.05 -0.01 0.41
Northern Amazon 26.08 26.08 14.78 14.78 23.89 23.89 -0.01 0.00 0.00
251
Annexes
Centre West Cerrado 19.33 19.17 12.66 12.78 23.47 23.71 0.86 0.91 1.03
South 4.27 4.26 6.67 6.67 4.81 4.80 0.26 -0.04 -0.13
Outside Brazil 14.82 14.82 11.28 11.28 15.53 15.53 0.00 0.00 -0.01
&
16-03-20 12:28
Annexes
A39. Sugarcane irrigation in Brazil
We determined the total area of sugarcane, as well as sugarcane area separated by

rainfed and irrigated cultivation for 2012, and 2030 for the ethanol scenario and
the reference per region (Table A39b). We used this to calculate the percentage of
sugarcane area that is irrigated and the percentage of sugarcane that is rainfed.
A map of average monthly gross water demand for irrigated sugarcane in 2030 was
used to determine average monthly water demand per region for both the ethanol
scenario and the reference (Table A39a). We also determined the difference between
the ethanol scenario and the reference.
Table A39a: Average monthly gross water demand for irrigation (m3/m) in 2030 per region for
irrigated non-paddy cropland, irrigated paddy cropland and irrigated sugarcane for the reference
(ref) and ethanol (eth) scenarios, and the difference between the ethanol and reference scenarios
(Δeth-ref).
Region Irrigated sugarcane (m3/m)

ref eth Δeth-ref
Southeast 1.02 1.87 0.85
Northeast Coast 0.44 0.49 0.06
Northeast Cerrado 0.00 0.02 0.02
Northern Amazon 0.00 0.00 0.00
Centre West Cerrado 2.53 3.48 0.95
South 0.00 0.00 0.00
Total 3.98 5.86 1.88
252

Table A39b: Sugarcane area (km2) and proportion of sugarcane that is irrigated and rainfed (%) for the year 2012 and 2030 for the reference and ethanol
scenario per region, separated for irrigated and rainfed sugarcane.
Irrigated Rainfed Irrigated Rainfed

Scenario Region Total (km2)
sugarcane (km2) sugarcane (km2) sugarcane (%) sugarcane (%)
Southeast
Northeast Coast
3,751 61,957 65,708 5.71 94.29
Northeast Cerrado
1,520 8,461 9,982 15.23 84.77
2012 Northern Amazon

89 1,127 1,216 7.31 92.69
Centre West Cerrado

0 130 130 0.00 100.00
South
3,404 12,577 15,981 21.30 78.70
Total
0 6,672 6,672 0.00 100.00
Southeast
8,764 90,924 99,689 8.79 91.21
Northeast Coast
3,470 58,234 61,704 5.62 94.38
Northeast Cerrado
1,163 3,683 4,845 24.00 76.00
2030 reference
Northern Amazon
scenario
0 285 285 0.00 100.00
Centre West Cerrado

0 72 72 0.00 100.00
South
6,758 25,732 32,490 20.80 79.20
Total
0 1,723 1,723 0.00 100.00
Southeast
11,391 89,729 101,119 11.26 88.74
Northeast Coast
5,728 69,333 75,061 7.63 92.37
Northeast Cerrado
1,227 4,870 6,096 20.12 79.88
2030 ethanol
Northern Amazon
scenario
0 564 564 0.00 100.00
Centre West Cerrado

0 104 104 0.00 100.00
South
11,071 39,606 50,677 21.85 78.15
Total
0 2,811 2,811 0.00 100.00
18,026 117,288 135,313 13.32 86.68
253
Annexes
&
16-03-20 12:28
References
References
Abt, K. L., Abt, R. C., & Galik, C. S. (2012). “Effect of Bioenergy Demands and Supply Response
on Markets, Carbon and Land Use.” Forest Science, 58(5), 523–539.
Abt, K. L., Abt, R. C., Galik, C. S., & Skog, K. E. (2014). Effect of Policies on Pellet Production
and Forests in the U.S. South: A Technical Document Supporting the Forest Service
Update of the 2010 RPA Assessment. General Technical Report SRS-202.
Abt, R. C., & Abt, K. L. (2013a). “Potential impact of bioenergy demand on the sustainability
of the southern forest resource.” Journal of Sustainable Forestry, 32, 175–194. https://
doi.org/10.1080/10549811.2011.652044
Abt, R. C., & Abt, K. L. (2013b). “Potential Impact of Bioenergy Demand on the Sustainability
of the Southern Forest Resource.” Journal of Sustainable Forestry, 32(1–2), 175–194.
https://doi.org/10.1080/10549811.2011.652044
Abt, R. C., Cubbage, F., & Abt, K. L. (2009a). “Projecting southern timber supply for multiple
products by subregion.” Forest Products Journal, 59(10409), 7–16.
Abt, R. C., Cubbage, F. W., & Abt, K. L. (2009b). “Projecting southern timber supply for
multiple products by subregion.” Forest Products Journal, 59(7/8), 7–16.
Abt, R. C., Cubbage, F. W., & Pacheco, G. (2000). “Southern Forest Source Assessment using
the subregional timber supply (SRTS) model.” Forest Products Journal, 50(4), 25–33.
Adami, M., Rudorff, B. F. T., Freitas, R. M., Aguiar, D. A., Sugawara, L. M., & Mello, M. P. (2012).
“Remote sensing time series to evaluate direct land use change of recent expanded
sugarcane crop in Brazil.” Sustainability, 4(4), 574–585. https://doi.org/10.3390/
su4040574
Agencia Nacional de Agua. (2013). Conjuntura dos Recursos Hidricos no Brasil 2013.
Agencia Nacional de Agua, Ministerio do Meio Ambiente, Brasilia, Brazil. https://doi.
org/10.5700/rausp1110
Aguilar, F. X., Cai, Z., & D’Amato, A. W. (2014). “Non-industrial private forest owner’s
willingness-to-harvest: How higher timber prices influence woody biomass supply.”
Biomass and Bioenergy, 71, 202–215. https://doi.org/10.1016/j.biombioe.2014.10.006
Alig, R. J., Plantinga, A. J., Ahn, S., & Kline, J. D. (2004). “Land use changes involving forestry
in the United States: 1952 to 1997, with projections to 2050.” General Technical Reports
of the US Department of Agriculture, Forest Service, 92.
Alkemade, R., van Oorschot, M., Miles, L., Nellemann, C., Bakkenes, M., & Ten Brink, B.
(2009). “GLOBIO3: A framework to investigate options for reducing global terrestrial
biodiversity loss.” Ecosystems, 12(3), 374–390. https://doi.org/10.1007/s10021-009-
9229-5
Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop evapotranspiration - Guidelines
for computing crop water requirements. Irrigation and drainage paper 56, FAO, Rome,
Italy.
Aycrigg, J., Beauvais, G., Gotthardt, T., Huettmann, F., Pyare, S., Andersen, M., … Walton, K.
(2015). “Novel Approaches to Modeling and Mapping Terrestrial Vertebrate Occurrence
in the Northwest and Alaska: An Evaluation.” Northwest Science, 89(4), 355–381.
https://doi.org/10.3955/046.089.0405
Beca, G., Vancine, M. H., Carvalho, C. S., Pedrosa, F., Alves, R. S. C., Buscariol, D., … Galetti,
M. (2017). “High mammal species turnover in forest patches immersed in biofuel
plantations.” Biological Conservation, 210, 352–359. https://doi.org/10.1016/j.
biocon.2017.02.033
254

References
Benítez-López, A., Alkemade, R., Schipper, A. M., Ingram, D. J., Verweij, P. A., Eikelboom, J.
A. J., & Huijbregts, M. A. J. (2017). “The impact of hunting on tropical mammal and bird
populations.” Science, 356(6334), 180–183. https://doi.org/10.1126/science.aaj1891
Berndes, G. (2002). “Bioenergy and water—the implications of large-scale bioenergy
production for water use and supply.” Global Environmental Change, 12, 253–271.
https://doi.org/10.1016/S0959-3780(02)00040-7
Berndes, G. (2008). “Water demand for global bioenergy production: trends, risks and
opportunities,” 48.
Berndes, G., Bird, N., & Cowie, A. (2010). Bioenergy , Land Use Change and Climate Change
Mitigation. IEA report for policy advisors and policy makers. IEA Bioenergy.
Bicheron, P., Defourny, P., Brockmann, C., Schouten, L., Vancutsem, C., Huc, M., … Arino, O.
(2008). GLOBCOVER: Products Description and Validation Report. ESA Bulletin. Medias-
France, Toulouse, France. https://doi.org/10013/epic.39884.d016
Boie, I., Breitschopf, B., Held, A., Ragwitz, M., Steinhilber, S., Resch, G., … Steinbäcker, S.
(2016). Policy Dialogue on the Assessment and Convergence of RES Policy in EU Member
States. Fraunhofer ISI.
Bontemps, S., Defourny, P., van Bogaert, E., Arino, O., Kalogirou, V., & Perez, J. R. (2011).
GLOBCOVER 2009: Products Description and Validation Report. Louvain-la-Neuve,
Belgium.
Booth, M. S. (2018). “Not carbon neutral: Assessing the net emissions impact of
residues burned for bioenergy.” Environmental Research Letters, 13(3). https://doi.
org/10.1088/1748-9326/aaac88
Bosmans, J. H. C., van Beek, L. P. H., Sutanudjaja, E. H., & Bierkens, M. F. P. (2017).
“Hydrological impacts of global land cover change and human water use.” Hydrology
and Earth System Sciences Discussions, 21, 5603–5626. https://doi.org/10.5194/hess- &
2016-621
Boyd, C., Brooks, T. M., Butchart, S. H. M., Edgar, G. J., Da Fonseca, G. A. B., Hawkins, F.,
… Van Dijk, P. P. (2008). “Spatial scale and the conservation of threatened species.”
Conservation Letters, 1(1), 37–43. https://doi.org/10.1111/j.1755-263X.2008.00002.x
BPBES. (2019). 1o Diagnóstico Brasileiro de Biodiversidade & Serviços Ecossistêmicos:
sumário para tomadores de decisão. Retrieved from www.bpbes.net.br.
Brockerhoff, E. G., Jactel, H., Parrotta, J. A., Quine, C. P., & Sayer, J. (2008). “Plantation forests
and biodiversity: Oxymoron or opportunity?” Biodiversity and Conservation, 17(5),
925–951. https://doi.org/10.1007/s10531-008-9380-x
Brown, D. G., Johnson, K. M., Loveland, T. R., & Theobald, D. M. (2005). “Rural Land-Use
Trends in the Conterminous United States, 1950-2000.” Ecological Applications, 15(6),
1851–1863.
Butler, S. M., Butler, B. J., & Markowski-Lindsay, M. (2017). “Family Forest Owner
Characteristics Shaped by Life Cycle, Cohort, and Period Effects.” Small-Scale Forestry,
16(1). https://doi.org/10.1007/s11842-016-9333-2
Cabral, O. M. R., Rocha, H. R., Gash, J. H., Ligo, M. A. V., Tatsch, J. D., Freitas, H. C., & Brasilio,
E. (2012). “Water use in a sugarcane plantation.” GCB Bioenergy, 4, 555–565. https://
doi.org/10.1111/j.1757-1707.2011.01155.x
Calderon, C., Colla, M., Jossart, J.-M., Hemeleers, N., Martin, A., Aveni, N., & Caferri, C. (2019).
Bioenergy Europe Statistical Report 2019.
Canadell, J., Jackson, R. B., Ehleringer, J. R., Mooney, H. A., Sala, O. E., & Schulze, E. D. (1996).
“Maximum rooting depth of vegetation types at the global scale.” Oecologia, 108, 583–
595.
255

References
Carnus, J.-M., Parrotta, J., Brockerhoff, E., Arbez, M., Jactel, H., Kremer, A., … Walters, B.
(2006). “Planted forests and biodiversity.” Journal of Forestry, 104(2), 65–77. https://
doi.org/10.1093/jof/104.2.65
Carvalho, F. M. V, De Marco, P., & Ferreira, L. G. (2009). “The Cerrado into-pieces: Habitat
fragmentation as a function of landscape use in the savannas of central Brazil.” Biological
Conservation, 142(7), 1392–1403. https://doi.org/10.1016/j.biocon.2009.01.031
CBD. (2010). Global Biodiversity Outlook 3. Montreal, Canada. https://doi.org/10.1002/
ejoc.201200111
Ceballos, G., & Ehrlich, P. R. (2006). “Global mammal distributions, biodiversity hotspots,
and conservation.” Proceedings of the National Academy of Sciences, 103(51), 19374–
19379. https://doi.org/10.1073/pnas.0609334103
Chapin, F. S., Zavaleta, E. S., Eviner, V. T., Naylor, R. L., Vitousek, P. M., Reynolds, H. L., …
Díaz, S. (2000). “Consequences of changing biodiversity.” Nature, 405(May), 234–242.
Chen, H., Tian, H., Liu, M., Melillo, J., Pan, S., & Zhang, C. (2006). “Effect of Land-Cover Change
on Terrestrial Carbon Dynamics in the Southern United States.” Journal of Environment
Quality, 35(4), 1533. https://doi.org/10.2134/jeq2005.0198
Cintas, O., Berndes, G., Cowie, A. L., Egnell, G., Holmström, H., Marland, G., & Agren, G. I.
(2017). “Carbon balances of bioenergy systems using biomass from forests managed
with long rotations : bridging the gap between stand and landscape assessments.” GCB
Bioenergy, 1238–1251. https://doi.org/10.1111/gcbb.12425
Cocchi, M., Nikolaisen, L., Junginger, M., Goh, C. S., Hess, R., Jacobson, J., … Schouwenberg, P. P.
(2011). Global Wood Pellet Industry Market and Trade Study. IEA report. IEA Bioenergy.
Colnes, A., Emick, H., Evans, A., Guild, F., Robards, T., Rambo, F., & High, C. (2012). “Biomass
Supply and Carbon Accounting for Southeastern Forests.” Report, (February), 132.
Conrad, J. L., Bolding, M. C., Smith, R. L., & Aust, W. M. (2011). “Wood-energy market impact
on competition, procurement practices, and profitability of landowners and forest
products industry in the U.S. south.” Biomass and Bioenergy, 35(1), 280–287. https://
doi.org/10.1016/j.biombioe.2010.08.038
Copley, A., & Lang, A. (2017). “Wood Bioenergy Update and Wood Pellet Exports: Q2 2017.”
Excerpt from Q2 2017 Forisk Research Quarterly (FRQ) Publication.
Costa, M. H., Botta, A., & Cardille, J. A. (2003). “Effects of large-scale changes in land cover
on the discharge of the Tocantins River, Southeastern Amazonia.” Journal of Hydrology,
283, 206–217. https://doi.org/10.1016/S0022-1694(03)00267-1
Costanza, J. K., Abt, R. C., Mckerrow, A. J., & Collazo, J. A. (2015). “Linking state-and-transition
simulation and timber supply models for forest biomass production scenarios.” AIMS
Environmental Science, 2(2), 180–202. https://doi.org/10.3934/environsci.2015.2.180
Costanza, J. K., Abt, R. C., McKerrow, A. J., & Collazo, J. A. (2017). “Bioenergy production
and forest landscape change in the southeastern United States.” GCB Bioenergy, 9(5),
924–939. https://doi.org/10.1111/gcbb.12386
Creutzig, F., Ravindranath, N. H., Berndes, G., Bolwig, S., Bright, R., Cherubini, F., … Masera,
O. (2015). “Bioenergy and climate change mitigation: An assessment.” GCB Bioenergy,
7(5), 916–944. https://doi.org/10.1111/gcbb.12205
Cubbage, F. W. (2006). “Fast-Grown Plantations , Forest Certification , and the U . S .
South : Environmental Benefits and Economic Sustainability *.” New Zealand Journal
of Forestry Science, 35(2005), 266–289.
Daigneault, A., Sohngen, B., & Sedjo, R. (2012). “Economic approach to assess the forest
carbon implications of biomass energy.” Environmental Science and Technology, 46(11),
5664–5671. https://doi.org/10.1021/es2030142
256

References
Dale, V. H., Brown, S., Haeuber, R. A., Hobbs, N. T., Huntly, N., Naiman, R. J., … Valone, T. J.
(2000). “Ecological Principles and Guidelines for Managing the Use of Land.” Ecological
Applications, 10(3), 639–670. https://doi.org/10.1890/04-0922
Dale, V. H., Kline, K. L., Parish, E. S., Cowie, A. L., Emory, R., Malmsheimer, R. W., … Wellisch,
M. (2017). “Status and prospects for renewable energy using wood pellets from the
southeastern United States.” GCB Bioenergy, 9(8), 1296–1305. https://doi.org/10.1111/
gcbb.12445
Dale, V. H., Kline, K. L., Wiens, J., & Fargione, J. (2010). “Biofuels : Implications for Land Use
and Biodiversity.” ESA Report, (January).
Dale, V. H., Parish, E., Kline, K. L., & Tobin, E. (2017). “How is wood-based pellet production
affecting forest conditions in the southeastern United States?” Forest Ecology and
Management, 396, 143–149. https://doi.org/10.1016/j.foreco.2017.03.022
Dallemand, E. J. F., Hilbert, J. A., & Monforti, F. (2015). Bioenergy and Latin America : A
Multi-Country Perspective. European Commission, Joint Research Centre, Institute for
Energy and Transport, Luxembourg. https://doi.org/10.2790/246697
Dauber, J., Hirsch, M., Simmering, D., Waldhardt, R., Otte, A., & Wolters, V. (2003).
“Landscape structure as an indicator of biodiversity: Matrix effects on species
richness.” Agriculture, Ecosystems and Environment, 98(1–3), 321–329. https://doi.
org/10.1016/S0167-8809(03)00092-6
de Abreu Bovo, A. A., Magioli, M., Percequillo, A. R., Kruszynski, C., Alberici, V., Mello,
M. A. R., … Ferraz, K. M. P. M. de B. (2018). “Human-modified landscape acts as
refuge for mammals in Atlantic Forest.” Biota Neotropica, 18(2), 1–12. https://doi.
org/10.1590/1676-0611-bn-2017-0395
de Cerqueira Leite, R. C., Verde Leal, M. R. L., Barbosa Cortez, L. A., Griffin, W. M., & Gaya
Scandiffio, M. I. (2009). “Can Brazil replace 5% of the 2025 gasoline world demand &
with ethanol?” Energy, 34(5), 655–661. https://doi.org/10.1016/j.energy.2008.11.001
De Vos, J. M., Joppa, L. N., Gittleman, J. L., Stephens, P. R., & Pimm, S. L. (2015). “Estimating
the normal background rate of species extinction.” Conservation Biology, 29(2), 452–
462. https://doi.org/10.1111/cobi.12380
Demuth, S., & Heinrich, B. (1997). “Temporal and spatial behaviour of drought in south
germany.” Proceedings of the Postojna, Slovenia, Conference, September-October 1997,
(246), 151–157.
Dermody, B. J., van Beek, R. P. H., Meeks, E., Klein Goldewijk, K., Scheidel, W., van der Velde,
Y., … Dekker, S. C. (2014). “A virtual water network of the Roman world.” Hydrology
and Earth System Sciences Discussions, 18, 5025–5040. https://doi.org/10.5194/hessd-
11-6561-2014
Díaz, S., Settele, J., Brondízio, E., Ngo, H., Guèze, M., Agard, J., … Chan, K. (2019). Summary for
policymakers of the global assessment report on biodiversity and ecosystem services of
the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services.
Diniz-Filho, J. A. F., Bini, L. M., Pinto, M. P., Terribile, L. C., De Oliveira, G., Vieira, C. M., …
Bastos, R. P. (2008). “Conservation planning: A macroecological approach using the
endemic terrestrial vertebrates of the Brazilian Cerrado.” Oryx, 42(4), 567–577. https://
doi.org/10.1017/S0030605308001129
Diogo, V., van der Hilst, F., Van Eijck, J., Verstegen, J. A., Hilbert, J., Carballo, S., … Faaij,
A. (2014). “Combining empirical and theory-based land-use modelling approaches
to assess economic potential of biofuel production avoiding iLUC: Argentina as a
case study.” Renewable and Sustainable Energy Reviews, 34, 208–224. https://doi.
org/10.1016/j.rser.2014.02.040
257

References
Dobson, A. P., Rodriguez, J. P., Roberts, W. M., & Wilcove, D. C. (1997). “Geographic
distribution of endangered species in the United States.” Science, 275(January), 550–
551.
Dornburg, V., Faaij, A. P. C., & Verweij, P. A. (2012). Biomass Assessment – Assessment of
global biomass potentials and their links to food , water , biodiversity , energy demand
and economy. Bilthoven, Netherlands. https://doi.org/10.1007/s00521-010-0414-4
Drummond, M. A., & Loveland, T. R. (2010). “Land-use Pressure and a Transition to Forest-
cover Loss in the Eastern United States.” BioScience, 60(4), 286–298. https://doi.
org/10.1525/bio.2010.60.4.7
Duarte, C. G., Gaudreau, K., Gibson, R. B., & Malheiros, T. F. (2013). “Sustainability
assessment of sugarcane-ethanol production in Brazil: A case study of a sugarcane
mill in São Paulo state.” Ecological Indicators, 30, 119–129. https://doi.org/10.1016/j.
ecolind.2013.02.011
Duden, A. S., Verweij, P. A., Junginger, H. M., Abt, R. C., Henderson, J. D., Dale, V. H., … van der
Hilst, F. (2017). “Modeling the impacts of wood pellet demand on forest dynamics in
southeastern United States.” Biofuels, Bioproducts and Biorefining, 11(6), 1–23. https://
doi.org/10.1002/bbb.1803
Eken, G., Bennun, L., Brooks, T. M., Darwall, W., Fishpool, L. D. C., Foster, M., … Tordoff, A.
(2004). “Key Biodiversity Areas as Site Conservation Targets.” BioScience, 54(12), 1110.
https://doi.org/10.1641/0006-3568(2004)054[1110:KBAASC]2.0.CO;2
European Biomass Association. (2016). European Bioenergy Outlook 2016. AEBIOM.
Brussels.
European Commission. (2017). Proposal for a directive of the European Parliament and
of the council on the promotion of the use of energy from renewable sources (recast).
Brussels, Belgium.
European Parliament and Council. “Regulation (EU) 2018/841 of the European Parliament
and of the Council of 30 May 2018 on the inclusion of greenhouse gas emissions and
removals from land use, land use change and forestry in the 2030 climate and energy
framework” (2018).
Evans, J. M., Fletcher, R. J., Alavalapati, J., Calabria, J., Geller, D., Smith, A. L., … Vasudev, D.
(2013). Forestry Bioenergy in the Southeast United States: Implications for Wildlife
Habitat and Biodiversity. National Wildlife Federation. Merrifield, VA, United States.
Fachinelli, N. P., & Pereira, A. O. (2015). “Impacts of sugarcane ethanol production in
the Paranaiba basin water resources.” Biomass and Bioenergy, 83, 8–16. https://doi.
org/10.1016/j.biombioe.2015.08.015
Faleiro, F. V., Machado, R. B., & Loyola, R. D. (2013). “Defining spatial conservation priorities
in the face of land-use and climate change.” Biological Conservation, 158, 248–257.
https://doi.org/10.1016/j.biocon.2012.09.020
FAO. (2015). Forest products Annual Market Review 2014-2015. United Nations Food and
Agriculture Organization. Geneva.
FAO. (2019). “FAOSTAT Statistical Database.” Retrieved May 26, 2019, from http://www.
fao.org/faostat/en/#home
Fargione, J., Hill, J., Tilman, D., Polasky, S., & Hawthorne, P. (2008). “Land clearing and
the biofuel carbon debt.” Science, 319(5867), 1235–1238. https://doi.org/10.1126/
science.1152747
Firbank, L. G. (2008). “Assessing the Ecological Impacts of Bioenergy Projects.” BioEnergy
Research, 1(1), 12–19. https://doi.org/10.1007/s12155-007-9000-8
258

References
Fischer, G., Nachtergaele, F. O., Prieler, S., Teixeira, E., Tóth, G., Velthuizen, H. van, … Wiberg,
D. (2012). Global Agro-ecological Zones (GAEZ v3.0); Model documentation. IIASA,
Laxenburg, Austria and FAO, Rome, Italy.
Flach, R., Ran, Y., Godar, J., Karlberg, L., & Suavet, C. (2016). “Towards more spatially
explicit assessments of virtual water flows: Linking local water use and scarcity to
global demand of Brazilian farming commodities.” Environmental Research Letters,
11, 1–10. https://doi.org/10.1088/1748-9326/11/7/075003
Fleishman, E., Blockstein, D. E., Hall, J. A., Mascia, M. B., Rudd, M. A., Scott, J. M., … Vedder,
A. (2011). “Top 40 Priorities for Science to Inform US Conservation and Management
Policy.” BioScience, 61(4), 290–300. https://doi.org/10.1525/bio.2011.61.4.9
Fleishman, E., Noss, R. F., & Noon, B. R. (2006). “Utility and limitations of species richness
metrics for conservation planning.” Ecological Indicators, 6(3), 543–553. https://doi.
org/10.1016/j.ecolind.2005.07.005
Fletcher, R. J., Robertson, B. A., Evans, J. M., Doran, P. J., Alavalapati, J. R. R., &
Schemske, D. W. (2011). “Biodiversity conservation in the era of biofuels: Risks and
opportunities.” Frontiers in Ecology and the Environment, 9(3), 161–168. https://doi.
org/10.1890/090091
Folke, C., Holling, C. S., & Perrings, C. (1996). “Biological Diversity, Ecosystem , and the
Human Scale.” Ecological Applications, 6(4), 1018–1024.
Fox, T. R., Jokela, E. J., & Allen, H. L. (2007). “The Development of Pine Plantation Silviculture
in the Southern United States.” Journal of Forestry, 105(November), 337–347.
Fricko, O., Havlik, P., Rogelj, J., Klimont, Z., Gusti, M., Johnson, N., … Riahi, K. (2017). “The
marker quantification of the Shared Socioeconomic Pathway 2: A middle-of-the-road
scenario for the 21st century.” Global Environmental Change, 42, 251–267. https://doi.
org/10.1016/j.gloenvcha.2016.06.004 &
Fritts, S., Moorman, C., Grodsky, S., Hazel, D., Homyack, J., Farrell, C., & Castleberry, S.
(2016). “Do biomass Harvesting Guidelines influence herpetofauna following harvests
of logging residues for renewable energy?” Ecological Applications, 26(3), 926–939.
https://doi.org/10.1890/14-2078/suppinfo
FSC. (2019). “Forest Stewardship Council.” Retrieved from https://www.fsc.org/en
Galik, C. S., & Abt, R. C. (2012). “The effect of assessment scale and metric selection on the
greenhouse gas benefits of woody biomass.” Biomass and Bioenergy, 44, 1–7. https://
doi.org/10.1016/j.biombioe.2012.04.009
Galik, C. S., & Abt, R. C. (2015). “Sustainability guidelines and forest market response: an
assessment of European Union pellet demand in the southeastern United States.” GCB
Bioenergy, 1–12. https://doi.org/10.1111/gcbb.12273
Galik, C. S., & Abt, R. C. (2016). “Sustainability guidelines and forest market response: An
assessment of European Union pellet demand in the southeastern United States.” GCB
Bioenergy, 8(3), 658–669. https://doi.org/10.1111/gcbb.12273
Galik, C. S., Abt, R. C., Latta, G., & Vegh, T. (2015). “The environmental and economic effects
of regional bioenergy policy in the southeastern U.S.” Energy Policy, 85, 335–346.
https://doi.org/10.1016/j.enpol.2015.05.018
Galik, C. S., Abt, R. C., & Wu, Y. (2009a). “Forest Biomass Supply in the Southeastern United
States—Implications for Industrial Roundwood and Bioenergy Production.” Journal of
Forestry, 107(March 2009), 8.
259

References
Galik, C. S., Abt, R. C., & Wu, Y. (2009b). “Forest Biomass Supply in the Southeastern United
States --Implications for Industrial Roundwood and Bioenergy Production.” Journal of
Forestry, 107(2), 69–77.
GBEP. (2019). “Global Bioenergy Partnership.” Retrieved from http://www.globalbioenergy.org/
Georgescu, M., Lobell, D. B., Field, C. B., & Mahalov, A. (2013). “Simulated hydroclimatic
impacts of projected Brazilian sugarcane expansion.” Geophysical Research Letters,
40, 972–977. https://doi.org/10.1002/grl.50206
Gerbens-Leenes, P. W., Hoekstra, A. Y., & van der Meer, T. H. (2009). “The water footprint
of bioenergy.” Proceedings of the National Academy of Sciences of the United States of
America, 106(25), 10219–10223. https://doi.org/10.1073/pnas.0812619106
Gerssen-Gondelach, S. J., Wicke, B., & Faaij, A. P. C. (2017). “GHG emissions and other
environmental impacts of indirect land use change mitigation.” GCB Bioenergy, 9(4),
725–742. https://doi.org/10.1111/gcbb.12394
Gestich, C. C., Arroyo-Rodríguez, V., Ribeiro, M. C., da Cunha, R. G. T., & Setz, E. Z. F. (2019).
“Unraveling the scales of effect of landscape structure on primate species richness and
density of titi monkeys (Callicebus nigrifrons).” Ecological Research, 34(1), 150–159.
https://doi.org/10.1111/1440-1703.1009
Gibbs, H. K., Johnston, M., Foley, J. A., Holloway, T., Monfreda, C., Ramankutty, N., & Zaks,
D. (2008). “Carbon payback times for crop-based biofuel expansion in the tropics: The
effects of changing yield and technology.” Environmental Research Letters, 3(3), 1–10.
https://doi.org/10.1088/1748-9326/3/3/034001
Given, D. R., & Norton, D. A. (1993). “A multivariate approach to assessing threat and for
priority setting in threatened species conservation.” Biological Conservation, 64, 57–66.
Global Runoff Data Centre (GRDC). (2009). Long‐term mean monthly discharges and annual
characteristics of GRDC stations. Global Runoff Data Centre, Koblenz, Germany.
Gottlieb, I. G. W., Fletcher, R. J., Nuñez-Regueiro, M. M., Ober, H., Smith, L., & Brosi, B. J.
(2017). “Alternative biomass strategies for bioenergy: implications for bird communities
across the southeastern United States.” GCB Bioenergy, 9(11), 1606–1617. https://doi.
org/10.1111/gcbb.12453
Green, R. E., Cornell, S. J., Scharlemann, J. P. W., & Balmford, A. (2005). “Farming and the fate
of wild nature.” Science (New York, N.Y.), 307(5709), 550–555. https://doi.org/10.1126/
science.1106049
Gregory, R. D., van Strien, A., Vorisek, P., Gmelig Meyling, A. W., Noble, D. G., Foppen, R. P.
B., & Gibbons, D. W. (2005). “Developing indicators for European birds.” Philosophical
Transactions of the Royal Society B: Biological Sciences, 360(1454), 269–288. https://
doi.org/10.1098/rstb.2004.1602
Grodsky, S. M., Moorman, C. E., Fritts, S. R., Castleberry, S. B., & Wigley, T. B. (2016).
“Breeding, early-successional bird response to forest harvests for bioenergy.” PLoS
ONE, 11(10), 1–20. https://doi.org/10.1371/journal.pone.0165070
Grodsky, S. M., Moorman, C. E., Fritts, S. R., Hazel, D. W., Homyack, J. A., Castleberry, S.
B., & Wigley, T. B. (2016). “Winter bird use of harvest residues in clearcuts and the
implications of forest bioenergy harvest in the southeastern United States.” Forest
Ecology and Management, 379(November 2016), 91–101. https://doi.org/10.1016/j.
foreco.2016.07.045
Guarenghi, M. ., & Walter, A. (2016). “Assessing potential impacts of sugarcane production
on water resources: A case study in Brazil.” Biofuels, Bioproducts and Biorefining, 10,
699–709. https://doi.org/10.1002/bbb
260

References
Gupta, H. V, Kling, H., Yilmaz, K. K., & Martinez, G. F. (2009). “Decomposition of the
mean squared error and NSE performance criteria: Implications for improving
hydrological modelling.” Journal of Hydrology, 377, 80–91. https://doi.org/10.1016/j.
jhydrol.2009.08.003
Hagemann, S., Botzet, M., Dümenil, L., & Machenhauer, B. (1999). Derivation of Global GCM
Boundary Conditions from 1 Km Land Use Satellite Data. Max Planck Institute for
Meteorology, Hamburg, Germany.
Hagemann, S., & Gates, L. D. (2003). “Improving a subgrid runoff parameterization
scheme for climate models by the use of high resolution data derived from satellite
observations.” Climate Dynamics, 21, 349–359. https://doi.org/10.1007/s00382-003-
0349-x
Hanssen, S. V., Duden, A. S., Junginger, M., Dale, V. H., & van der Hilst, F. (2017). “Wood pellets,
what else? Greenhouse gas parity times of European electricity from wood pellets
produced in the south-eastern United States using different softwood feedstocks.” GCB
Hardie, I., Parks, P., Gottleib, P., & Wear, D. (2000). “Responsiveness of Rural and Urban
Land Uses to Land Rent Determinants in the U.S. South.” Land Economics, 76(4), 659.
https://doi.org/10.2307/3146958
Hernandes, T. A. D., Bof Bufon, V., & Seabra, J. E. A. (2013). “Water footprint of biofuels
in Brazil: assessing regional differences.” Biofuels, Bioproducts and Biorefining, 6,
241–252. https://doi.org/10.1002/bbb
Hernandes, T. A. D., Scarpare, F. V., & Seabra, J. E. A. (2018). “Assessment of impacts on basin
stream flow derived from medium-term sugarcane expansion scenarios in Brazil.”
Agriculture, Ecosystems & Environment, 259(March), 11–18. https://doi.org/10.1016/j.
agee.2018.02.026 &
Hisdal, H., Tallaksen, L., Clausen, B., Peters, E., & Gustard, A. (2004). “Hydrological
drought characteristics, Hydrological drought, Processes and estimation methods for
streamflow and groundwater.” Developments in Water Science, 48, 139–198.
Hoekstra, A. Y., Mekonnen, M. M., Chapagain, A. K., Mathews, R. E., & Richter, B. D. (2012).
“Global monthly water scarcity: Blue water footprints versus blue water availability.”
PLoS ONE, 7(2). https://doi.org/10.1371/journal.pone.0032688
Hof, C., Araújo, M. B., Jetz, W., & Rahbek, C. (2011). “Additive threats from pathogens, climate
and land-use change for global amphibian diversity.” Nature, 480(7378), 516–519.
https://doi.org/10.1038/nature10650
Holdridge, L. R. (1967). Life Zone Ecology. Tropical Science Center, San Jose, Costa Rica.
Homer, C. G., Dewitz, J. A., Yang, L., Jin, S., Danielson, P., Xian, G., … Megown, K. (2015).
“Completion of the 2011 National Land Cover Database for the conterminous United
States-Representing a decade of land cover change information.” Photogrammetric
Engineering and Remote Sensing, 81(5), 345–354.
Homer, C. G., Dewitz, J., Fry, J., Coan, M., Hossain, N., Larson, C., … Wickham, J. (2007).
“Completion of the 2001 National Land Cover Database for the Conterminous United
States.” Photogrammetric Engineering and Remote Sensing, 73(4), 337–341.
Hunke, P., Mueller, E. N., Schröder, B., & Zeilhofer, P. (2015). “The Brazilian Cerrado:
Assessment of water and soil degradation in catchments under intensive agricultural
use.” Ecohydrology, 8(6), 1154–1180. https://doi.org/10.1002/eco.1573
IEA. (2014). World Energy Outlook 2014. https://doi.org/10.1787/weo-2014-en
261

References
IEA. (2018). Renewables 2018: Analysis and Forecasts to 2023. International Energy
Agency, Paris, France.
IEA. (2019). World Energy Outlook 2019. https://doi.org/DOE/EIA-0383(2012) U.S.
Immerzeel, D. J., Verweij, P. A., van der Hilst, F., & Faaij, A. P. C. (2014). “Biodiversity impacts
of bioenergy crop production: A state-of-the-art review.” GCB Bioenergy, 6(3), 183–209.
https://doi.org/10.1111/gcbb.12067
Ince, P. J., & Nepal, P. (2012). “Effects on U . S . Timber Outlook of Recent Economic Recession
, Collapse in Housing Construction , and Wood Energy Trends.” General Technical Report
FPK-GTR-219, 18.
IPBES. (2018). Biodiversity and Ecosystem Services. Encyclopedia of Biodiversity. IPBES,
Bonn, Germany. https://doi.org/10.1016/B978-0-12-384719-5.00349-X
IPCC. (2011). Renewable energy sources and climate change mitigation: Special report of
the intergovernmental panel on climate change. Renewable Energy Sources and Climate
Change Mitigation: Special Report of the Intergovernmental Panel on Climate Change.
https://doi.org/10.1017/CBO9781139151153
IPCC. (2019). Climate Change and Land: an IPCC special report on climate change,
desertification, land degradation, sustainable land management, food security, and
greenhouse gas fluxes in terrestrial ecosystems. https://doi.org/http://www.ipcc.ch/
publications_and_data/ar4/wg2/en/spm.html
IRENA. (2019). Global energy transformation: A roadmap to 2050 (2019 edition).
International Renewable Energy Agency.
IUCN. (2019). “Red List of Threatened Species.” Retrieved July 12, 2019, from https://www.
iucnredlist.org/
Jenkins, C. N., Alves, M. A. S., Uezu, A., & Vale, M. M. (2015). “Patterns of vertebrate diversity
and protection in Brazil.” PLoS ONE, 10(12), 1–13. https://doi.org/10.1371/journal.
pone.0145064
Jenkins, C. N., Van Houtan, K. S., Pimm, S. L., & Sexton, J. O. (2015). “US protected lands
mismatch biodiversity priorities.” Proceedings of the National Academy of Sciences,
112(16), 5081–5086. https://doi.org/10.1073/pnas.1418034112
Jetz, W., Wilcove, D. S., & Dobson, A. P. (2007). “Projected impacts of climate and land-use
change on the global diversity of birds.” PLoS Biology, 5(6), 1211–1219. https://doi.
org/10.1371/journal.pbio.0050157
Jonker, J. G. G., Junginger, M., & Faaij, A. (2014). “Carbon payback period and carbon offset
parity point of wood pellet production in the South-eastern United States.” GCB
Jonker, J. G. G., van der Hilst, F., Junginger, H. M., Cavalett, O., Chagas, M. F., & Faaij, A. P. C.
(2015). “Outlook for ethanol production costs in Brazil up to 2030, for different biomass
crops and industrial technologies.” Applied Energy, 147, 593–610.
Jonker, J. G. G., van der Hilst, F., Markewitz, D., Faaij, A. P. C., & Junginger, H. M. (2018).
“Carbon balance and economic performance of pine plantations for bioenergy
production in the Southeastern United States.” Biomass and Bioenergy, 117(June),
44–55. https://doi.org/10.1016/j.biombioe.2018.06.017
Joshi, S., & Arano, K. G. (2009). “Determinants of private forest management decisions: A
study on West Virginia NIPF landowners.” Forest Policy and Economics, 11(2), 132–139.
https://doi.org/10.1016/j.forpol.2008.10.005
Joudrey, J., McDow, W., Smith, T., & Larson, B. (2012). European Power from U.S. Forests: How
Evolving EU Policy Is Shaping the Transatlantic Trade in Wood Biomass. Environmental
Defense Fund, New York, United States.
262

References
Junginger, H. M., Mai-Moulin, T., Daioglou, V., Fritsche, U., Guisson, R., Hennig, C., … Wild,
M. (2019). “The future of biomass and bioenergy deployment and trade: a synthesis of
15 years IEA Bioenergy Task 40 on sustainable bioenergy trade.” Biofuels, Bioproducts
and Biorefining, 13(2), 247–266. https://doi.org/10.1002/bbb.1993
Kavallari, A., Smeets, E., & Tabeau, A. (2014). “Land use changes from EU biofuel use: A
sensitivity analysis.” Operational Research, 14(2), 261–281. https://doi.org/10.1007/
s12351-014-0155-8
Kennedy, C. M., Hawthorne, P. L., Miteva, D. A., Baumgarten, L., Sochi, K., Matsumoto, M.,
… Kiesecker, J. (2016). “Optimizing land use decision-making to sustain Brazilian
agricultural profits, biodiversity and ecosystem services.” Biological Conservation,
204, 221–230. https://doi.org/10.1016/j.biocon.2016.10.039
Kier, G., & Barthlott, W. (2001). “Measuring and mapping endemism and species richness:
a new methodological approach and its application on the flora of Africa.” Biodiversity
and Conservation, 10, 1513–1529.
Kirby, K. R., Laurance, W. F., Albernaz, A. K., Schroth, G., Fearnside, P. M., Bergen, S., … da
Costa, C. (2006). “The future of deforestation in the Brazilian Amazon.” Futures, 38(4),
432–453. https://doi.org/10.1016/j.futures.2005.07.011
Klink, C. A., & Machado, R. B. (2005). “Conservation of the Brazilian Cerrado. Biodiversity
and conservation of plants in Brazil.” Conservation Biology, 19(3), 707–713.
Koh, L. P., & Ghazoul, J. (2008). “Biofuels, biodiversity, and people: Understanding the
conflicts and finding opportunities.” Biological Conservation, 141(10), 2450–2460.
https://doi.org/10.1016/j.biocon.2008.08.005
Laffan, S. W., & Crisp, M. D. (2003). “Assessing endemism at multiple spatial scales, with an
example from the Australian vascular flora.” Journal of Biogeography, 30(4), 511–520.
https://doi.org/10.1046/j.1365-2699.2003.00875.x &
Lamb, E. G., Bayne, E., Holloway, G., Schieck, J., Boutin, S., Herbers, J., & Haughland, D.
L. (2009). “Indices for monitoring biodiversity change: Are some more effective
than others?” Ecological Indicators, 9(3), 432–444. https://doi.org/10.1016/j.
ecolind.2008.06.001
Lamers, P., Hoefnagels, R., Junginger, M., Hamelinck, C., & Faaij, A. (2015). “Global solid
biomass trade for energy by 2020: An assessment of potential import streams and
supply costs to North-West Europe under different sustainability constraints.” GCB
Lamers, P., Junginger, M., Hamelinck, C., & Faaij, A. (2012). “Developments in international
solid biofuel trade - An analysis of volumes, policies, and market factors.” Renewable
and Sustainable Energy Reviews, 16(5), 3176–3199. https://doi.org/10.1016/j.
rser.2012.02.027
Lapola, D. M., Schaldach, R., Alcamo, J., Bondeau, A., Koch, J., Koelking, C., & Priess, J. A.
(2010). “Indirect land-use changes can overcome carbon savings from biofuels in
Brazil.” Proceedings of the National Academy of Sciences, 107(8), 3388–3393. https://
doi.org/10.1073/pnas.0907318107
Lawler, J. J., O’Connor, R. J., Hunsaker, C. T., Jones, K. B., Loveland, T. R., & White, D. (2004).
“The effects of habitat resolution on models of avian diversity and distributions: A
comparison of two land-cover classifications.” Landscape Ecology, 19(5), 517–532.
https://doi.org/10.1023/B:LAND.0000036151.28327.01
Lawton, J. H., Bignell, D. E., Bolton, B., Bloemers, G. F., Eggleton, P., Hammond, P. M., … Watt,
A. D. (1998). “Biodiversity inventories, indicator taxa and effects of habitat modification
in tropical forest.” Nature, 391, 72–76.
263

References
Lefcheck, J. S., Byrnes, J. E. K., Isbell, F., Gamfeldt, L., Griffin, J. N., Eisenhauer, N., … Duffy, J.
E. (2015). “Biodiversity enhances ecosystem multifunctionality across trophic levels
and habitats.” Nature Communications, 6, 1–7. https://doi.org/10.1038/ncomms7936
Liang, J., Crowther, T. W., Picard, N., Wiser, S., Zhou, M., Alberti, G., … Reich, P. B. (2016).
“Positive biodiversity-productivity relationship predominant in global forests.” Science,
354(6309), 196–207. https://doi.org/10.1126/science.aaf8957
Loarie, S. R., Lobell, D. B., Asner, G. P., Mu, Q., & Field, C. B. (2011). “Direct impacts on local
climate of sugar-cane expansion in Brazil.” Nature Climate Change, 1(2), 105–109.
https://doi.org/10.1038/nclimate1067
Louette, G., Maes, D., Alkemade, J. R. M., Boitani, L., de Knegt, B., Eggers, J., … Delbaere,
B. (2010). “BioScore-Cost-effective assessment of policy impact on biodiversity using
species sensitivity scores.” Journal for Nature Conservation, 18(2), 142–148. https://
doi.org/10.1016/j.jnc.2009.08.002
Loveland, T. R., Reed, B. C., Brown, J. F., Ohlen, D. O., Zhu, Z., Yang, L., & Merchant, J. W.
(2000). “Development of a global land cover characteristics database and IGBP DISCover
from 1 km AVHRR data.” International Journal of Remote Sensing, 21(6), 1303–1330.
Lubowski, R. N., Plantinga, A. J., Stavins, R. N., Ruben, N., Plantinga, J., & Robert, N. (2008).
“What drives land-use change in the United States? A national analysis of landowner
decisions.” Land Economics, 84(4), 529–550. https://doi.org/10.3368/le.84.4.529
Maneta, M. P., Torres, M. O., Wallender, W. W., Vosti, S., Howitt, R., Rodrigues, L., … Panday,
S. (2009). “A spatially distributed hydroeconomic model to assess the effects of drought
on land use, farm profits, and agricultural employment.” Water Resources Research,
45(11), 1–19. https://doi.org/10.1029/2008WR007534
Marin, F. R., Jones, J. W., Royce, F., Suguitani, C., Donzeli, J. L., Filho, W. J. P., & Nassif, D.
S. P. (2011). “Parameterization and evaluation of predictions of DSSAT/CANEGRO for
Brazilian sugarcane.” Agronomy Journal, 103(2), 304–315. https://doi.org/10.2134/
agronj2010.0302
Martinelli, L. A., & Filoso, S. (2008). “Expansion of sugarcane ethanol production in Brazil:
Environmental and social challenges.” Ecological Applications, 18(4), 885–898. https://
doi.org/10.1890/07-1813.1
Marx, A., Kumar, R., Thober, S., Rakovec, O., Wanders, N., Zink, M., … Samaniego, L. (2018).
“Climate change alters low flows in Europe under global warming of 1.5, 2, and 3 ◦ C.”
Hydrology and Earth System Sciences, 22, 1017–1032.
Master, L. L., Faber-Langendoen, D., Bittman, R., Hammerson, G. A., Heidel, B., Ramsay,
L., … Tomaino, A. (2012). NatureServe Conservation Status Assessments: Factors for
Assessing Extinction Risk. NatureServe, Arlington, VA, United States.
Maxwell, J., & Gergely, K. (2005). Gap Analysis--A geographic approach to planning for
biological diversity. U.S. Geological Survey Gap Analysis Program Bulletin No. 13.
Maxwell, J., Gergely, K., Aycrigg, J., & Davidson, A. (2009). Gap Analysis--A geographic
approach to planning for biological diversity. U.S. Geological Survey Gap Analysis
Program Bulletin No. 16.
McKerrow, A., Tarr, N., Rubino, M., & Williams, S. (2018). “Patterns of species richness
hotspots and estimates of their protection are sensitive to spatial resolution.” Diversity
and Distributions, 00, 14.
McKerrow, A., Williams, S., & Collazo, J. (2006). The North Carolina Gap Analysis Project
Final Report. North Carolina Cooperative Fish and Wildlife Research Unit, North
Carolina State University, Raleigh, NC. Cooperative research agreement 1434-HQ-97-
RU-01568.
264

References
McKinney, M. L. (2008). “Effects of urbanization on species richness: A review of plants

and animals.” Urban Ecosystems, 11(2), 161–176. https://doi.org/10.1007/s11252-007-
0045-4
Melo, D. D. C. D., Scanlon, B. R., Zhang, Z., Wendland, E., & Yin, L. (2016). “Reservoir storage
and hydrologic responses to droughts in the Paraná River basin, south-eastern Brazil.”
Hydrology and Earth System Sciences, 20, 4673–4688. https://doi.org/10.5194/hess-
20-4673-2016
Meyer, M. A., Chand, T., & Priess, J. A. (2015). “Comparing bioenergy production sites in the
southeastern US regarding ecosystem service supply and demand.” PLoS ONE, 10(3),
1–24. https://doi.org/10.1371/journal.pone.0116336
Millenium Ecosystem Assessment. (2005). Ecosystems and Human Well-Being: Synthesis.
Island Press, Washington, DC, United States.
Miller, D. A., Wigley, T. B., & Miller, K. V. (2009). “Managed Forests and Conservation of
Terrestrial Biodiversity in the Southern United States.” Journal of Forestry, 107(June),
197–203.
Miner, R. A., Abt, R. C., Bowyer, J. L., Buford, M. A., Malmsheimer, R. W., Laughlin, J. O., …
Skog, K. E. (2014). “Forest Carbon Accounting Considerations in US Bioenergy Policy.”
Journal of Forestry, 112(November), 591–606. https://doi.org/10.5849/jof.14-009
Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M., & Gascon, C. (2011).
“Biodiversity Hotspots.” https://doi.org/10.1007/978-3-642-20992-5
MMA. (2014). Lista Nacional de espécies ameaçadas portaria MMA no 444. Retrieved
from http://www.icmbio.gov.br/cepsul/images/stories/legislacao/Portaria/2014/p_
mma_444_2014_lista_espécies_ameçadas_extinção.pdf
MMA. (2015). Fifth National Report to the CBD - Brazil. https://doi.org/10.1044/leader.
PPL.20012015.20 &
MMA. (2018). “Taxa de desmatamento na Amazônia Legal.” Retrieved August 4, 2019, from
http://www.mma.gov.br/informma/item/15259-governo-federal-divulga-taxa-de-
desmatamento-na-amazônia.html
MMA. (2019). “Mapas de Cobertura Vegetal dos Biomas Brasileiros: Cerrado.” Retrieved
from http://mapas.mma.gov.br/mapas/aplic/probio/datadownload.htm
Molden, D. (2007). Comprehensive Assessment of Water Management in Agriculture.
Earthscan, London and International Water Management Institute, Colombo. https://
doi.org/10.1007/978-3-540-85156-1
Moreira, J. R. (2007). “Water Use and Impacts Due Ethanol Production in Brazil.” In
Proceedings Linkages between Energy and Water Management for Agriculture in
Developing Countries. Hyderabad, India, January 29– 30, 2007 (p. 24).
Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R., Rose, S. K., Van Vuuren, D. P., …
Wilbanks, T. J. (2010). “The next generation of scenarios for climate change research
and assessment.” Nature, 463, 747–756. https://doi.org/10.1038/nature08823
Myers, N. (2003). “Biodiversity Hotspots Revisited.” BioScience, 53(10), 916. https://doi.
org/10.1641/0006-3568(2003)053[0916:BHR]2.0.CO;2
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000).
“Biodiversity hotspots for conservation priorities.” Nature, 403(6772), 853–858.
https://doi.org/10.1038/35002501
Napton, D. E., Auch, R. F., Headley, R., & Taylor, J. L. (2010). “Land changes and their driving
forces in the Southeastern United States.” Regional Environmental Change, 10(1),
37–53. https://doi.org/10.1007/s10113-009-0084-x
265

References
NatureServe. (2017). “NatureServe Explorer.”

Naylor, R., Liska, A., Burke, M., Falcon, W., Gaskell, J., Rozelle, S., & Cassman, K. G. (2007).
“The ripple effect. Biofuels, food security, and the environment.” Environment, 49(9),
30–43.
Nobre, C. A., Marengo, J. A., Seluchi, M. E., Cuartas, L. A., & Alves, L. M. (2016). “Some
Characteristics and Impacts of the Drought and Water Crisis in Southeastern Brazil
during 2014 and 2015.” Journal of Water Resource and Protection, 08(02), 252–262.
https://doi.org/10.4236/jwarp.2016.82022
Norris, K. (2008). “Agriculture and biodiversity conservation: opportunity knocks.”
Conservation Letters, 1(1), 2–11. https://doi.org/10.1111/j.1755-263x.2008.00007.x
Noss, R. F. (1983). “A Regional Landscape Approach to Maintain Diversity,” 33(11), 700–706.
Noss, R. F. (1990). “Indicators for monitoring biodiversity: a hierarchical approach.”
Conservation Biology, 4(December), 355–364.
Noss, R. F., Platt, W. J., Sorrie, B. A., Weakley, A. S., Means, D. B., Costanza, J., & Peet, R. K.
(2015). “How global biodiversity hotspots may go unrecognized: Lessons from the
North American Coastal Plain.” Diversity and Distributions, 21(2), 236–244. https://
doi.org/10.1111/ddi.12278
NRDC. (2015). In the U . S . Southeast , Natural Forests Are Being Felled to Send Fuel
Overseas.
O’Neill, B. C., Kriegler, E., Ebi, K. L., Kemp-Benedict, E., Riahi, K., Rothman, D. S., … Solecki,
W. (2017). “The roads ahead: Narratives for shared socioeconomic pathways describing
world futures in the 21st century.” Global Environmental Change, 42, 169–180. https://
doi.org/10.1016/j.gloenvcha.2015.01.004
OECD, & FAO. (2015). Agricultural Outlook 2015-2024. OECD Publishing, Paris, France.
Olesen, A. S., Kittler, B., Price, W., & Aguilar, F. X. (2016). Environmental Implications of
Increased Reliance of the EU on Biomass from the South East US. European Commission,
Brussels, Belgium. https://doi.org/10.2779/30897
Oliveira, P. T. S., Nearing, M. A., Moran, M. S., Goodrich, D. C., Wendland, E., & Gupta, H. V.
(2014). “Trends in water balance components across the Brazilian Cerrado.” Water
Resources Research, 7100–7114. https://doi.org/10.1002/2013WR015202.Received
Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood,
E. C., … Kassem, K. R. (2001). “Terrestrial Ecoregions of the World: A New Map of Life on
Earth.” BioScience, 51(11), 933. https://doi.org/10.1641/0006-3568(2001)051[0933:te
otwa]2.0.co;2
Orme, C. D. L., Davies, R. G., Burgess, M., Eigenbrod, F., Pickup, N., Olson, V. A., … Owens, I.
P. F. (2005). “Global hotspots of species richness are not congruent with endemism or
threat.” Nature, 436(7053), 1016–1019. https://doi.org/10.1038/nature03850
Oswalt, S. N., Smith, W. B., Miles, P. D., & Pugh, S. A. (2014a). Forest resources of the
United States, 2012: A technical document supporting the Forest Service update of
the 2010 RPA Assessment. USDA General Technical Report WO-91. U.S. Department of
Agriculture, Washington, DC, United States.
Oswalt, S. N., Smith, W. B., Miles, P. D., & Pugh, S. A. (2014b). Forest Resources of the United
States, 2012. A Technical Document Supporting the Forest Service Update of the 2010
RPA Assessment.
Parish, E. S., Herzberger, A. J., Phifer, C. C., & Dale, V. H. (2018). “Transatlantic wood pellet
trade demonstrates telecoupled benefits.” Ecology and Society, 23(1), 1–15. https://doi.
org/10.5751/ES-09878-230128
266

References
Pebesma, E. J., de Jong, K., & Briggs, D. (2007). “Interactive visualization of uncertain
spatial and spatio-temporal data under different scenarios: An air quality example.”
International Journal of Geographical Information Science, 21(5), 515–527. https://doi.
org/10.1080/13658810601064009
Pedro-Monzonís, M., Solera, A., Ferrer, J., Estrela, T., & Paredes-Arquiola, J. (2015). “A review
of water scarcity and drought indexes in water resources planning and management.”
Journal of Hydrology, 527, 482–493. https://doi.org/10.1016/j.jhydrol.2015.05.003
Pelkmans, L., Goh, C. S., Junginger, H. M., Parhar, R., Bianco, E., Pellini, A., & Benedetti,
L. (2014). Impact of promotion mechanisms for advanced and low-iLUC biofuels on
biomass markets: Summary Report. IEA Bioenergy Task 40.
Pereira, F. F., Tursunov, M., & Uvo, C. B. (2013). “Towards the response of water balance
to sugarcane expansion in the Rio Grande Basin, Brazil.” Hydrology and Earth System
Sciences Discussions, 10(5), 5563–5603. https://doi.org/10.5194/hessd-10-5563-2013
Pfister, S., Bayer, P., Koehler, A., & Hellweg, S. (2011). “Projected water consumption in future
global agriculture: Scenarios and related impacts.” Science of the Total Environment,
409(20), 4206–4216. https://doi.org/10.1016/j.scitotenv.2011.07.019
Pimm, S. L., & Raven, P. (2000). “Biodiversity. Extinction by numbers.” Nature, 403(6772),
843–845. https://doi.org/10.1038/35002708
Pinchot Institute. (2013). The Transatlantic Trade in Wood for Energy : A Dialogue on
Sustainability Standards and Greenhouse Gas Emissions. Workshop proceedings.
Pinchot Institute, Washington, DC, United States.
Plantinga, A. J., & Birdsey, R. A. (1993). “Carbon fluxes resulting from U.S. private timberland
management.” Climatic Change, 23(1), 37–53. https://doi.org/10.1007/BF01092680
Portmann, F. T., Siebert, S., & Doll, P. (2010). “MIRCA2000—Global monthly irrigated and
rainfed crop areas around the year 2000: A new high‐resolution data set for agricultural &
and hydrological modeling.” Global Biogeochemical Cycles, 24, 221–231. https://doi.
org/10.1029/2008GB003435
Potter, K. M., & Woodall, C. W. (2014). “Does biodiversity make a difference? Relationships
between species richness, evolutionary diversity, and aboveground live tree biomass
across U.S. forests.” Forest Ecology and Management, 321, 117–129. https://doi.
org/10.1016/j.foreco.2013.06.026
Poyry. (2014). “The Risk of Indirect Wood Use Change.” Final Report, (July), 33.
Prestele, R., Alexander, P., Rounsevell, M. D. A., Arneth, A., Calvin, K., Doelman, J., … Verburg,
P. H. (2016). “Hotspots of uncertainty in land-use and land-cover change projections:
a global-scale model comparison.” Global Change Biology, 22(12), 3967–3983. https://
doi.org/10.1111/gcb.13337
Prestemon, J. P., & Abt, R. C. (2002). “The southern timber market to 2040.” Journal of
Forestry, 100(November), 16–23.
Prestemon, J. P., Wear, D. N., & Foster, M. O. (2015). “The Global Position of the U . S . Forest
Products Industry.” Technical Report SRS-204, (March), 24.
Price, S. J., Dorcas, M. E., Gallant, A. L., Klaver, R. W., & Willson, J. D. (2006). “Three decades
of urbanization: Estimating the impact of land-cover change on stream salamander
populations.” Biological Conservation, 133(4), 436–441. https://doi.org/10.1016/j.
biocon.2006.07.005
Prudhomme, C., Giuntoli, I., Robinson, E. L., Clark, D. B., Arnell, N. W., Dankers, R., … Wisser,
D. (2014). “Hydrological droughts in the 21st century, hotspots and uncertainties from
a global multimodel ensemble experiment.” Proceedings of the National Academy of
Sciences, 111, 1–6. https://doi.org/10.1073/pnas.1222473110
267

References
Redford, K. H. (2010). “The Empty of neotropical forest where the vegetation still appears
intact.” BioScience, 42(6), 412–422.
Riffell, S., Verschuyl, J., Miller, D., & Wigley, T. B. (2011a). “A meta-analysis of bird and
mammal response to short-rotation woody crops.” GCB Bioenergy, 3(4), 313–321.
https://doi.org/10.1111/j.1757-1707.2010.01089.x
Riffell, S., Verschuyl, J., Miller, D., & Wigley, T. B. (2011b). “Biofuel harvests, coarse woody
debris, and biodiversity - A meta-analysis.” Forest Ecology and Management, 261(4),
878–887. https://doi.org/10.1016/j.foreco.2010.12.021
Rodrigues, A. S. L., Andelman, S. J., Bakarr, M. I., Boitani, L., Brooks, T. M., Cowling, R. M.,
… Yan, X. (2004). “Effectiveness of the global protected area network in representing
species diversity.” Nature, 428(April), 9–12. https://doi.org/10.1038/nature02459.1.
Rondinini, C., Di Marco, M., Chiozza, F., Santulli, G., Baisero, D., Visconti, P., … Boitani,
L. (2011). “Global habitat suitability models of terrestrial mammals.” Philosophical
Transactions of the Royal Society B: Biological Sciences, 366(1578), 2633–2641. https://
doi.org/10.1098/rstb.2011.0113
RSPO. (2019). “Roundtable on Sustainable Palm Oil.” Retrieved from https://rspo.org/
Rupp, S. P., Bies, L., Glaser, A., Kowaleski, C., McCoy, T., Rentz, T., … Wigley, T. (2012). Effects
of Bioenergy Production on Wildlife and Wildlife Habitat. Wildlife Society Technical
Review 12-03.
Sala, O. E., Chapin, F. S., Armesto, J. J., Berlow, E., Bloomfield, J., Dirzo, R., … Wall16, D. H.
(2000). “Global Biodiversity Scenarios for the Year 2100.” Science, 287(5459), 1770–
1774. https://doi.org/10.1126/science.287.5459.1770
Sano, E. E., Rosa, R., Brito, J. L. S., & Ferreira, L. G. (2010). “Land cover mapping of the
tropical savanna region in Brazil.” Environmental Monitoring and Assessment, 166(1–
4), 113–124. https://doi.org/10.1007/s10661-009-0988-4
Scarpare, F. V., Hernandes, T. A. D., Ruiz-Corrêa, S. T., Kolln, O. T., Gava, G. J. D. C., Dos Santos,
L. N. S., & Victoria, R. L. (2016). “Sugarcane water footprint under different management
practices in Brazil: Tietê/Jacaré watershed assessment.” Journal of Cleaner Production,
112, 4576–4584. https://doi.org/10.1016/j.jclepro.2015.05.107
Scarpare, F. V., Hernandes, T. A. D., Ruiz-Corrêa, S. T., Picoli, M. C. A., Scanlon, B. R., Chagas,
M. F., … Cardoso, T. de F. (2016). “Sugarcane land use and water resources assessment
in the expansion area in Brazil.” Journal of Cleaner Production, 133, 1318–1327. https://
doi.org/10.1016/j.jclepro.2016.06.074
Schipper, A. M., Hilbers, J. P., Meijer, J. R., Antão, L. H., Benítez-López, A., de Jonge, M. M. J., …
Huijbregts, M. A. J. (2019). “Projecting terrestrial biodiversity intactness with GLOBIO
4.” Global Change Biology, (August), 1–12. https://doi.org/10.1111/gcb.14848
Schulze, E. D., Körner, C., Law, B. E., Haberl, H., & Luyssaert, S. (2012). “Large-scale
bioenergy from additional harvest of forest biomass is neither sustainable nor
greenhouse gas neutral.” GCB Bioenergy, 4(6), 611–616. https://doi.org/10.1111/j.1757-
1707.2012.01169.x
Searchinger, T., Heimlich, R., Houghton, R. A., Dong, F., Elobeid, A., Fabiosa, J., … Yu, T.
H. (2008). “Use of U.S. croplands for biofuels increases greenhouse gases through
emissions from land-use change.” Science, 319(5867), 1238–1240. https://doi.
org/10.1126/science.1151861
Shannon-Wiener, C. E., Weaver, W., & Weater, W. J. (1950). “The mathematical theory of
communication.” Physics Today, 3, 31–37.
Sheffield, J., & Wood, E. F. (2012). Drought: past problems and future scenarios. Routledge.
268

References
Siebert, S., & Döll, P. (2008). The Global Crop Water Model (GCWM): Documentation and
first results for irrigated crops. Institute of Physical Geography, University of Frankfurt
(Main).
Sierra, A. M., Vanderpoorten, A., Gradstein, S. R., Pereira, M. R., Bastos, C. J. P., & Zartman,
C. E. (2018). “Bryophytes of Jaú National Park (Amazonas, Brazil): Estimating species
detectability and richness in a lowland Amazonian megareserve.” The Bryologist,
121(4), 571–588. https://doi.org/10.1639/0007-2745-121.4.571
Simpson, E. H. (1949). “Measurement of Diversity.” Nature, 163(4148), 688.
Sinclair, A. . R. E. (2003). “The Role of Mammals As Ecosystem Landscapers.” Alces, 39,
161–176. https://doi.org/10.1017/S1431927607070213
Soares-Filho, B. S., Nepstad, D. C., Curran, L. M., Cerqueira, G. C., Garcia, R. A., Ramos, C.
A., … Schlesinger, P. (2006). “Modelling conservation in the Amazon basin.” Nature,
440(7083), 520–523. https://doi.org/10.1038/nature04389
Soliveres, S., Van Der Plas, F., Manning, P., Prati, D., Gossner, M. M., Renner, S. C., …
Allan, E. (2016). “Biodiversity at multiple trophic levels is needed for ecosystem
multifunctionality.” Nature, 536(7617), 456–459. https://doi.org/10.1038/nature19092
Souza, G. M., Victoria, R. L., Joly, C. A., & Verdade, L. M. (2015). Bioenergy & Sustainability:
bridging the gaps.
Steck, C. E., Bürgi, M., Coch, T., & Duelli, P. (2007). “Hotspots and richness pattern of
grasshopper species in cultural landscapes.” Biodiversity and Conservation, 16(7),
2075–2086. https://doi.org/10.1007/s10531-006-9089-7
Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M. M., Buchori, D., Erasmi, S., …
Tscharntke, T. (2007). “Tradeoffs between income, biodiversity, and ecosystem
functioning during tropical rainforest conversion and agroforestry intensification.”
Proceedings of the National Academy of Sciences of the United States of America, &
104(12), 4973–4978. https://doi.org/10.1073/pnas.0608409104
Stephens, S., & Wagner, M. (2007). “Forest plantations and biodiversity: a fresh perspective.”
Journal of Forestry, 105(September), 307–313.
Stephenson, A. L., & MacKay, D. J. (2014a). “Life Cycle Impacts of Biomass Electricity in
2020.” UK DECC Report, (July), 154.
Stephenson, A. L., & MacKay, D. J. (2014b). Life Cycle Impacts of Biomass Electricity in 2020.
Department of Energy & Climate Change. London.
Strassburg, B. B. N., Brooks, T., Feltran-Barbieri, R., Iribarrem, A., Crouzeilles, R., Loyola,
R., … Balmford, A. (2017). “Moment of truth for the Cerrado hotspot.” Nature Ecology
and Evolution, 1(4), 1–3. https://doi.org/10.1038/s41559-017-0099
Sutanudjaja, E. H., Van Beek, R., Wanders, N., Wada, Y., Bosmans, J. H. C., Drost, N., … Bierkens,
M. F. P. (2018). “PCR-GLOBWB 2: A 5 arcmin global hydrological and water resources
model.” Geoscientific Model Development, 11, 2429–2453. https://doi.org/10.5194/
gmd-11-2429-2018
Tallaksen, L. M., & Hisdal, H. (1997). “Regional analysis of extreme streamflow drought
duration and deficit volume.” In Proceedings of the Postojna, Slovenia, Conference,
September-October 1997 (p. 10). https://doi.org/10.1212/WNL.0b013e31823ed0a4
Tangdamrongsub, N., Ditmar, P. G., Steele-dunne, S. C., Gunter, B. C., & Sutanudjaja, E. H.
(2016). “Remote Sensing of Environment Assessing total water storage and identifying
fl ood events over Tonlé Sap basin in Cambodia using GRACE and MODIS satellite
observations combined with hydrological models.” Remote Sensing of Environment,
181, 162–173. https://doi.org/10.1016/j.rse.2016.03.030
269

References
Tarr, N. M., Rubino, M. J., & Costanza, J. K. (2016). “Projected gains and losses of wildlife
habitat from bioenergy-induced landscape change.” GCB Bioenergy, 9(5), 1–15. https://
doi.org/10.1111/gcbb.12383
Tarr, N. M., Rubino, M. J., Costanza, J. K., McKerrow, A. J., Collazo, J. A., & Abt, R. C. (2017).
“Projected gains and losses of wildlife habitat from bioenergy-induced landscape
change.” GCB Bioenergy, 9(5), 909–923. https://doi.org/10.1111/gcbb.12383
Taverna, K., Urban, D. L., & McDonald, R. I. (2005). “Modeling landscape vegetation pattern
in response to historic land-use: A hypothesis-driven approach for the North Carolina
Piedmont, USA.” Landscape Ecology, 20(6), 689–702. https://doi.org/10.1007/s10980-
004-5652-3
Taylor, K. E., Stouffer, R. J., & Meehl, G. A. (2012). “An overview of CMIP5 and the experiment
design.” Bulletin of the American Meteorological Society, 93(4), 485–498. https://doi.
org/10.1175/BAMS-D-11-00094.1
Thomas, C. D., Cameron, A., Green, R. E., Bakkenes, M., Beaumont, L. J., Collingham, Y. C.,
… Williams, S. E. (2004). “Extinction risk from climate change.” Nature, 427, 145–148.
Titeux, N., Henle, K., Mihoub, J. B., Regos, A., Geijzendorffer, I. R., Cramer, W., … Brotons, L.
(2016). “Biodiversity scenarios neglect future land-use changes.” Global Change Biology,
22(7), 2505–2515. https://doi.org/10.1111/gcb.13272
U.S. Fish and Wildlife Service. (1979). “Endangered Species Act of 1973.”
UNESCO. (2019). The United Nations World Water Development Report 2019.
UNICA. (2019). “Total ethanol production data.” Retrieved August 15, 2019, from http://
www.unicadata.com.br
United States Department of Agriculture. (1994). Major World Crop Areas and Climatic
Profiles. Agricultural Handbook No. 664. United States Department of Agriculture,
Washington, DC, United States.
US EIA. (2019). “Densified Biomass Fuel Report.”
US Geological Survey Gap Analysis Program (GAP). (2011). “National Land Cover, Version 2.”
USDA. (2013). “Summary Report: 2010 National Resources Inventory,” 162.
USDA Foreign Agricultural Service. (2019a). Biofuels Annual 2019. Global Agricultural
Information Network (GAIN) Report. Retrieved from http://gain.fas.usda.gov/Recent
GAIN Publications/Biofuels Annual_Ottawa_Canada_11-24-2014.pdf
USDA Foreign Agricultural Service. (2019b). Sugar Annual 2019. Sugar annual.
USDA FS. (2012). Future of America’s forests and Rangelands: Forest Service 2010
Resources Planning Act Assessment. General Technical Report WO-87. Washington, DC.
USDA FS. (2016a). “Forest Inventory and Analysis Glossary.”
USDA FS. (2016b). “Forest Inventory and Analysis Program (FIA) EVALIDator
webapplication.”
USDA FS. (2017a). “Forest inventory and analysis, FIA data and tools.”
USDA FS. (2017b). “Timber Product Output (TPO).”
USGS. (2011). “U.S. Geological Survey Gap Analysis Program - Land Cover Data v2.2.”
USGS. (2014). “Federal lands of the United States.”
Van Beek, L. P. H., & Bierkens, M. F. P. (2009). The Global Hydrological Model PCR-GLOBWB:
Conceptualization, Parameterization and Verification. Department of Physical
Geography Faculty of Earth Sciences Utrecht University, Netherlands.
van Beek, L. P. H., Wada, Y., & Bierkens, M. F. P. (2011). “Global monthly water stress: 1.
Water balance and water availability.” Water Resources Research, 47, 1–25. https://
doi.org/10.1029/2010WR009791
270

References
van der Hilst, F., Van Eijck, J., Verstegen, J., Diogo, V., Batidzirai, B., & Faaij, A. (2013).
“Impacts of Biofuel Production. Case studies: Mozambique, Argentina and Ukraine.
Final Report.”
van der Hilst, F., Hoefnagels, R., Junginger, M., Londo, M., Shen, L., & Wicke, B. (2018).
“Biomass Provision and Use, Sustainability Aspects.” Encyclopedia of Sustainability
Science and Technology, 1–30. https://doi.org/10.1007/978-1-4939-2493-6_1048-1
van der Hilst, F., Lesschen, J. P., Van Dam, J. M. C., Riksen, M., Verweij, P. A., Sanders, J. P. M.,
& Faaij, A. P. C. (2012). “Spatial variation of environmental impacts of regional biomass
chains.” Renewable and Sustainable Energy Reviews, 16(4), 2053–2069. https://doi.
org/10.1016/j.rser.2012.01.027
Van der Hilst, F., Verstegen, J. A., Karssenberg, D., & Faaij, A. P. C. (2012). “Spatiotemporal land
use modelling to assess land availability for energy crops - illustrated for Mozambique.”
GCB Bioenergy, 4(6), 859–874. https://doi.org/10.1111/j.1757-1707.2011.01147.x
van der Hilst, F., Verstegen, J. A., & Woltjer, G. (2018). “Mapping direct and indirect land use
changes resulting from biofuel production and the effect of LUC mitigation measures.”
Global Change Biology, 1–21.
van der Hilst, F., Verstegen, J. a., Zheliezna, T., Drozdova, O., & Faaij, A. P. C. (2014).
“Integrated spatiotemporal modelling of bioenergy production potentials, agricultural
land use, and related GHG balances; demonstrated for Ukraine.” Biofuels, Bioproducts
and Biorefining, 8, 391–411. https://doi.org/10.1002/bbb.1471
van Vuuren, D. P., Edmonds, J., Kainuma, M., Riahi, K., Thomson, A., Hibbard, K., … Rose, S.
K. (2011). “The representative concentration pathways: An overview.” Climatic Change,
109, 5–31. https://doi.org/10.1007/s10584-011-0148-z
van Vuuren, D. P., Kriegler, E., O’Neill, B. C., Ebi, K. L., Riahi, K., Carter, T. R., … Winkler,
H. (2014). “A new scenario framework for Climate Change Research: Scenario matrix &
architecture.” Climatic Change, 122(3), 373–386. https://doi.org/10.1007/s10584-013-
0906-1
Verdade, L. M., Gheler-Costa, C., Penteado, M., & Dotta, G. (2012). “The Impacts of Sugarcane
Expansion on Wildlife in the State of Sao Paulo, Brazil.” Journal of Sustainable Bioenergy
Systems, 02(04), 138–144. https://doi.org/10.4236/jsbs.2012.24020
Verschuyl, J., Riffell, S., Miller, D., & Wigley, T. B. (2011). “Biodiversity response to intensive
biomass production from forest thinning in North American forests - A meta-analysis.”
Forest Ecology and Management, 261(2), 221–232. https://doi.org/10.1016/j.
foreco.2010.10.010
Verstegen, J. A., Karssenberg, D., van der Hilst, F., & Faaij, A. (2012). “Spatio-temporal
uncertainty in Spatial Decision Support Systems: A case study of changing land
availability for bioenergy crops in Mozambique.” Computers, Environment and Urban
Systems, 36(1), 30–42. https://doi.org/10.1016/j.compenvurbsys.2011.08.003
Verstegen, J. A., Karssenberg, D., van der Hilst, F., & Faaij, A. P. C. (2014). “Identifying a
land use change cellular automaton by Bayesian data assimilation.” Environmental
Modelling and Software, 53, 121–136. https://doi.org/10.1016/j.envsoft.2013.11.009
Verstegen, J. A., Karssenberg, D., van der Hilst, F., & Faaij, A. P. C. (2016). “Detecting systemic
change in a land use system by Bayesian data assimilation.” Environmental Modelling
and Software, 75, 424–438. https://doi.org/10.1016/j.envsoft.2015.02.013
Verstegen, J. A., van der Hilst, F., Woltjer, G., Karssenberg, D., de Jong, S. M., & Faaij, A. P. C.
(2016). “What can and can’t we say about indirect land-use change in Brazil using an
integrated economic - land-use change model?” GCB Bioenergy, 8(3), 561–578. https://
doi.org/10.1111/gcbb.12270
271

References
Visconti, P., Pressey, R. L., Giorgini, D., Maiorano, L., Bakkenes, M., Boitani, L., … Rondinini, C.
(2011). “Future hotspots of terrestrial mammal loss.” Philosophical Transactions of the
Royal Society B: Biological Sciences, 366(1578), 2693–2702. https://doi.org/10.1098/
rstb.2011.0105
Wada, Y., van Beek, L. P. H., & Bierkens, M. F. P. (2011). “Modelling global water stress of
the recent past: on the relative importance of trends in water demand and climate
variability.” Hydrology and Earth System Sciences, 15, 3785–3808. https://doi.
org/10.5194/hess-15-3785-2011
Wada, Y., Van Beek, L. P. H., Viviroli, D., Drr, H. H., Weingartner, R., & Bierkens, M. F. P. (2011).
“Global monthly water stress: 2. Water demand and severity of water stress.” Water
Resources Research, 47, 1–17. https://doi.org/10.1029/2010WR009792
Wada, Y., Wisser, D., & Bierkens, M. F. P. (2014). “Global modeling of withdrawal, allocation
and consumptive use of surface water and groundwater resources.” Earth System
Dynamics, 5, 15–40. https://doi.org/10.5194/esd-5-15-2014
Walter, A., Galdos, M. V., Scarpare, F. V., Leal, M. R. L. V., Seabra, J. E. A., Da Cunha, M. P.,
… De Oliveira, C. O. F. (2014). “Brazilian Sugarcane Ethanol: Developments so far and
Challenges for the Future.” Advances in Bioenergy: The Sustainability Challenge,
3(February), 373–394. https://doi.org/10.1002/9781118957844.ch24
Wanders, N., & Van Lanen, H. A. J. (2015). “Future discharge drought across climate regions
around the world modelled with a synthetic hydrological modelling approach forced
by three general circulation models.” Natural Hazards and Earth System Sciences, 15,
487–504. https://doi.org/10.5194/nhess-15-487-2015
Wanders, N., & Wada, Y. (2015). “Human and climate impacts on the 21st century
hydrological drought.” Journal of Hydrology, 526, 208–220. https://doi.org/10.1016/j.
jhydrol.2014.10.047
Wanders, N., Wada, Y., & Van Lanen, H. A. J. (2015). “Global hydrological droughts in the
21st century under a changing hydrological regime.” Earth System Dynamics, 6(1),
1–15. https://doi.org/10.5194/esd-6-1-2015
Wang, W., Dwivedi, P., Abt, R. C., & Khanna, M. (2015a). “Carbon savings with transatlantic
trade in pellets: accounting for market-driven effects.” Environmental Research Letters,
10, 1–13. https://doi.org/10.1088/1748-9326/10/11/114019
Wang, W., Dwivedi, P., Abt, R. C., & Khanna, M. (2015b). “Carbon savings with transatlantic
trade in pellets: Accounting for market-driven effects.” Environmental Research
Letters, 10(11), 114019. https://doi.org/10.1088/1748-9326/10/11/114019
Warszawski, L., Frieler, K., Huber, V., Piontek, F., Serdeczny, O., & Schewe, J. (2014). “The
Inter-Sectoral Impact Model Intercomparison Project (ISI–MIP): Project framework.”
Proceedings of the National Academy of Sciences, 111(9), 3228–3232. https://doi.
org/10.1073/pnas.1312330110
Watkins, D. W., de Moraes, M. M. G. A., Asbjornsen, H., Mayer, A. S., Licata, J., Lopez, J. G.,
… da Nobrega Germano, B. (2015). “Bioenergy Development Policy and Practice Must
Recognize Potential Hydrologic Impacts: Lessons from the Americas.” Environmental
Management, 56(6), 1295–1314. https://doi.org/10.1007/s00267-015-0460-x
Wear, D. N. (2011). “Forecasts of County-Level Land Uses Under Three Future Scenarios.”
A Technical Document Supporting the Forest Service 2010 RPA Assessment, 41.
Wear, D. N., & Greis, J. G. (2013a). The southern Forest Futures Project: Technical report.
Wear, D. N., & Greis, J. G. (2013b). The Southern Forest Futures Project. Technical report.
USDA Forest Service, Southern Research Station. Asheville, NC, United States.
272

References
Wicke, B., van der Hilst, F., Daioglou, V., Banse, M., Beringer, T., Gerssen-Gondelach, S., …
Faaij, A. P. C. (2015). “Model collaboration for the improved assessment of biomass
supply, demand, and impacts.” GCB Bioenergy, 7(3), 422–437. https://doi.org/10.1111/
gcbb.12176
Wicke, B., Verweij, P., Van Meijl, H., Van Vuuren, D. P., & Faaij, A. P. C. (2012). “Indirect land
use change: Review of existing models and strategies for mitigation.” Biofuels, 3(1),
87–100. https://doi.org/10.4155/bfs.11.154
Williams, P. H., & Gaston, K. J. (1994). “Measuring more of Biodiversity: Can Higher-taxon
Richness Predict Wholesale Species Richness?” Biological Conservation, 67(1994),
211–217.
WWF. (2012). Living Planet Report 2012. https://doi.org/10.1159/000255356
Yassa, S. (2015). “Think Wood Pellets are Green? Think Again.” NRDC Issue Brief, (may), 1–5.
Yevjevich, V. (1967). “An Objective Approach to Definitions and Investigations of Continental
Hydrologic Droughts.” Hydrology Papers, 23, 1967.
Young, T., Wang, Y., Guess, F., Fly, M., Hodges, D., & Poudyal, N. (2015). “Understanding
the Characteristics of Non-industrial Private Forest Landowners Who Harvest Trees.”
Small-Scale Forestry, 14(3), 273–285. https://doi.org/10.1007/s11842-015-9287-9
Zavaleta, E. S., Pasari, J. R., Hulvey, K. B., & Tilman, D. G. (2010a). “Sustaining multiple
ecosystem functions in grassland communities requires higher biodiversity.”
Proceedings of the National Academy of Sciences of the United States of America, 107(4),
1443–1446. https://doi.org/10.1073/pnas.0906829107
Zavaleta, E. S., Pasari, J. R., Hulvey, K. B., & Tilman, G. D. (2010b). “Sustaining multiple
ecosystem functions in grassland communities requires higher biodiversity.”
Proceedings of the National Academy of Sciences, 107(4), 1443–1446. https://doi.
org/10.1073/pnas.0906829107 &
Zhu, Z., Bergamaschi, B., Bernknopf, R., & Clow, D. (2010). Public Review Draft: A Method
for Assessing Carbon Stocks, Carbon Sequestration, and Greenhouse-Gas Fluxes
in Ecosystems of the United States Under Present Conditions and Future Scenario.
Scientific Investigations Report 2010–5233, DRAFT.
Zuurbier, P. (2008). Sugarcane ethanol: contributions to climate change mitigation and the
environment (1st ed). Wageningen Academic Publishers: Wageningen, the Netherlands.
https://doi.org/10.3920/978-90-8686-652-6
273

Anna Duden - Environmental Impacts of Bioenergy - Ex-Ante Spatially-Explicit Assessment of Biodiversity and Hydrology Impacts of Increase

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Anna Duden - Environmental Impacts of Bioenergy - Ex-Ante Spatially-Explicit Assessment of Biodiversity and Hydrology Impacts of Increase

Caricato da

Copyright:

Formati disponibili

Environmental impacts

Anna Sarah Duden

AnnaDuden_BNW.indd 1 16-03-20 12:26

© A.S. Duden 2020

AnnaDuden_BNW.indd 2 16-03-20 12:26

Ex-ante ruimtelijk expliciete analyse van de

(met een samenvatting in het Nederlands)

Ter verkrijging van de graad van doctor

AnnaDuden_BNW.indd 3 16-03-20 12:26

AnnaDuden_BNW.indd 4 16-03-20 12:26

Chapter 2 Modelling the impacts of wood pellet demand on forest 25

Chapter 3 Impact of increased wood pellet demand on biodiversity 55

Chapter 4 Biodiversity impacts of increased ethanol production in 85

Chapter 5 Hydrological impacts of ethanol-driven sugarcane 109

Chapter 6 Synthesis and conclusions 127

Nederlandse samenvatting 274

About the author 281

AnnaDuden_BNW.indd 5 16-03-20 12:26

AID Area In Drought

AnnaDuden_BNW.indd 6 16-03-20 12:26

AnnaDuden_BNW.indd 7 16-03-20 12:26

AnnaDuden_BNW.indd 8 16-03-20 12:26

AnnaDuden_BNW.indd 9 16-03-20 12:26

1.1 Risks and benefits of increased use of bioenergy

Global consumption of energy is projected to increase by over 25% by 2040 (IEA,

AnnaDuden_BNW.indd 10 16-03-20 12:26

Another type of environmental impact relates to the effects of land-use change on

As a result of these sustainability concerns, a range of sustainability guidelines for

AnnaDuden_BNW.indd 11 16-03-20 12:26

1.2 Quantification of environmental impacts

Increased bioenergy demand puts additional pressure on natural resources already

Apart from direct impacts of expansion of bioenergy production, indirect impacts

AnnaDuden_BNW.indd 12 16-03-20 12:26

resources varies strongly geographically. Availability of fresh water for instance, is

AnnaDuden_BNW.indd 13 16-03-20 12:26

1.3 Aim and research questions

I. How can environmental impacts of increased bioenergy demand be

This thesis focuses on an ex-ante assessment of increased bioenergy demand on

AnnaDuden_BNW.indd 14 16-03-20 12:26

1.4.1 Wood pellet production in the southeastern United States

The southeastern US predominantly produces industrial wood pellets, which are

AnnaDuden_BNW.indd 15 16-03-20 12:26

Urbanization has been projected to be the largest driver of deforestation in the

AnnaDuden_BNW.indd 16 16-03-20 12:26

AnnaDuden_BNW.indd 17 16-03-20 12:26

inform US conservation managers and policy makers (Fleishman, 2011). A number of

1.4.2 Ethanol production in Brazil

AnnaDuden_BNW.indd 18 16-03-20 12:26

As a result, Brazil is projected to show the strongest growth in biofuel production

The projected increase in ethanol production is expected to be realized through

AnnaDuden_BNW.indd 19 16-03-20 12:26

AnnaDuden_BNW.indd 20 16-03-20 12:26

1.5 Thesis outline

The research questions described in paragraph 1.3 are addressed in chapters 2 to 5.

Chapter 2 uses a combined modelling approach to assess land-use changes in the

AnnaDuden_BNW.indd 21 16-03-20 12:26

AnnaDuden_BNW.indd 22 16-03-20 12:26

AnnaDuden_BNW.indd 23 16-03-20 12:26

AnnaDuden_BNW.indd 24 16-03-20 12:26

and hydrology impacts of

AnnaDuden_BNW.indd 25 16-03-20 12:26

AnnaDuden_BNW.indd 26 16-03-20 12:27

1 Natural timberland is defined as ‘Productive forests composed of trees established by natural