Soil Biology & Biochemistry 70 (2014) 22e32

Contents lists available at ScienceDirect

Soil Biology & Biochemistry

journal homepage: www.elsevier.com/locate/soilbio

Response of osmolytes in soil to drying and rewetting

Charles R. Warren*
School of Biological Sciences, The University of Sydney, NSW 2006, Australia

a r t i c l e i n f o a b s t r a c t

Article history: The accumulation and subsequent release of microbial osmolytes in response to drying and rewetting are
Received 19 October 2013 thought to be key players in C and N dynamics, yet studies on soils have failed to support this hypothesis.
Received in revised form The aim of this experiment was to determine how low-molecular weight compounds, and osmolytes in
2 December 2013
particular, are affected by drying and rewetting. Water decits were imposed slowly by withholding
Accepted 13 December 2013
water for 21 weeks from large (200 L) mesocosms vegetated with a globally widespread grass Themeda
Available online 22 December 2013
triandra. A broad spectrum of small molecules in extracts was identied and quantied by capillary
electrophoresisemass spectrometry and gas chromatographyemass spectrometry. Compared with
Keywords:
Drought
controls, drought-stressed mesocosms contained >10-fold larger amounts of known microbial osmo-
Water decits lytes: ectoine, hydroxyectoine, betaine, proline-betaine, trigonelline, proline, trehalose, arabitol. The pool
Osmolyte of osmolytes accounted for 3.6% of CHCl3 labile TOC in control mesocosms and 17% of CHCl3 labile TOC in
Osmotic adjustment drought-stressed mesocosms. There was no evidence that rewatering led to a large pulse of osmolytes in
Mass spectrometry free solution. Instead osmolytes decreased to control concentrations within 1e3 h of rewatering e
Birch effect probably indicating rapid uptake by microbes and plants. Results of this study suggest that osmolytes can
Rewatering account for a substantial fraction of microbial C, and are at least one of the ways that soil microbes cope
Grassland
with water decits.
2013 Elsevier Ltd. All rights reserved.

1. Introduction reduce microbial activity of soils while rewetting commonly in-

Water decits are a common feature of large tracts of the
terrestrial biosphere. Some biomes are regularly exposed to water
decits (e.g. arid, semi-arid and Mediterranean), while prolonged
periods of below-average precipitation may lead to drought in bi-
omes that do not normally experience water decits. For example,
recent years have seen prolonged periods of below-average pre-
cipitation give rise to several sub-continental scale droughts:
central/south-west Asia (1998e2003), western North America
(1999e2007), Australia (2002e2003), Europe (2003) and Ama-
zonia (2005) (Cook et al., 2004; Trenberth et al., 2007; Thomas
et al., 2009). There is evidence to suggest that the geographic
area affected by droughts has increased in the past four decades
(Dai et al., 2004) and climate change projections suggest decreased
precipitation and increased occurrence and/or severity of drought
in most sub-tropical and mid-latitude regions of the globe (Meehl
et al., 2007).
One globally important consequence of water decits is their
effect on key soil processes such as C and N mineralization. For
example, it has been known for several decades that water decits
* Tel.: 61 2 9351 2678; fax: 61 2 9351 4119.
E-mail address: charles.warren@sydney.edu.au.
creases microbial activity and leads to a pulse of C and N mineral-
ization (e.g. Birch, 1958, 1964; Bottner, 1985). The mechanisms
underpinning the response of soil to drying and rewetting remain
controversial (Borken and Matzner, 2009), but almost certainly
involve processes altered at the molecular scale (Schimel et al.,
2007; Boot et al., 2013).
One mechanism that may at least partially underpin responses
to water decits and rewetting is the synthesis and metabolism of
low-molecular weight organic solutes by soil microbes. In response
to decreasing water potentials, soil microbes lower their solute
potential by synthesizing or importing organic solutes (Lippert and
Galinski, 1992; Kempf and Bremer, 1998a,b; Halverson et al., 2000;
Hasegawa et al., 2000; Wood et al., 2001). Typically organic solutes
accumulate in the cytosol to counter concentrations of inorganic
ions in vacuoles (thus contributing to osmotic balance between
compartments, and overall turgor). The organic solutes that accu-
mulate are termed compatible solutes, or osmolytes, and tend to be
low-molecular weight compounds that do not interfere with
metabolism even at very high concentrations. Osmolytes include C-
and N-containing compounds such as some sugars and sugar al-
cohols (e.g. trehalose and arabitol), quaternary ammonium com-
pounds (e.g. betaine) and pyrimidine derivatives (e.g. ectoine)
(Csonka, 1989; Lippert and Galinski, 1992; Hasegawa et al., 2000;
Wood et al., 2001). At least two mechanisms may underpin how

0038-0717/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soilbio.2013.12.008
 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
the rapid increase in water potential that follows rewetting of dried
soils leads to a ush of C- and N-containing substrates and meta-
bolism. First, rewetting of dried soils leads to hydration and lysis of
dead microbial cells that accumulated during the drying period.
Lysed cells may subsequently serve as substrates for those microbes
that survived (Kieft et al., 1987; Wu and Brookes, 2005; Borken and
Matzner, 2009). Second, rewetting pose a major stress for those
microbes that survived the drying cycle (Schimel et al., 2007). The
stress arises because at the end of a drying cycle the surviving soil
microbes have a strongly negative solute potential due to accu-
mulated osmolytes, and thus microbes need to dispose of accu-
mulated intracellular osmolytes so as to avoid uncontrolled inux
of water (Kieft et al., 1987; Csonka, 1989; Schimel et al., 2007;
Borken and Matzner, 2009).
Culture-based and modeling studies suggest accumulation and
subsequent release of osmolytes in response to drying and rewet-
ting cycles are key players in C and N dynamics (Schimel et al.,
2007), yet there is little empirical support from studies with soil.
If microbes use osmolytes as constitutive or inducible defenses
against water decits one would predict that osmolytes would
comprise a substantial proportion of microbial biomass in dry soil
(Boot et al., 2013) and rewetting would lead to a ush of osmolytes
in the extracellular matrix (i.e. soil solution) (Williams and Xia,
2009). At present there is little support for these predictions with
recent studies failing to nd large constitutive or inducible
amounts of osmolytes in a eld experiment on seasonally dry
grassland soil (Boot et al., 2013) or forest soil (Gransson et al.,
2013) or soils exposed to laboratory water stress treatments
(Williams and Xia, 2009; Kakumanu et al., 2013). Partial support
comes from a study of soil extracts from seasonally dry eld sites
that reported abundant osmolytes in ve out of seven sites
(Warren, 2013b). However, limited conclusions could be drawn
from the latter experiment because it did not examine seasonal
variation in osmolytes or manipulate water availability.
The inconsistency of results may reect true biological differ-
ences, but a proportion can probably be explained by varying du-
rations of water stress (Borken and Matzner, 2009), and the
comprehensiveness with which osmolytes were proled (Warren,
2013a). For example, studies that impose water stress for a short
periods (e.g. 4 days: Williams and Xia, 2009) may be too rapid for
prompting signicant osmolyte accumulation (Turner, 1986) and
would not encompass changes in the microbial response to rewa-
tering that occur only after prolonged drought (Meisner et al.,
2013). A diversity of different C- and N-containing molecules can
be accumulated as osmolytes (Csonka, 1989; Lippert and Galinski,
1992; Hasegawa et al., 2000; Wood et al., 2001), and thus studies
that quantify only N-containing osmolytes (Boot et al., 2013;
Warren, 2013b) may fail to see quantitatively signicant changes
in C-based osmolytes (e.g. sugars and sugar alcohols) while data on
compound classes (e.g. monosaccharides and amino acids,
Gransson et al., 2013) are difcult to interpret in terms of osmolyte
accumulation because the assays do not distinguish osmolytes from
the large background of non-osmolytes.
The aim of this experiment was to determine how low-molec-
ular weight osmolytes are affected by water decit and rewetting.
Mesocosms were lled with soil and seedlings from Themeda tri-
andra Forssk. grassland. The response of T. triandra grassland to
drying and rewetting may be globally signicant because it is one of
the most widespread grasses in grassland ecosystems of Africa, Asia
and Australia and is regularly exposed to water decits (DellAcqua
et al., 2013). Moreover, climate change projections suggest the
future will see increased frequency and severity of water decits
across much of the species range (Meehl et al., 2007). To assess the
signicance of osmolyte accumulation in responses to water de-
cits and rewetting, I tested a) if osmolytes were a larger fraction of
23
microbial C in mesocosms from which water had been withheld
than control mesocosms kept well watered, and b) if rewetting led
to a large pulse of osmolytes in free solution. Water decits were
imposed slowly by withholding water from large (200 L) meso-
cosms so as to mimic the situation in nature and allow sufcient
time for osmolytes to accumulate. A broad spectrum of C- and N-
containing osmolytes was quantied by gas chromatographyemass
spectrometry (Roessner et al., 2000; Warren et al., 2012) and
capillary electrophoresisemass spectrometry (Warren, 2013a),
while measurements of CO2 efux from the soil surface was used as
an index of microbial activity and to place the putative ush of
osmolytes in the context of the large ush of CO2 efux induced by
rewetting of dried soils (Birch, 1958, 1964; Jarvis et al., 2007).
2. Materials and methods
2.1. Chemicals
Methanol, acetonitrile and formic acid were LC/MS (Optima)
grade from Fisher Chemical (Scoresby, Vic, Australia). Ammonium
formate (Acros Organics, Geel, Belgium), ammonium hydroxide
(28e30% NH3, Sigma, Sydney, Australia), potassium sulfate (Sigma),
methoxyamine hydrochloride (Sigma) and iodomethane (Sigma)
were analytical grade, while pyridine, chloroform, and N-Methyl-N-
triuoroacetamide (MSTFA) with 1% trimethylchlorosilane (TMCS)
were derivatization or GC grade.
All electrolytes, rinsing solutions, standards and samples were
prepared with 18.2 MU cm resistivity ultra-pure water (Arium,
Sartorius, Goettingen, Germany). Approximately 140 standards
(comprising organic N monomers, small carbohydrates, and
organic acids) were prepared from their free acids or salts pur-
chased from Sigma. a-N-methyl-histidine, a-N,N-dimethyl-histi-
dine and N-methyl-proline were from Chem-Impex (Chem-Impex
International, Wood Dale, IL, USA). All standards of chiral amino
acids were L enantiomers, while carbohydrates were D enantio-
mers. Hercynine (Na,Na,Na-trimethyl-L-histidine) was synthesized
according to Reinhold et al. (1968), as described recently (Warren,
2013b).
2.2. Soil mesocosms
In June 2009 eight replicate mesocosms (painted steel drums,
572 mm diameter, 851 mm high) were lled with loam soil
collected from A1 and A2 horizons of T. triandra grassland in
western Sydney (34.0 S, 150.6 E, 75 m above sea-level). The intact
soil was an abruptic lixisol and chemical properties have been
described recently (Warren, 2013b). After collecting in the eld, soil
was sieved to 4 mm, mixed, and then mesocosms were lled with
200 L of soil at approximately the bulk density of eld soil. Cali-
brated soil moisture probes (Hobo EC-5 Soil Moisture Smart Sensor,
Onset Corp, Pocasset, MA, USA) were installed horizontally at a
depth of 15 cm in four mesocosms (two control and two drought
treatment). Volumetric water content was recorded every 30 min
and stored on a datalogger (Hobo). In November 2009 mesocosms
were planted with six-month-old seedlings of two perennial native
grasses T. triandra and Microlaena stipoides. Mesocosms were held
within a sunlit polythene-covered greenhouse that transmitted
around 70% of sunlight. A ventilation system ensured that air
within the greenhouse was well mixed and exchanged with
external air, while a thermostated cooling system maintained
maximum temperatures 5  C above ambient from May to October
and at ambient temperature from October to May (Fig. 1). Meso-
cosms were watered every 5e15 days so as to avoid development of
water stress. When treatments commenced in September 2012 the
above-ground dry mass of plants varied between 1000 and
24 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
Fig. 1. Maximum and minimum air temperatures (a) within the greenhouse that housed mesocosms, and volumetric soil water content at a depth of 15 cm (b) of two control and
two treatment (droughted) mesocosms. Measurements were made immediately before rewatering (T0) until 28 days after rewatering (T28).
1500 g m2, with approximately 90% of the dry mass accounted for
by T. triandra, 5e10% by M. stipoides and less than 5% by herbaceous
species.
2.3. Treatments and experimental design
In the Austral spring (September 2012), four mesocosms were
assigned to a control treatment that continued to receive frequent
watering to near eld capacity while another four mesocosms were
assigned to a water stress treatment. To mimic the cycles of
increasingly severe water stress that can occur in water limited
habitats, the water stress treatment involved ve dryingerewa-
tering cycles of increasing duration (Fig. 1). The rst three dryinge
rewatering cycles involved withholding water for three weeks, the
next drying cycle involved withholding water six weeks, while the
nal drying cycle was the most severe and involved withholding
water for 21 weeks. The nal drying cycle coincided with hot
summer and autumn weather. When rewatering mesocosms, water
was applied slowly to allow inltration of water.
2.4. Soil CO2 efux
One soil respiration collar (203 mm diameter PVC pipe) was
installed in each of two control and two treatment (droughted)
mesocosms. Collars were positioned in-between swards of the
grasses, and thus measured CO2 efux does not include any direct
contribution from respiration of above-ground plant parts. CO2
efux was measured using a soil respiration system (LI-8100, LI-Cor
Inc., Lincoln, Nebraska, U.S.A.) and 20-cm survey chamber (8100-
103). For the four measured mesocosms CO2 efux was measured
between 10AM and 2PM several days before mesocosms were
rewatered, immediately after rewatering, and then periodically for
the next 28 days. In addition, CO2 efux of one drought-stressed
mesocosm was measured every 30 min for ten days after rewa-
tering so as to obtain information on diel variation. Respiration
measurements were made using normal protocols for the LI-8100
instrument, viz., maximum ow rate, 120-s chamber closure,
averaging of three replicate measurements.
2.5. Extraction of soil samples
For each of four droughted and four control mesocosms H2O
extracts, K2SO4 extracts and combined CHCl3 K2SO4 extracts were
made at the conclusion of the nal drying cycle (i.e. immediately
before rewatering) and 28 days after rewatering. To assess if
rewatering led to a pulse of osmolytes in free solution, additional
H2O extracts were made on droughted mesocosms 1 h, 3 h, 1 day, 3
days, 7 days, 21 days after rewatering. For these time-course H2O
extracts parallel measurements on control mesocosms were not
necessary because solute concentrations in control mesocosms
varied by less than 20% among days (e.g. see consistency of solute
concentrations between days 0 and 28 in control mesocosms,
Fig. 6). Soil samples (0e10 cm depth) were collected with a 2.5 cm
diameter corer. To minimize artefacts associated with soil distur-
bance from repeated sampling, all soil samples were at least 15 cm
away from previous soil cores. In all cases soil samples were
collected between 10AM and 2PM so as to minimize impacts of
possible diel variation. Soil samples were immediately transported
to the laboratory and extracted within 10 min of collection.
Soil samples were extracted with ultra-pure water (4.0 g FW
soil: 20.0 mL ultra-pure water) by shaking end-to-end at 100 rpm
for 10 min, centrifuging (3200g, 10 min, 20  C) and then trans-
ferring the supernatant to a clean container and immediately
freezing at 80  C. Blanks (20.0 mL ultra-pure H2O) were carried
through the same extraction procedures and subsequently
analyzed alongside samples. Samples were extracted for 10 min
rather than the more usual longer extractions (60e120 min) so as to
minimize metabolism during extraction (Rousk and Jones, 2010).
K2SO4 and combined CHCl3 K2SO4 extracts were made at the
conclusion of the nal drying cycle and 28 days after rewatering so
as to obtain exchangeable and CHCl3 extractable fractions. Chlo-
roform extraction and K2SO4 extraction were combined used the
 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
so-called direct extraction method (Setia et al., 2012; Boot et al.,
2013). The chloroform direct extraction method was used rather
than gas fumigation for two reasons: 1) direct extraction minimizes
the problem of post-collection metabolism of soils compared with
what would occur during 1e4 day CHCl3 gas fumigation; and 2)
with direct extraction the microbial extraction efciency should be
equivalent for samples irrespective of their water content whereas
microbial extraction efciencies with CHCl3 gas fumigation can be
confounded by differences in soil water content due to their effect
on gaseous diffusion. One sub-sample of soil was extracted with
0.5 M K2SO4 (4.0 g FW soil: 20.0 mL K2SO4), while the other was
extracted with 0.5 M K2SO4 that contained CHCl3 (4.0 g FW soil:
20.0 mL K2SO4 with 1.5% CHCl3). K2SO4 (1.5% CHCl3) extracts were
shaken end-to-end at 100 rpm for 30 min, centrifuged (3200g,
10 min, 20  C) and then ltered through Whatman #1 lter paper
and then frozen. Blanks (20.0 mL K2SO4, and 20.0 mL K2SO4 with
1.5% CHCl3) were carried through the same extraction procedures
and subsequently analyzed alongside samples. Initial experiments
established that direct extraction extracted 60e80% of the organic C
that was extracted by CHCl3 gas fumigation for 2 days. No correc-
tions were made for the discrepancy between direct extraction and
CHCl3 gas fumigation methods. Similarly, no corrections were made
for extraction (fumigation) efciency.
2.6. Capillary electrophoresisemass spectrometry of organic N
monomers
Capillary electrophoresisemass spectrometry (CEeMS) was
used for untargeted proling of organic N monomers in soil ex-
tracts, essentially as described previously (Warren, 2013a). CEeMS
was performed with a capillary electrophoresis system (P/ACE
MDQ, BeckmaneCoulter, Fullerton, CA, USA) equipped with a bare
fused silica capillary (50 mm i.d.  100 cm long) interfaced via a co-
axial sheath-ow sprayer (G1607A, Agilent, Waldbronn, Germany)
to an ion trap mass spectrometer (AmaZon SL, Bruker Daltonics,
Bremen, Germany). Sheath liquid of 50% (v/v) methanol with 0.1%
(v/v) formic acid was delivered at 4 mL min1 by a syringe pump
(NE-1002X Microuidics Syringe Pump, New Era Pump Systems,
Farmingdale, NY, USA) driving a 10-mL PTFE-tipped gas tight sy-
ringe (SGE, Ringwood, Vic, Australia). Ion source parameters were
as described previously (Warren, 2013a). Water extracts were
concentrated 10-fold while K2SO4 and K2SO4 CHCl3 extracts were
concentrated 1.5-fold by evaporating under reduced pressure
(Vacufuge, Eppendorf). Samples were made up in 100 mM
ammonium formate (pH 10) in 25% (v/v) acetonitrile that contained
an internal standard (0.4 mg mL1 methionine sulfone). Samples
were injected at 3 psi for 30 s and separated with an electrolyte of
2 M formic acid with 20% (v/v) methanol under 30 kV positive
polarity. The mass spectrometer was set to a scan a range of 50e250
m/z in enhanced resolution mode (8100 u/s) and data were recor-
ded as the average of ve scans. Between runs the capillary was
ushed with electrolyte for 10 min (50 psi). Compounds were
identied and quantied based on comparison of migration times,
[M H], MS2 and (for some compounds) MS3 with 63 authentic
standards run under the same conditions on the same instrument,
as described previously (Warren, 2013a).
2.7. Gas chromatographyemass spectrometry
Methoximated TMS derivatives were prepared essentially as
described previously (Lisec et al., 2006). A 5 mL aliquot of
0.02 mg mL1 ribitol (internal standard) was added to 1000 mL of
H2O extract or 200 mL of K2SO4 extract (CHCl3) and dried under
reduced pressure (Vacufuge, Eppendorf). To derivatise samples,
40 mL of methoxyamination reagent (20 mg mL1 methoxyamine
25
hydrochloride in pyridine) was added and tubes were incubated for
90 min at 37  C in a shaking incubator (400 rpm), then 70 mL of N-
Methyl-N-triuoroacetamide (MSTFA) with 1% trimethyl-
chlorosilane (TMCS) was added and tubes were incubated for
30 min at 37  C in a shaking incubator (400 rpm). A 1 mL sample was
injected splitless into an injection port liner (single gooseneck
Siltek-treated, Restek, Bellefonte, PA, USA) at 250  C and separated
by capillary gas chromatography on an arylene-modied 5%
diphenyle95% dimethyl polysiloxane stationary phase (30 m
long  0.25 mm ID  0.25 mm lm thickness with a 10-m guard
column; Rxi-5SilMS, Restek, Bellefonte, USA). The column was
held at 70  C for 2 min, raised to 330  C at 6  C min1, and then held
at 330  C for 10 min. Helium (99.999%, BOC, North Ryde, NSW,
Australia) was used as the carrier gas at a constant ow of
1 mL min1. The transfer line was held at 280  C and the ion source
at 250  C. The column eluent was ionized by electron impact
(70 eV) and mass spectra were collected from 70 to 600 amu at 6.67
scans s1 (GCeMS-QP2010Plus, Shimadzu, Kyoto, Japan). To iden-
tify methoximated TMS metabolites, chromatograms were expor-
ted from the proprietary Shimadzu format to netCDF and
deconvoluted (AnalyzerPro, Spectralworks Ltd., Runcorn, UK).
Metabolites were identied by comparing retention indices and
mass spectra with a laboratory mass spectral/retention index li-
brary based on 130 chemical standards plus the Golm Metabolome
Database (GMD, Schauer et al., 2005), Agilent Fiehn and NIST
libraries.
2.8. Capillary electrophoresis of nitrate and ammonium
Nitrate and ammonium in H2O extracts were determined by
capillary electrophoresis with indirect UV detection, essentially as
described previously (Warren and Adams, 2004). Separations and
quantication were performed with a CE system (P/ACE MDQ,
BeckmaneCoulter Inc, Fullerton, CA, USA) with a 50 mm i.d.  50 cm
capillary of bare fused silica. Samples were analyzed without any
pre-treatment. Ammonium was analyzed by pressure injection
(0.5 psi  30 s); separation at 25 kV (normal polarity) with a
background electrolyte of 10 mM imidazole, 2 mM 18-crown-6 at
pH 4.2; and detection by indirect UV at 200 nm. Nitrate was
injected by pressure (0.5 psi  30 s); separation at 25 kV (reverse
polarity) with a background electrolyte of 20 mM 2,6-pyr-
idinedicarboxylic acid, 0.5 mM cetyltrimethylammonium bromide
at pH 5.6; and detection by indirect UV at 214 nm. Nitrate and
ammonium were identied and quantied with external standards.
2.9. Total oxidizable carbon
Total oxidizable carbon in K2SO4 (1.5% CHCl3) extracts was
determined colorimetrically (Bartlett and Ross, 1988) using a
microplate reader (Synery 2, BioTek, Winooski, VT, USA).
2.10. Statistics
For samples collected at the end of the nal drying cycle,
Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA,
Bylesj et al., 2006; Warren et al., 2012) was used to separate
predictive variation related to water stress from non-predictive
(orthogonal) variation and thereby help identify compounds that
differed between control and drought treatments. To perform
multivariate statistics, I constructed sample matrices that com-
bined data of GCeMS and CEeMS. Data were pre-processed using
typical procedures for mass spectrometry based metabolite data
(Wiklund et al., 2008): omission of those compounds that were not
present in at least 50% of samples in a treatment, Pareto scaling and
log transformation. Data were normalized (to the sum of DON
26 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
monomers or sum of small carbohydrates) to allow visualization of
patterns in relative amounts of metabolites. To identify compounds
that led to discrimination between groups (i.e. control
versus drought stressed), an S-plot was used to visualize covariance
and correlation loading proles. Multivariate statistics were per-
formed with SIMCA P 12.01 (Umetrics AB, Sweden).
3. Results
3.1. Soil water content
Volumetric water content (at 15 cm) of control mesocosms
varied in between 13 and 15% (Fig. 1). Volumetric water content of
treatment (droughted) mesocosms decreased below control levels
when water was withheld, but effects were modest for the three
three-week-long drying cycles. At the conclusion of the nal 21-
week-long drying cycle water content of the droughted meso-
cosms was 3e5%. In all cases rewatering led to almost immediate
increases in soil water content to control levels.
3.2. Soil respiration
At the conclusion of the nal drying cycle, CO2 efux measured
between 10AM and 2PM was approximately three times as large in
control mesocosms (1.5e1.9 mmol m2 s1) as in treatment
(droughted) mesocosms (0.5e0.6 mmol m2 s1) (Fig. 2). Rewater-
ing of drought-stressed mesocosms led to 5e10-fold increases in
CO2 efux within minutes. CO2 efux of rewatered mesocosms
remained higher than control mesocosms for approximately 10
days.
3.3. Small molecules in soil extracts: overall prole and differences
between extract types
Forty low-molecular weight compounds were above quanti-
cation limits in at least 50% of samples from a treatment. Of the 40
compounds, 31 were organic N monomers that were separated and
quantied by CEeMS, while 9 were sugars and sugar alcohols
separated and quantied by GCeMS. The 31 organic N monomers
comprised all protein amino acids except cysteine (not quantiable
Fig. 2. Soil CO2 efux of two control and two treatment (drought) mesocosms as a
function of time after rewatering (where time 0 is the moment of rewatering).
Measurements were made several days before rewatering (i.e. when treatment mes-
ocosms were drought stressed), and then for 28 days after rewatering. In one drought
stressed replicate CO2 efux was measured every 30 min for ten days after rewatering.
CO2 efux of other mesocosms was measured between 10AM and 2PM. All data points
are means of three replicate measurements.
with the methods used), three non-protein amino acids (citrulline,
ornithine, GABA), eight quaternary ammonium compounds
(choline, g-butyrobetaine, hercynine, carnitine, acetyl carnitine,
trigonelline, proline-betaine, betaine), one pyridine derivative
(nicotinic acid) and two hydroxyrimidine derivatives (ectoine and
hydroxyectoine). In addition to the 31 organic N monomers quan-
tiable in most samples, there were at least 15 additional com-
pounds that were present at concentrations too low for reliable
quantication and/or in a minority of samples. Examples of such
compounds include guanine, cytosine, adenine, creatinine, b-
alanine, hydroxyproline, ergothioneine, hydroxyproline-betaine,
glucosamine. GCeMS detected only nine compounds that were
above quantication limits in at least 50% of samples from a
treatment. All nine compounds reported by GCeMS were sugars or
sugar alcohols (fructose, glucose, sucrose, rafnose, trehalose, myo-
inositol, arabitol, mannitol). Compounds detected by GCeMS but
present at concentrations too low for reliable quantication and/or
in a minority of samples included several organic acids (malic acid,
succinic acid, fumaric acid, citric acid, quinic acid), additional sugars
and sugar alcohols (threitol, erythitol, arabinose, mannose,
maltose) and some of the more abundant protein amino acids
(glutamic and aspartic acids, serine, glycine). Nitrate and ammo-
nium in H2O extracts were quantied by CE with indirect UV
detection, but data are not reported for nitrate because it was ab-
sent or below detection limits in most samples.
Absolute concentrations of small molecules were several times
larger in soil extracted with K2SO4 that contained 1.5% CHCl3
(hereafter referred to as CHCl3 extracts) than either K2SO4 extracts
or H2O extracts (Fig. 3, upper panels). Hence, concentrations in the
chloroform labile pool (calculated as CHCl3 extract  K2SO4 extract)
were some 2e5 times greater than in the H2O-extractable free pool
or K2SO4-extractable exchangeable pool (see also Table 1 for data
on the basis of moles of C). Differences among extract types in
relative abundances of different compounds were generally small
(Fig. 3 bottom panels and Fig. 4). For example, the molecular
composition of the chloroform labile pool (calculated as CHCl3
extract  K2SO4 extract) was very similar to CHCl3 extracts and
K2SO4 extracts but more variable (due to mathematical accumula-
tion of uncertainty), so data are not presented separately. The most
notable differences between extract types were that relative
abundance of arginine, lysine and choline was 3e8-fold greater in
K2SO4 and CHCl3 extracts than water extracts, while aspartic and
glutamic acids were 3e6-fold less abundant in K2SO4 and CHCl3
extracts than water extracts (Fig. 4). The most probable explanation
is that K2SO4 affects the interaction of organic cations and anions
with negatively charged soil colloids, while differential effects of
K2SO4 on solubility may also play a part (Macedo, 2005).
3.4. Small molecules in soil extracts: univariate differences between
control and drought
At the conclusion of the nal drying cycle the pool of organic N
monomers in soil from drought-stressed mesocosms was larger
than in control mesocosms by a factor of three (in K2SO4 and CHCl3
extracts)eseven (in water extracts). The larger pool size of organic
N monomers in drought-stressed mesocosms was not due to a
general increase in concentration of all compounds, but could
instead be attributed to increases in a modest number of com-
pounds (Fig. 3). Among the three extract types, drought-stressed
mesocosms contained 9e28-fold more betaine and this accoun-
ted for 31e46% of the larger pool of organic N monomers in
drought-stressed than control mesocosms. Ectoine and hydrox-
yectoine were absent or below detection limits in control meso-
cosms, but were among the ten most abundant organic N
monomers in drought-stressed mesocosms and each individually
 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32 27

Fig. 3. Absolute concentrations (a, c) and relative concentrations (b, d) of monomeric organic N compounds (a, b) and sugars and sugar alcohols (c, d) of soil collected at the end of a
21-week-long drying cycle in control (C) and drought-stressed (D) mesocosms. Soil was extracted with water (H2O), 0.5 M K2SO4, or CHCl3 (0.5 M K2SO4 1.5% CHCl3). For organic N
monomers data are shown as broad compound classes: ect h-ect ectoine plus hydroxyectoine, QAC quaternary ammonium compounds, NPAA non-protein amino acids,
protein AA protein amino acids. For sugars and sugar alcohols data are shown as individual compounds: raff rafnose, treh trehalose, suc sucrose, myo myo-inositol,
glc glucose, frc fructose, mann mannose, arab arabitol, glyc glycerol. Data are mean of four replicate mesocosms. Standard error bars are shown for the total pool of
organic N monomers and sugars sugar alcohols.

accounted for 5e8% of the greater concentration of organic N
monomers in drought-stressed than control mesocosms. Other
compounds present at larger concentrations in drought-stressed
mesocosms were: proline (17e25-fold greater concentration),
asparagine (10e28-fold greater concentration), glutamine (4e28-
fold greater concentration), proline-betaine (11e196-fold greater
concentration).
The pools of organic N monomers in control and drought-
stressed mesocosms were characterized by differences in relative
abundance of compound classes (Fig. 3 lower panel) and individual
compounds (Fig. 4). In drought-stressed mesocosms, protein amino
acids accounted for a smaller proportion of organic N monomers
(49e52% versus 64e85% for control mesocosms), quaternary
ammonium compounds accounted for a larger proportion of
organic N monomers (31e37% versus 12e29% in controls), and two
hydroxypyrimidine derivatives (ectoine and hydroxyectoine) went
from at or below detection limits in controls to accounting for 8e
11% in drought-stressed mesocosms. The most striking differences
in abundance of individual compounds were that the pool of
organic N monomers in drought-stressed mesocosms contained
large amounts of several putative osmolytes (betaine, proline,
ectoine, hydroxyectoine: Fig. 4). For example, in drought-stressed
mesocosms betaine was 2e6 times more abundant than the next
most abundant compound and accounted for 25e30% of the pool of
organic N monomers, whereas in control mesocosms betaine was
4the6th most abundant; while proline, ectoine and hydroxyectoine
were among the ten most abundant compounds in drought-
stressed mesocosms, but not in control mesocosms.
Compared with control mesocosms, the pool of sugars and sugar
alcohols in drought-stressed mesocosms was nine times larger in
water extracts, ve times larger in K2SO4 extracts and two times
larger in CHCl3 extracts (Fig. 3). The large difference in pool sizes
was not due to a general increase in concentration of all com-
pounds. In CHCl3 extracts the larger pool of sugars and sugar al-
cohols in drought-stressed mesocosms could be attributed to 3-fold
larger amounts of trehalose (accounting for 71% of the difference in
pool of carbohydrates between control and drought) and 5-fold
larger amounts of arabitol (accounting for 29% of the difference in
pool size between control and drought). In H2O and K2SO4 extracts
the larger pool of sugars and sugar alcohols in drought-stressed

Table 1
Concentrations of broad compound classes in K2SO4 extracts and CHCl3 extracts (0.5 M K2SO4 1.5% CHCl3) of soil collected at the end of a 21-week-long drying cycle. Data are
monomeric organic N compounds (determined by CEeMS), sugars and sugar alcohols (determined by GCeMS), and total oxidizable carbon (TOC, determined colorimetrically).
The CHCl3 labile fraction was calculated by subtracting the concentration in 0.5 M K2SO4 extracts from the concentration (of TOC or osmolytes) in CHCl3-treated 0.5 M K2SO4
extracts. The pool of osmolytes was dened here as: organic N monomers ectoine hydroxyectoine betaine proline-betaine trigonelline proline; sugars and sugar
alcohols trehalose arabitol. All data are expressed on the basis of moles of C (i.e. after accounting for number of C atoms in different compounds) per kg of dry soil, and are
means (SE) of four replicate mesocosms.

Organic N monomers (mmol C kg1 soil) Sugars & sugar alcohols (mmol C kg1 soil) TOC (mmol C kg1 soil)

K2SO4 CHCl3 CHCl3 labile CHCl3 labile K2SO4 CHCl3 CHCl3 labile CHCl3 labile K2SO4 CHCl3 CHCl3 labile
osmolytes osmolytes

Control 50 (10) 205 (25) 155 (27) 20 (4) 30 (10) 620 (80) 590 (81) 189 (28) 1820 (230) 7440 (685) 5620 (717)
Drought-stressed 165 (60) 580 (52) 415 (79) 229 (45) 165 (60) 1125 (250) 960 (257) 611 (173) 2265 (140) 7200 (415) 4935 (436)
28 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
Fig. 4. Relative concentrations (molar basis) of monomeric organic N compounds of
soil collected at the end of a 21-week-long drying cycle in control (aec) and drought-
stressed mesocosms (def). Soil was extracted with water (a, d), 0.5 M K2SO4 (b, e) or
CHCl3 extracts (c, f). The ten most abundant compounds are named, while the
remaining 21 lower abundance compounds are grouped and presented as others. Data
are mean of four replicate mesocosms.
mesocosms could be attributed to trehalose (17e20% of difference),
arabitol (29e36%), fructose (7%), glucose (12e22%), sucrose (12e
22%).
3.5. Small molecules in soil extracts: multivariate differences
between control and drought
As a check on inferences drawn from univariate relationships
(see above), OPLS-DA was used to separate multivariate relation-
ships into predictive variation (related to water stress treatment)
and orthogonal variation (unrelated to treatment). Use of non-
normalized data did not produce strong multivariate models (CV-
ANOVA > 0.05) because patterns were completely dominated by
treatment differences in absolute concentrations (see Fig. 3 top
panel). In contrast, data normalized based on the pool of organic N
monomers and small carbohydrates led to signicant multivariate
models that reected relative differences in the overall prole of
compounds. Multivariate solutions are shown here only for H2O
extracts, but were very similar for soil extracted with H2O, K2SO4, or
CHCl3. OPLS-DA produced reasonable models (R2Y > 0.95;
Q2 > 0.75; CV-ANOVA, P < 0.05) and analysis of the scores plots
(rst predictive component versus rst orthogonal component)
showed complete separation of control mesocosms from drought-
stressed mesocosms (data not shown). Visual analysis of S-plots
(Fig. 5) led to identication of compounds at signicantly greater
relative concentration in drought-stressed than control mesocosms
(upper right) and signicantly lower concentrations in drought-
stressed than control mesocosms (lower left). A consistent picture
that emerged from OPLS-DA was that for all three extract types
drought-stressed mesocosms contained greater relative concen-
trations of ectoine, hydroxyectoine, betaine, proline-betaine, trig-
onelline, trehalose, arabitol, and proline. Compounds that were at
signicantly lower relative concentrations in drought-stressed
mesocosms were Gly, Ser, Thr, Ala, Leu, Ile, carnitine, acetyl carni-
tine, choline, hercynine, myo-inositol, mannitol.
3.6. Small molecules in soil extracts: contribution to total oxidizable
carbon
There was no difference in total oxidizable carbon between
control and drought-stressed mesocosms (Table 1), despite the
larger pool of small molecules in drought-stressed than control
mesocosms (Fig. 3, Table 1). Hence, small molecules accounted for
approximately twice as large a proportion of TOC in drought-
stressed mesocosms as in control mesocosms. For example, in the
CHCl3 labile fraction organic N monomers accounted for 3% of TOC
in control mesocosms and 8% in drought stressed, while small
carbohydrates accounted for 10% of TOC in control mesocosms and
20% in drought-stressed mesocosms. The pool of osmolytes
(dened here as ectoine, hydroxyectoine, betaine, proline-betaine,
trigonelline, trehalose, arabitol, and proline) accounted for 3.6% of
CHCl3 labile TOC in control mesocosms and 17% of CHCl3 labile TOC
in drought-stressed mesocosms.
3.7. Small molecules in soil extracts: response to rewatering
Rewatering led to rapid decreases in concentrations of organic N
monomers, ammonium and small carbohydrates such that by 24 h
after rewatering there were no consistent differences in concen-
trations between control and rewatered mesocosms (Fig. 7). In the
case of organic N monomers and ammonium there was a large
decrease in concentrations in the rst hour after rewatering, and
only minor decreases thereafter. Concentrations of small carbohy-
drates were somewhat slower to decrease upon rewatering, but the
bulk of the decrease occurred within the rst 3 h. The decrease in
concentration of organic N monomers in rewatered mesocosms
was associated with large reductions in amounts of betaine, pro-
line, ectoine, hydroxyectoine, proline-betaine, aspartic and gluta-
mic acids (Fig. 7). The decrease in concentration of small
carbohydrates in rewatered mesocosms was associated with large
decreases in absolute and relative amounts of trehalose and ara-
bitol. Interestingly, in the rst hour after rewatering while the
concentration of arabitol was decreasing by 23 mmol kg1 and
trehalose by 11 mmol kg1 there were increases in rafnose of
12 mmol kg1 and sucrose of 5 mmol kg1.
4. Discussion
4.1. Prole of small molecules in soil extracts
Few studies have reported broad, non-targeted characterization
of small molecules in soil exposed to dryewet cycles (Williams and
Xia, 2009) at least in part because this poses a considerable
analytical challenge. The scope of the challenge is best illustrated by
the surprisingly limited molecular information obtained from GCe
MS. GCeMS of methoximated trimethylsilyl derivatives has found
widespread use for complex samples of biological origin because it
provides good coverage of many small molecules (Roessner et al.,
2000; Koek et al., 2006, 2011). However, for the soil extracts
 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32 29

2013). One unavoidable problem is that concentrations of small

molecules in H2O and K2SO4 extracts are overestimated to an un-
known extent by compounds extracted from lysed microbes.
Absolute concentrations should be interpreted cautiously, but
there is reasonable evidence that none of this works substantive
conclusions are unduly affected by limitations of aqueous extracts.
For example, the quantitative signicance of osmolytes was esti-
mated from the CHCl3 labile fraction (i.e. CHCl3eK2SO4), and thus
would have been underestimated if lysing of microbes by K2SO4 was
signicant. The veracity of data based on H2O and K2SO4 extracts is
strengthened by the observation there were treatment differences
not only in absolute concentrations but also in relative concentra-
tions (e.g. >10-fold differences in relative abundance of osmolytes).
Hence, inferences drawn from multivariate statistics based on
normalized data (i.e. describing the pattern of relative abundance,
Fig. 5. S-plot of orthogonal partial least square-discriminant analysis (OPLS-DA) of Fig. 5) arrived at effectively the same conclusions as inferences
small molecules in H2O extracts of soil from control and drought-stressed mesocosms drawn from univariate analysis of absolute concentrations.
at the end of a 21-week-long drying cycle. The plot shows the relative contribution of
small molecules to discrimination of control from drought-stressed mesocosms. The S-
plot combines the covariance and correlation loading proles. This corresponds to
combining the contribution or magnitude (covariance) with the effect and reliability
(correlation) for the model variables with respect to model component scores. In this
plot, the p loading prole (p[1]) of the rst predictive (related to drought) OPLS-DA
component is plotted on the x-axis and represent X-variable contribution (covari-
ance). The p correlation loading vector (p(corr)[1])of the rst predictive (class-
discriminating) component is plotted on the y-axis and represents the correlation
(reliability) of each X-variable with the rst predictive score component. The further
away a small molecule is from the value 0, the better is its correlation from replicate to
replicate. As a result, the compounds on both ends of the curve are the strongest
contributors to discriminating control from drought-stressed mesocosms.

analyzed here only nine sugars and sugar alcohols were consis-
tently detected, and similarly sparse molecular proles were re-
ported by other GCeMS studies of soil extracts (Williams and Xia,
2009; Kakumanu et al., 2013). To put this in perspective, studies
analyzing plant samples or microbial cultures with the same GCe
MS approach commonly report in excess of 100 compounds from a
variety of compound classes (Roessner et al., 2000; Koek et al.,
2006, 2011; Warren et al., 2011, 2012). The simplest explanation
as to why so few compounds were detected in soil extracts is that
most of the compounds in soil were below detection limits due to a
combination of inherently low concentrations and poor detection
limits for some compounds (particularly those with amine, phos-
phoric, amide, thiol, or sulfonic functional groups). Detection of
non-carbohydrate compounds in soil extracts might be possible
with more extensive pre-concentration of extracts prior to injec-
tion, but then concentrations of sugars and sugar alcohols would be
well above the linear range (typically 103e104) of GCeMS. In any
case, even with extensive concentration and fractionation of sam-
ples prior to analysis a considerable number of nitrogenous com-
pounds cannot be derivatized and will remain invisible to common
GCeMS approaches (Warren, 2013a).
The choice of how to extract small molecules from soil is a vexed
issue that affects subsequent interpretation of data (e.g. Jones and
Willett, 2006; Inselsbacher et al., 2011). For dry soils minimally
invasive techniques such as microdialysis and suction lysimetry do
not work, and thus there is little choice but to use aqueous extracts.
Aqueous extracts are beset by a raft of problems, some can be
quantied, some can be minimized while others are unavoidable. In
the present study the problem of metabolism during extraction Fig. 6. Response of organic N monomers (a), ammonium (b), sugars and sugar alcohols
(Rousk and Jones, 2010) was minimized by extracting samples (c) to rewatering. Mesocosms were rewatered (at T0) after a 21-week-long drying
rapidly (10 min for H2O extracts, 30 min for K2SO4 and CHCl3 ex- cycle. Soil samples from drought-stressed mesocosms were collected immediately
tracts). The likely modest extent of metabolism during extraction is before rewatering (T0), 1 h, 3 h, 1 day, 3 days, 7, days, 14 days, 21 days, 28 days after
rewatering. Soil was extracted with water. Samples from control mesocosms were
supported by a standard addition experiment that established collected at 0 and 28 days only. The overwhelming majority of the change in con-
there was limited microbial consumption or abiotic adsorption of centrations occurred within the rst 24 h, so the inset expands upon this region. Data
amino acids during extraction of drought-stressed soils (Boot et al., are means (SE) of four replicate mesocosms.
30 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
Fig. 7. Response of individual organic N monomers (a), sugars and sugar alcohols (b) to
rewatering. Mesocosms were rewatered (at T0) after a 21-week-long drying cycle.
Samples were collected immediately before rewatering in control and drought-
stressed mesocosms. Samples from drought-stressed mesocosms were collected 1 h,
3 h, 1 day, 3 days, 7, days, 14 days, 21 days, 28 days after rewatering. Soil was extracted
with water. Data are shown only for the rst 24 h after rewatering (see Fig. 6 for 28-day
time course). Data are means of four replicate mesocosms. Standard error bars are
shown for the total pool of organic N monomers and sugars sugar alcohols.
4.2. Occurrence of water decits and osmolyte accumulation
To mimic the slow development of water stress that occurs
under eld conditions, water was withheld from large (200 L)
vegetated mesocosms and water stress developed slowly over a
period of 21 weeks (Fig. 1). At the conclusion of the drying cycle
volumetric water content was 3e5%, which based on water release
characteristics for similar textured soils would equate to water
potential more negative than the nominal permanent wilting point
of 1.5 MPa. The 3-fold slower rate of CO2 efux from droughted
mesocosms than control mesocosms (Fig. 2) is consistent with a
strong effect of water decits on microbial activity (Orchard and
Cook, 1983; Skopp et al., 1990; Manzoni et al., 2012). There was
no effect of water decits on microbial biomass (CHCl3 labile TOC,
Table 1), consistent with other studies showing no effect of sea-
sonal water stress on microbial biomass (Boot et al., 2013), but
contradicting others showing decreases in microbial biomass (e.g.
Sparling et al., 1986; Kieft et al., 1987; Jensen et al., 2003; Wu and
Brookes, 2005; Bapiri et al., 2010). In any case, it is possible that
water decit led to marked shifts in structure of the microbial
community (Fernandez et al., 2012; Hueso et al., 2012; Kakumanu
et al., 2013; Kwon et al., 2013) without altering overall microbial
biomass.
Slow development of water decits permitted testing of the
hypothesis that osmolytes accumulate under drought stress. Pre-
vious experiments with soils have provided little support for this
prediction with recent studies failing to nd large constitutive or
inducible amounts of osmolytes ineld experiments on grassland
soil (Boot et al., 2013) or forest soils (Gransson et al., 2013) or
laboratory experiments exposing soils to water stress treatments
(Williams and Xia, 2009; Kakumanu et al., 2013). In contrast with
these studies but in agreement with the hypothesis, pools of
quantied small molecules were several times larger in soil from
drought-stressed mesocosms than control mesocosms, primarily
due to >10-fold larger amounts of known osmolytes (ectoine,
hydroxyectoine, betaine, proline-betaine, trigonelline, proline,
trehalose, arabitol: Csonka, 1989; Lippert and Galinski, 1992;
Hasegawa et al., 2000; Wood et al., 2001) (Fig. 3, Table 1). Hence,
data are consistent with results of various culture-based studies
suggesting that soil microbes accumulate osmolytes as a means of
coping with osmotic stress (e.g. Kempf and Bremer, 1998b) and
contradict claims that results of culture studies of individual mi-
croorganisms may not apply to mixed microbial consortia in soil
(Boot et al., 2013).
Why did the present study detect signicant accumulation of
osmolytes in soil, whereas other studies (Williams and Xia, 2009;
Boot et al., 2013; Gransson et al., 2013; Kakumanu et al., 2013)
did not? A large part of the explanation is that there are likely
biological differences among soil ecosystems in the quantitative
signicance of osmolyte accumulation. For example, differences
among ecosystems in the microbial community that lead to varying
proportions of strategies employed to cope with water decits (e.g.
osmolyte accumulation, dormancy, mucilage and exopoly-
saccharides). A complementary explanation is that experimental
procedures and measurements may have contributed to differences
among studies. In the case of the two laboratory-based studies
(Williams and Xia, 2009; Kakumanu et al., 2013) water decits may
have been imposed for too short a duration (3e4 days) for signi-
cant osmolytes to accumulate (e.g. as has been argued for plants:
Turner, 1986) or for longer-term changes in microbial metabolism
(Gransson et al., 2013; Meisner et al., 2013). The problem of short-
duration may have been compounded by the absence of labile C
inputs from plants (i.e. rhizodeposits) further limiting the rate of
synthesis of osmolytes. Another partial explanation is that past
studies used analytical approaches unlikely to capture the diversity
of C- and N-containing osmolytes (Figs. 3 and 4, and see also:
Csonka, 1989; Lippert and Galinski, 1992; Hasegawa et al., 2000;
Wood et al., 2001). For example, in the present study GCeMS was
able to reliably detect only sugars and sugar alcohols (see
Discussion above) and thus the studies that relied upon GCeMS
(Williams and Xia, 2009; Kakumanu et al., 2013) may have missed
N-containing osmolytes. On the other hand, the eld based study
that quantied only N-containing osmolytes (Boot et al., 2013) may
have failed to see accumulation of C-containing osmolytes (e.g.
arabitol and trehalose).
4.3. Quantitative signicance of osmolytes and response to
rewatering
Osmolytes not only increased under water stress but also made
a substantial contribution to microbial C, indicating that osmolyte
accumulation is a signicant and costly means of coping with
drought. This is borne out by the observation that the pool of
osmolytes (dened here as ectoine, hydroxyectoine, betaine, pro-
line-betaine, trigonelline, proline, trehalose, and arabitol) accoun-
ted for 3.6% of CHCl3 labile TOC in control mesocosms and 17% of
CHCl3 labile TOC in drought-stressed mesocosms (Table 1). Data are
consistent with previous theoretical estimates of the C-cost of
osmolyte accumulation that suggested osmolytes might account for
7e20% of microbial C (Schimel et al., 2007), and a laboratory study
nding that the pool of microbial metabolites accounted for 20e
 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
50% of microbial C (Kakumanu et al., 2013). In contrast, a eld study
of nitrogenous osmolytes reported that glutamate was consistently
2e3% of microbial C while betaine and proline were below detec-
tion limits (Boot et al., 2013). Clearly additional studies are required
to determine whether osmolytes are consistently large proportions
(e.g. 10 s of %) of microbial C.
Interest in the quantitative signicance of osmolytes has been
fueled by suggestions that osmolytes could (at least partially) un-
derpin the large pulse of CO2 efux (respiration) induced by
rewatering (Schimel et al., 2007; Borken and Matzner, 2009). To
date there has been little experimental support for this hypothesis
with recent studies nding pools of osmolytes or small labile C
compounds were small and/or not correlated with respiratory re-
sponses (Williams and Xia, 2009; Gransson et al., 2013), and
suggesting rewatering leads to a pulse of respiration by physically
perturbing soil and thereby increasing availability of substrates
(Gransson et al., 2013). In contrast, very different conclusion can
be drawn from results of the present study with a simple budgeting
approach indicating microbial osmolytes have the potential to
make a signicant contribution to the pulse of respiration after
rewatering. The pulse of CO2 efux after rewatering was approxi-
mately 0.34 mol C m2 (based on the ten days CO2 efux from
rewatered mesocosms was greater than control mesocosms, Fig. 2)
of which half (0.17 mol C m2) can probably be attributed to mi-
crobes. This estimate of the microbial pulse of CO2 is of the same
order of magnitude as the difference in microbial osmolytes be-
tween drought-stressed and control mesocosms of 0.12 mol C m2
(assuming microbial C is restricted to upper 20 cm of soil, Table 1).
One way to address the role of microbial osmolytes in sup-
porting CO2 efux is to determine if, as hypothesized, rewatering
leads to a large pulse of osmolytes in free solution. Contrary to the
hypothesis there was no evidence that rewatering led to a large
pulse of osmolytes in free solution. On the contrary, concentrations
of small molecules decreased massively within the rst hour of
rewatering, and within 1e3 h of rewatering concentrations of
organic N monomers, sugars and sugar alcohols had decreased to
approximately the same concentration as in controls (Fig. 7). The
rapid decrease in organic N monomers and small carbohydrates
upon rewatering is entirely consistent with two related observa-
tions: small molecules are rapidly taken up by plants and microbes
(Jones and Murphy, 2007; Boddy et al., 2008; Jones and Kielland,
2012; Roberts and Jones, 2012; Warren, 2013b), and consequently
steady-state concentrations in free solution tend to be low (van
Hees et al., 2005; Jones and Willett, 2006). Interestingly, in the
rst hour after rewatering while the concentration of various
osmolytes was decreasing there were substantial increases in
rafnose and sucrose. Sucrose is more commonly plant than
microbe associated, and thus the increase may have come about
due to the resumption of plant inputs into the soil (i.e. root
exudation). Isotope labeling (Fierer and Schimel, 2003) and direct
measurements of labeled molecules and CO2 is necessary to
determine the sources of small molecules in the soil solution and
CO2 efux (e.g. derived from recent photosynthesis versus accu-
mulated under stress) and provide direct evidence of a link be-
tween osmolyte accumulation under drought and subsequent CO2
efux upon rewetting.
5. Conclusions
Consistent with the rst hypothesis it was found that prolonged
drought led to substantial accumulation of osmolytes. Moreover,
osmolytes not only increased under water stress but also made a
substantial contribution to microbial C, and may have at least
partially fueled the large pulse of soil respiration upon rewatering.
Additional studies on other soils are required to establish whether
31
osmolyte accumulation is widespread among ecosystems given
that previous studies did not observe signicant osmolyte accu-
mulation (Williams and Xia, 2009; Boot et al., 2013; Gransson
et al., 2013; Kakumanu et al., 2013). In contrast with the second
hypothesis there was no evidence that rewatering led to a large
pulse of osmolytes in free solution. Instead rewatering decreased
concentrations of osmolytes within hours, which was consistent
with rapid microbial uptake and the role of osmolytes in fueling the
pulse of soil respiration.
Acknowledgments
Charles Warren is supported by a Future Fellowship from the
Australian Research Council.
References
Bapiri, A., Baath, E., Rousk, J., 2010. Drying-rewetting cycles affect fungal and bac-
terial growth differently in an arable soil. Microbial Ecology 60, 419e428.
Bartlett, R.J., Ross, D.S., 1988. Colorimetric Determination of oxidizable carbon in
acid soil solutions. Soil Science Society of America Journal 52, 1191e1192.
Birch, H.F., 1958. The effect of soil drying on humus decomposition and nitrogen
availability. Plant and Soil 10, 9e31.
Birch, H.F., 1964. Mineralisation of plant nitrogen following alternate wet and dry
conditions. Plant and Soil 20, 43e49.
Boddy, E., Roberts, P., Hill, P.W., Farrar, J., Jones, D.L., 2008. Turnover of low mo-
lecular weight dissolved organic C (DOC) and microbial C exhibit different
temperature sensitivities in Arctic tundra soils. Soil Biology and Biochemistry
40, 1557e1566.
Boot, C.M., Schaeffer, S.M., Schimel, J.P., 2013. Static osmolyte concentrations in
microbial biomass during seasonal drought in a California grassland. Soil
Biology and Biochemistry 57, 356e361.
Borken, W., Matzner, E., 2009. Reappraisal of drying and wetting effects on C and N
mineralization and uxes in soils. Global Change Biology 15, 808e824.
Bottner, P., 1985. Response of microbial biomass to alternate moist and dry condi-
tions in a soil incubated with C-14-labeled and N-15-labelled plant-material.
Soil Biology and Biochemistry 17, 329e337.
Bylesj, M., Rantalainen, M., Cloarec, O., Nicholson, J.K., Holmes, E., Trygg, J., 2006.
OPLS discriminant analysis: combining the strengths of PLS-DA and SIMCA
classication. Journal of Chemometrics 20, 341e351.
Cook, E.R., Woodhouse, C.A., Eakin, C.M., Meko, D.M., Stahle, D.W., 2004. Long-term
aridity changes in the western United States. Science 306, 1015e1018.
Csonka, L.N., 1989. Physiological and genetic responses of bacteria to osmotic-stress.
Microbiological Reviews 53, 121e147.
Dai, A., Trenberth, K.E., Qian, T.T., 2004. A global dataset of Palmer Drought Severity
Index for 1870e2002: relationship with soil moisture and effects of surface
warming. Journal of Hydrometeorology 5, 1117e1130.
DellAcqua, M., Gomarasca, S., Porro, A., Bocchi, S., 2013. A tropical grass resource for
pasture improvement and landscape management: Themeda triandra Forssk.
Grass and Forage Science 68, 205e215.
Fernandez, D.A., Roldan, A., Azcon, R., Caravaca, F., Baath, E., 2012. Effects of water
stress, organic amendment and mycorrhizal inoculation on soil microbial
community structure and activity during the establishment of two heavy
metal-tolerant native plant species. Microbial Ecology 63, 794e803.
Fierer, N., Schimel, J.P., 2003. A proposed mechanism for the pulse in carbon dioxide
production commonly observed following the rapid rewetting of a dry soil. Soil
Science Society of America Journal 67, 798e805.
Gransson, H., Godbold, D.L., Jones, D.L., Rousk, J., 2013. Bacterial growth and
respiration responses upon rewetting dry forest soils: impact of drought-legacy.
Soil Biology and Biochemistry 57, 477e486.
Halverson, L.J., Jones, T.M., Firestone, M.K., 2000. Release of intracellular solutes by
four soil bacteria exposed to dilution stress. Soil Science Society of America
Journal 64, 1630e1637.
Hasegawa, P.M., Bressan, R.A., Zhu, J.K., Bohnert, H.J., 2000. Plant cellular and mo-
lecular responses to high salinity. Annual Review of Plant Physiology and Plant
Molecular Biology 51, 463e499.
Hueso, S., Garcia, C., Hernandez, T., 2012. Severe drought conditions modify the
microbial community structure, size and activity in amended and unamended
soils. Soil Biology and Biochemistry 50, 167e173.
Inselsbacher, E., Ohlund, J., Jmtgrd, S., Huss-Danell, K., Nsholm, T., 2011. The
potential of microdialysis to monitor organic and inorganic nitrogen com-
pounds in soil. Soil Biology and Biochemistry 43, 1321e1332.
Jarvis, P.G., et al., 2007. Drying and wetting of Mediterranean soils stimulates
decomposition and carbon dioxide emission: the Birch effect. Tree Physiology
27, 929e940.
Jensen, K.D., Beier, C., Michelsen, A., Emmett, B.A., 2003. Effects of experimental
drought on microbial processes in two temperate heathlands at contrasting
water conditions. Applied Soil Ecology 24, 165e176.
32 C.R. Warren / Soil Biology & Biochemistry 70 (2014) 22e32
Jones, D.L., Willett, V.B., 2006. Experimental evaluation of methods to quantify
dissolved organic nitrogen (DON) and dissolved organic carbon (DOC) in soil.
Soil Biology and Biochemistry 38, 991e999.
Jones, D.L., Murphy, D.V., 2007. Microbial response time to sugar and amino acid
additions to soil. Soil Biology and Biochemistry 39, 2178e2182.
Jones, D.L., Kielland, K., 2012. Amino acid, peptide and protein mineralization dy-
namics in a taiga forest soil. Soil Biology and Biochemistry 55, 60e69.
Kakumanu, M.L., Cantrell, C.L., Williams, M.A., 2013. Microbial community response
to varying magnitudes of desiccation in soil: a test of the osmolyte accumula-
tion hypothesis. Soil Biology and Biochemistry 57, 644e653.
Kempf, B., Bremer, E., 1998a. Stress responses of Bacillus subtilis to high osmolarity
environments: uptake and synthesis of osmoprotectants. Journal of Biosciences
23, 447e455.
Kempf, B., Bremer, E., 1998b. Uptake and synthesis of compatible solutes as mi-
crobial stress responses to high-osmolality environments. Archives of Micro-
biology 170, 319e330.
Kieft, T.L., Soroker, E., Firestone, M.K., 1987. Microbial biomass response to a rapid
increase in water potential when dry soil is wetted. Soil Biology and
Biochemistry 19, 119e126.
Koek, M.M., Muilwijk, B., van der Werf, M.J., Hankemeier, T., 2006. Microbial
metabolomics with gas chromatography/mass spectrometry. Analytical Chem-
istry 78, 1272e1281.
Koek, M.M., Jellema, R.H., van der Greef, J., Tas, A.C., Hankemeier, T., 2011. Quanti-
tative metabolomics based on gas chromatography mass spectrometry: status
and perspectives. Metabolomics 7, 307e328.
Kwon, M.J., Haraguchi, A., Kang, H., 2013. Long-term water regime differentiates
changes in decomposition and microbial properties in tropical peat soils
exposed to the short-term drought. Soil Biology and Biochemistry 60, 33e44.
Lippert, K., Galinski, E.A., 1992. Enzyme stabilization by ectoine-type compatible
solutes e protection against heating, freezing and drying. Applied Microbiology
and Biotechnology 37, 61e65.
Lisec, J., Schauer, N., Kopka, J., Willmitzer, L., Fernie, A.R., 2006. Gas chromatography mass
spectrometry-based metabolite proling in plants. Nature Protocols 1, 387e396.
Macedo, E.A., 2005. Solubility of amino acids, sugars, and proteins. Pure and Applied
Chemistry 77, 559e568.
Manzoni, S., Schimel, J.P., Porporato, A., 2012. Responses of soil microbial commu-
nities to water stress: results from a meta-analysis. Ecology 93, 930e938.
Meehl, G.A., et al., 2007. Global climate projections. In: Solomon, S., et al. (Eds.),
Climate Change 2007: The Physical Science Basis. Contribution of Working Group
I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge, UK and New York, NY, USA.
Meisner, A., Baath, E., Rousk, J., 2013. Microbial growth responses upon rewetting
soil dried for four days or one year. Soil Biology and Biochemistry 66, 188e192.
Orchard, V.A., Cook, F.J., 1983. Relationship between soil respiration and soil-
moisture. Soil Biology and Biochemistry 15, 447e453.
Reinhold, V.N., Ishikawa, Y., Melville, D.B., 1968. Synthesis of alpha-N-methylated
histidines. Journal of Medicinal Chemistry 11, 258e260.
Roberts, P., Jones, D.L., 2012. Microbial and plant uptake of free amino sugars in
grassland soils. Soil Biology and Biochemistry 49, 139e149.
Roessner, U., Wagner, C., Kopka, J., Trethewey, R.N., Willmitzer, L., 2000. Simulta-
neous analysis of metabolites in potato tuber by gas chromatography-mass
spectrometry. Plant Journal 23, 131e142.
Rousk, J., Jones, D.L., 2010. Loss of low molecular weight dissolved organic carbon
(DOC) and nitrogen (DON) in H2O and 0.5 M K2SO4 soil extracts. Soil Biology
and Biochemistry 42, 2331e2335.
Schauer, N., et al., 2005. GC-MS libraries for the rapid identication of metabolites
in complex biological samples. FEBS Letters 579, 1332e1337.
Schimel, J., Balser, T.C., Wallenstein, M., 2007. Microbial stress-response physiology
and its implications for ecosystem function. Ecology 88, 1386e1394.
Setia, R., Verma, S.L., Marschner, P., 2012. Measuring microbial biomass carbon by
direct extraction e comparison with chloroform fumigation-extraction. Euro-
pean Journal of Soil Biology 53, 103e106.
Skopp, J., Jawson, M.D., Doran, J.W., 1990. Steady-state aerobic microbial activity as a
function of soil-water content. Soil Science Society of America Journal 54, 1619e
1625.
Sparling, G.P., Speir, T.W., Whale, K.N., 1986. Changes in microbial biomass-C, ATP
content, soil phospho-monoesterase and phospho-diesterase activity following
air-drying of soils. Soil Biology and Biochemistry 18, 363e370.
Thomas, C.K., Law, B.E., Irvine, J., Martin, J.G., Pettijohn, J.C., Davis, K.J., 2009. Sea-
sonal hydrology explains interannual and seasonal variation in carbon and
water exchange in a semiarid mature ponderosa pine forest in central Oregon.
Journal of Geophysical Research: Biogeosciences 114.
Trenberth, K.E., et al., 2007. Observations: surface and atmospheric climate change.
In: Solomon, S., et al. (Eds.), Climate Change 2007: The Physical Science Basis.
Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press,
Cambridge, UK and New York, NY, USA.
Turner, N.C., 1986. Adaptation to water decits e a changing perspective. Australian
Journal of Plant Physiology 13, 175e190.
van Hees, P.A.W., Jones, D.L., Finlay, R., Godbold, D.L., Lundstomd, U.S., 2005. The
carbon we do not see e the impact of low molecular weight compounds on
carbon dynamics and respiration in forest soils: a review. Soil Biology and
Biochemistry 37, 1e13.
Warren, C.R., 2013a. High diversity of small organic N observed in soil water. Soil
Biology and Biochemistry 57, 444e450.
Warren, C.R., 2013b. Quaternary ammonium compounds can be abundant in some
soils and are taken up as intact molecules by plants. New Phytologist 198, 476e
485.
Warren, C.R., Adams, M.A., 2004. Capillary electrophoresis of the major anions
and cations in leaf extracts of woody species. Phytochemical Analysis 15,
407e413.
Warren, C.R., Aranda, I., Cano, F.J., 2011. Responses to water stress of gas exchange
and metabolites in Eucalyptus and Acacia spp. Plant Cell and Environment 34,
1609e1629.
Warren, C.R., Aranda, I., Cano, F.J., 2012. Metabolomics demonstrates diver-
gent responses of two Eucalyptus species to water stress. Metabolomics 8,
186e200.
Wiklund, S., et al., 2008. Visualization of GC/TOF-MS-based metabolomics data for
identication of biochemically interesting compounds using OPLS class models.
Analytical Chemistry 80, 115e122.
Williams, M.A., Xia, K., 2009. Characterization of the water soluble soil organic pool
following the rewetting of dry soil in a drought-prone tallgrass prairie. Soil
Biology and Biochemistry 41, 21e28.
Wood, J.M., et al., 2001. Osmosensing and osmoregulatory compatible solute
accumulation by bacteria. Comparative Biochemistry and Physiology Part A:
Molecular and Integrative Physiology 130, 437e460.
Wu, J., Brookes, P.C., 2005. The proportional mineralisation of microbial biomass
and organic matter caused by air-drying and rewetting of a grassland soil. Soil
Biology and Biochemistry 37, 507e515.