Received: 6 September 2016 | Revised: 25 April 2017 | Accepted: 2 May 2017

DOI: 10.1111/1755-0998.12690

RESOURCE ARTICLE

Environmental DNA enables detection of terrestrial mammals

from forest pond water

Masayuki Ushio1,2,3,4 | Hisato Fukuda4 | Toshiki Inoue4 | Kobayashi Makoto5 |

Osamu Kishida5,6 | Keiichi Sato7 | Koichi Murata8,9 | Masato Nikaido10 |
Tetsuya Sado11 | Yukuto Sato12 | Masamichi Takeshita13 | Wataru Iwasaki13 |
4,14 4 11
Hiroki Yamanaka | Michio Kondoh | Masaki Miya

1
Center for Ecological Research, Kyoto University, Otsu, Japan
2
PRESTO, Japan Science and Technology Agency, Kawaguchi, Japan
3
Joint Research Center for Science and Technology, Ryukoku University, Otsu, Japan
4
Department of Environmental Solution Technology, Ryukoku University, Otsu, Japan
5
Teshio Experimental Forest, Field Science Center for Northern Biosphere, Hokkaido University, Hokkaido, Japan
6
Tomakomai Experimental Forest, Field Science Center for Northern Biosphere, Hokkaido University, Hokkaido, Japan
7
Okinawa Churashima Research Center, Okinawa, Japan
8
College of Bioresource Sciences, Nihon University, Kanagawa, Japan
9
Yokohama Zoological Gardens ZOORASIA, Kanagawa, Japan
10
School of Life Science and Technology, Tokyo Institute of Technology, Tokyo, Japan
11
Natural History Museum and Institute, Chiba, Japan
12
Tohoku Medical Megabank Organization, Tohoku University, Miyagi, Japan
13
Department of Biological Sciences, The University of Tokyo, Tokyo, Japan
14
The Research Center for Satoyama Studies, Ryukoku University, Shiga, Japan

Correspondence
Masayuki Ushio, Center for Ecological Abstract
Research, Kyoto University, Otsu, Japan. Terrestrial animals must have frequent contact with water to survive, implying that
Email: ong8181@gmail.com
and environmental DNA (eDNA) originating from those animals should be detectable
Masaki Miya, Natural History Museum and from places containing water in terrestrial ecosystems. Aiming to detect the pres-
Institute, Chiba, Japan.
Email: miya@chiba-muse.or.jp ence of terrestrial mammals using forest water samples, we applied a set of univer-
sal PCR primers (MiMammal, a modified version of fish universal primers) for
Funding information
Core Research for Evolutional Science and metabarcoding mammalian eDNA. The versatility of MiMammal primers was tested
Technology, Grant/Award Number: in silico and by amplifying DNAs extracted from tissues. The results suggested that
JPMJCR13A2
MiMammal primers are capable of amplifying and distinguishing a diverse group of
mammalian species. In addition, analyses of water samples from zoo cages of mam-
mals with known species composition suggested that MiMammal primers could suc-
cessfully detect mammalian species from water samples in the field. Then, we
performed an experiment to detect mammals from natural ecosystems by collecting
five 500-ml water samples from ponds in two cool-temperate forests in Hokkaido,
northern Japan. MiMammal amplicon libraries were constructed using eDNA
extracted from water samples, and sequences generated by Illumina MiSeq were
subjected to data processing and taxonomic assignment. We thereby detected mul-
tiple species of mammals common to the sampling areas, including deer (Cervus

Mol Ecol Resour. 2017;17:e63–e75. wileyonlinelibrary.com/journal/men © 2017 John Wiley & Sons Ltd | e63
e64 | USHIO ET AL.

nippon), mouse (Mus musculus), vole (Myodes rufocanus), raccoon (Procyon lotor), rat
(Rattus norvegicus) and shrew (Sorex unguiculatus). Many previous applications of the
eDNA metabarcoding approach have been limited to aquatic/semiaquatic systems,
but the results presented here show that the approach is also promising even for
forest mammal biodiversity surveys.

KEYWORDS
environmental DNA, Forest, Illumina MiSeq, Mammal, metabarcoding, terrestrial ecosystem

1 | INTRODUCTION frequent opportunities to contact aquatic systems. For example,

Environmental DNA (eDNA) is genetic material that is found in an
environment and derived from organisms living in that habitat, and
researchers have recently been using eDNA to detect the presence
of macro-organisms, particularly those living in an aquatic environ-
ment. In the case of macro-organisms, eDNA originates from various
sources such as metabolic waste, damaged tissue or sloughed skin
cells (Kelly et al., 2014a), and the eDNA contains information about
the species identity of organisms that produced it. Ficetola, Miaud,
Pompanon, and Taberlet (2008) first demonstrated the usefulness of
eDNA for detecting the presence of an aquatic vertebrate (American
bullfrog invasive in France) in natural wetlands. Subsequently, eDNA
in aquatic ecosystems has been repeatedly used as a tool for moni-
toring the distributions of fish species in ponds, rivers and seawater
(e.g., Jerde et al., 2013; Minamoto, Yamanaka, Takahara, Honjo, &
Kawabata, 2011; Sigsgaard, Carl, Møller, & Thomsen, 2015; Yama-
moto et al., 2016) as well as the distributions of other aquatic/semi-
aquatic nonfish vertebrates such as giant salamanders (Fukumoto,
Ushimaru, & Minamoto, 2015). Those studies have shown that the
use of eDNA can be a promising approach for an efficient noninva-
sive monitoring tool for biodiversity in aquatic ecology.
Although earlier studies used quantitative PCR and species/
taxon-specific primers to amplify a particular region of eDNA (e.g.,
Ficetola et al., 2008; Fukumoto et al., 2015; Minamoto et al., 2011;
Rodgers & Mock, 2015; Takahara, Minamoto, & Doi, 2013),
researchers have begun to apply massively parallel sequencing tech-
nology (e.g., Illumina MiSeq) and universal primer sets for eDNA
studies (e.g., Bista et al., 2017; Cannon et al., 2016; Deiner, Fron-
hofer, M€achler, Walser, & Altermatt, 2016; Miya et al., 2015; Taber-
let, Coissac, Pompanon, Brochmann, & Willerslev, 2012; Yamamoto
et al., 2017). This approach enables detection of multiple species at
one time (e.g., more than 100 species) and has greatly improved the
efficiency of eDNA detection. A previous study demonstrated that
an eDNA metabarcoding approach using fish-targeting universal pri-
mers (MiFish primers) enabled the detection of more than 230 fish
species from seawater in a single study (Miya et al., 2015). Accord-
ingly, the eDNA metabarcoding approach has become a cost- and
labour-effective approach for capturing aquatic biodiversity.
The use of eDNA is, however, not necessarily limited to aquatic/
semi-aquatic vertebrates, because terrestrial animals also have
Rodgers and Mock (2015) tested the potential of drinking water for
mammals as a medium that contains eDNA originating from terres-
trial mammals. They hypothesized that when a terrestrial mammal
(coyote; Canis latrans) drinks from a water source, DNA from its sal-
iva and mouth tissues is shed and can be used as a tool for species
identification. In their study, they successfully amplified a specific
region of eDNA using carnivore-specific primers. In addition, some
recent studies showed that eDNA of terrestrial animals (e.g., mam-
mals and arthropods) living in a watershed area can be detected
from river/lake water samples (Bista et al., 2017; Cannon et al.,
2016; Deiner et al., 2016), and other studies used eDNA extracted
from soil samples to detect terrestrial animals (Andersen et al., 2012;
Ficetola et al., 2015). Those studies demonstrated that eDNA is a
potentially useful tool even for detecting and monitoring the pres-
ence of mammals living in a terrestrial ecosystem if one can collect
appropriate media that contain mammalian DNA. Mammals living in
a forest ecosystem often contact/utilize water in a pond (Matsub-
ayashi et al., 2006), and thus, a forest pond may provide a source
for researchers to collect mammalian eDNA efficiently.
In this study, we hypothesized that, when mammals utilize/con-
tact water, they shed their tissues into the water as a source of
eDNA, and that this eDNA can be used to detect the presence of
mammals living in a forest ecosystem. To improve the efficiency of
the eDNA approach, we modified the previously developed fish-tar-
geting universal primers (MiFish primers: Miya et al., 2015) by
accommodating them to mammal-specific variations, and conducted
mammalian eDNA metabarcoding. We performed three analyses to
test the versatility of the designed primers: (i) in silico examinations
of the primers, (ii) amplification of extracted tissue DNAs of mam-
mals belonging to various taxa and (iii) field tests in which we anal-
ysed water samples from zoo cages containing mammals of known
species composition. Finally, we examined the effectiveness of the
new primer set using water samples from natural forest ecosystems.
2 | MATERIALS AND METHODS
All of the critical information regarding our study is described below,
but is also listed in Table S1 to facilitate comparisons with other
studies, following the recommendations of Goldberg et al. (2016).
USHIO ET AL. | e65

Using custom Ruby and Python scripts, the number of mismatches

2.1 | Primer design
To facilitate primer design based on comparisons of diverse mam-
malian sequences, we first batch downloaded 741 mammalian
sequences from RefSeq (https://www.ncbi.nlm.nih.gov/refseq/) on
June 9, 2015. Then, the base composition for a selected position in a
conserved region was shown using the “Show Selection Summary
Strip” function in MESQUITE (Maddison & Maddison, 2011). The base
compositions in selected regions were manually recorded in a spread-
sheet for use for the primer design. In the primer design process, we
took into consideration a number of technical tips that enhance the
primer annealing to the template without the use of degenerate
bases (Palumbi, 1996); primers should include some G/C at the 30 -
ends to strengthen primer-template annealing at this position, but a
string of either Gs or Cs at the 30 -end should be avoided; considering
the unconventional base pairing in the T/G bond, the designed pri-
mers use G rather than A when the template is variably C or T, and T
rather than C when the template is variably A or G; G/C contents of
the primers fall between 40 and 60% with an almost identical melting
temperature (Tm). Tm was calculated using a nearest-neighbour ther-
modynamic model implemented in OLIGOCALC (Kibbe, 2007).
Taking into consideration the basic tips described above, we
designed our primers by modifying previously developed MiFish pri-
mers (Miya et al., 2015), which corresponded to regions in the mito-
chondrial 12S rRNA gene (median insert length = ~171 bp), and we
named our primers MiMammal-U (Table 1). In addition, we also
designed MiMammal-E and MiMammal-B primers after preliminary
experiments to accommodate sequence variations in the priming sites
of elephants and bears, respectively. Primer sequences with MiSeq
adaptors (for the first- and second-round PCR) are listed in Table 1.
2.2 | In silico evaluation of interspecific variation of
MiMammal sequences
The binding capacity of MiMammal-U primers was computationally
evaluated using the 741 batch-downloaded mammalian sequences.
between MiMammal-U primers and the mammalian sequences was
calculated.
Interspecific differences within the amplified DNA sequences are
required to assign taxonomic categories. Levels of interspecific varia-
tion in the target region (hereafter called “MiMammal sequence”)
across different taxonomic groups of mammals were computationally
evaluated using the 741 downloaded mammalian sequences. After
checking for duplicate sequences (e.g., multiple sequences from sub-
species) and annotation errors, 740 MiMammal sequences were
extracted and subjected to calculation of pairwise edit distances
using custom Python scripts. The edit distance quantifies dissimilarity
of sequences in bioinformatics and is defined as the minimum num-
ber of single-nucleotide substitutions, insertions or deletions that are
required to transform one sequence into the other.
In addition, the binding capacity and the levels of interspecific
variations of the target region were further evaluated using the “pri-
merTree” package (Cannon et al., 2016) of R version 3.3.1 (R Core
Team, 2016). Briefly, primerTree performs the following analysis: (i)
in silico PCR against the NCBI database; (ii) retrieval of DNA
sequences predicted to be amplified; (iii) taxonomic identification of
these sequences; (iv) multiple DNA sequence alignment; (v) recon-
struction of a phylogenetic tree; and (vi) visualization of the tree
with taxonomic annotation. Using the primerTree package, species
that can be amplified, phylogenetic relationships among the amplified
species, and interspecific variations in the amplified sequences can
be rapidly visualized. Further information and instructions for using
primerTree package can be found in Cannon et al. (2016).
2.3 | Primer testing with extracted DNA
We tested the versatility of MiMammal-U (no adapter sequences)
using extracted DNA from 25 species representing major groups of
mammals (Table 2). Double-stranded DNA concentrations from these
samples were measured with a NanoDrop Lite spectrophotometer
(Thermo Fisher Scientific, Wilmington, DE, USA), and the extracted

T A B L E 1 Detailed information for MiMammal primers

Primer name Information
Primers for the first PCRa,b (with MiSeq sequencing primer and six random bases)
MiMammal-U (forward) ACACTCTTTCCCTACACGACGCTCTTCCGATCT NNNNNN GGGTTGGTAAATTTCGTGCCAGC
MiMammal-U (reverse) GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT NNNNNN CATAGTGGGGTATCTAATCCCAGTTTG
MiMammal-E (forward) ACACTCTTTCCCTACACGACGCTCTTCCGATCT NNNNNN GGACTGGTCAATTTCGTGCCAGC
MiMammal-E (reverse) GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT NNNNNN CATAGTGAGGTATCTAATCTCAGTTTG
MiMimmal-B (forward) ACACTCTTTCCCTACACGACGCTCTTCCGATCT NNNNNN GGGTTGGTTAATTTCGTGCCAGC
MiMimmal-B (reverse) GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT NNNNNN CATAGTGGGGTATCTAATCCCAGTTTG
Primers for the second PCRc,d
2nd PCR-F AATGATACGGCGACCACCGAGATCTACAC XXXXXXXX ACACTCTTTCCCTACACGACGCTCTTCCGATCT
2nd PCR-R CAAGCAGAAGACGGCATACGAGAT XXXXXXXX GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT
a
Italic characters indicate the MiSeq sequencing primers.
b
Bold Ns indicate random bases to improve the quality of MiSeq sequencing.
c
Bold Xs indicate index sequences to identify each sample.
d
Underlined characters indicate P5/P7 adapter sequences for MiSeq sequencing.
e66 | USHIO ET AL.

T A B L E 2 Extract DNAs used to test the performance of the developed primer set
Common name Scientific name Order Family Accession no.
Opossum Monodelphis domestica Didelphimorphia Didelphidae LC104790
Cape hyrax Procavia capensis Hyracoidea Procaviidae LC104791
Asian elephant Elephas maximus Proboscidea Elephantidae LC104792
African elephant Loxodonta africana Proboscidea Elephantidae LC104793
Dugong Dugong dugon Sirenia Dugongidae LC104794
Aardvark Orycteropus afer Tubulidentata Orycteropodidae LC104795
Lesser hedgehog tenrec Echinops telfairi Afrosoricida Tenrecidae LC104796
Brazilian three-banded armadillo Tolypeutes tricinctus Cingulata Dasypodidae LC104797
Giant anteater Myrmecophaga tridactyla Pilosa Myrmecophagidae LC104798
Lesser anteater Tamandua tetradactyla Pilosa Myrmecophagidae LC104799
Northern treeshrew Tupaia belangeri Scandentia Tupaiidae LC104800
Cat Felis catus Carnivora Felidae LC104801
Sea otter Enhydra lutris Carnivora Mustelidae LC104802
European otter Lutra lutra Carnivora Mustelidae LC104803
Minke whale Balaenoptera acutorostrata Cetartiodactyla Balaenopteridae LC104804
Sei whale Balaenoptera borealis Cetartiodactyla Balaenopteridae LC104805
Bryde’s whale Balaenoptera brydei Cetartiodactyla Balaenopteridae LC104806
Cattle Bos taurus Cetartiodactyla Bovidae LC104807
Sheep Ovis aries Cetartiodactyla Bovidae LC104808
Pantropical spotted dolphin Stenella attenuata Cetartiodactyla Delphinidae LC104809
Pantropical spotted dolphin Stenella attenuata Cetartiodactyla Delphinidae LC104810
Sperm whale Physeter macrocephalus Cetartiodactyla Physeteroidea LC104811
Wild boar Sus scrofa Cetartiodactyla Suidae LC104812
Ryukyu flying fox Pteropus dasymallus Chiroptera Pteropodidae LC104813
Horse Equus caballus Perissodactyla Equidae LC104814

DNA was diluted to 15 ng/ll using Milli-Q water. PCR was carried
out with 30 thermal cycles of a 15-ll reaction volume containing
4.5 ll sterile distilled H2O, 7.5 ll 2 9 Gflex PCR Buffer (Mg2+,
dNTPs plus) (Takara, Otsu, Japan), 0.7 ll of each primer (5 lM),
0.3 ll Taq polymerase (Tks Gflex DNA Polymerase; Takara) and
1.2 ll template. The thermal cycle profile after an initial 1-min
denaturation at 94°C was as follows: denaturation at 98°C for 10 s;
annealing at 50°C for 10 s; and extension at 68°C for 10 s, with a
final extension at the same temperature for 7 min.
2.4 | Study site and water sampling for primer
testing with eDNA from zoo samples
To test the versatility of the newly designed primers for metabarcod-
ing eDNA from mammals, we sampled water from cages in Zoorasia
Yokohama Zoological Gardens, Yokohama, Japan (35˚290 4200 N,
139˚310 3500 E). We chose this zoo as a sampling site because the infor-
mation about the fauna in each cage was precisely known, and
because this zoo housed diverse taxonomic groups of animals (i.e.,
>100 species, including many mammals, were housed there). Ten
cages were selected as sampling places; cages of brown fur seal (Arcto-
cephalus pusillus), black rhinoceros (Diceros bicormis), Asian elephant
(Elephas maximus), Japanese macaque (Macaca fuscata), red kangaroo
(Macropus rufus), lion (Panthera leo), tiger (Panthera tigris), Malayan
tapir (Tapirus indicus) and polar bear (Ursus maritimus), and the
“Savanna cage” (in which four animal species, zebra [Equus quagga], gir-
affe [Giraffa camelopardalis], eland [Traurotragus oryx] and cheetah
[Acinonyx jubatus] were housed in the same cage) (Figure 1a–d). These
samples represent diverse taxonomic groups of mammals (Table 3).
Water samples were collected from either a pool (for brown fur
seal, tiger and polar bear), a small bathing place (for Asian elephant
and Malayan tapir), drinking water (for black rhinoceros, red kanga-
roo, lion and Savanna cage) or a moat surrounding a cage (for Japa-
nese macaque). All sampling and filtering equipment was washed
with a 10% commercial bleach solution before use. Approximately
500-ml water samples were collected using polyethylene bottles.
Two negative controls (distilled water) were taken to the zoo to
monitor contaminations during water sampling and transportation.
2.5 | Study site and water sampling for primer
testing with eDNA from natural forest ecosystems
To test the potential usefulness of the MiMammal primers in a natu-
ral ecosystem, we collected water from ponds in two cool-temperate
forests in Hokkaido, Japan (Teshio Research Forest, 44˚540 5500 N,
142˚10 2100 E; Tomakomai Research Forest, 42˚390 3200 N, 141˚360 1900 E)
USHIO ET AL. | e67

(a) (b)

(c) (d)
F I G U R E 1 Example images of target
mammal species. a–d, Mammals in the zoo
cages. Asian elephant, Elephas maximus (a),
tiger, Panthera tigris (b), Malayan tapir,
Tapirus indicus (c) and Zebra, Equus quagga,
in the Savanna cage (d). Approximate body
sizes of animals are as follows: 6 m (Asian
elephant; from the head to the rump),
70 cm (tiger), 2 m (Malayan tapir) and
2.5 m (Zebra). Field survey in cool-
temperate forests in Hokkaido, Japan.
Location of the two cool-temperate forests (e) (f) (g)
(e). The upper and lower black triangles
indicate Teshio and Tomakomai
Experimental Forests, respectively. One of
our sampling ponds in Teshio (f)
(approximately 60 cm from the bottom
side to the other side). Sika deer, Cervus
nippon, commonly found in Hokkaido (g).
Approximate body height of Sika deer is
1.5 m (from the top of head to the
forefoot)

to preliminarily examine the use of the primers for metabarcoding
eDNA from natural forest ponds with unknown mammal composi-
tions and abundances in an open ecosystem (Figure 1e,g). We
collected five water samples from four ponds in the two cool-tempe-
rate forests (in Teshio, one sample from Pond 1, and two samples
from Pond 2; in Tomakomai, one sample from each pond). Two neg-
ative controls (distilled water) were taken to the forests to monitor
contaminations during water sampling and transportation (one was
taken to Teshio forest, and the other was taken to Tomakomai for-
est). The water samples were transported to the laboratory. While
they were transported, the samples were kept at 4°C (for up to
48 hr before filtration) to minimize eDNA degradation.
2.6 | DNA extraction
The collected water samples were taken back to the laboratory and fil-
tered using 47-mm-diameter glass-fibre filters (nominal pore size,
0.7 lm; Whatman, Maidstone, UK). The sampling bottles were shaken
vigorously before the filtration. After the filtration, each filter was
wrapped in commercial aluminium foil and stored at –20°C before eDNA
extraction. Five hundred ml of Milli-Q water was used as the negative
control, and sampling bottles filled with the Milli-Q water were treated
identically to the eDNA samples, in order to monitor contamination dur-
ing the bottle handling, water filtering and subsequent DNA extraction.
DNA was extracted from the filters using a DNeasy Blood and Tis-
sue Kit (Qiagen, Hilden, Germany) in combination with a spin column
(EZ-10; Bio Basic, Markham, Ontario, Canada). After removing the
attached membrane from the spin column (EZ-10; note that this spin
column is not included in the DNeasy Blood and Tissue Kit, and thus,
we separately prepared the additional spin column), the filter was
tightly folded into a small cylindrical shape and placed in the spin col-
umn. The spin column was centrifuged at 6,000 g for 1 min to remove
excess water from the filter. The column was then placed in a new 2-ml
tube, and materials on the column were subjected to proteolysis using
proteinase K. Before the proteolysis, Milli-Q water (300 ll), proteinase
K (10 ll) and buffer AL (100 ll) were mixed and the mixed solution was
gently pipetted onto the folded filter in the spin column. The column
was then placed on a 56°C preheated aluminium heat block and incu-
bated for 15 min. After the incubation, the spin column was cen-
trifuged at 6,000 g for 1 min to collect DNA. To increase DNA yields
from the filter, 200 ll of sterilized TE buffer was gently pipetted onto
the folded filter and the spin column was again centrifuged at 6,000 g
for 1 min. The collected DNA solution (~500 ll) was purified using a
DNeasy Blood and Tissue Kit following the manufacturer’s protocol.
e68 | USHIO ET AL.

T A B L E 3 Classification of the target mammal species in the 3 min denaturation at 95°C was as follows (35 cycles): denaturation
Zoorasia experiment at 98°C for 20 s; annealing at 65°C for 15 s; and extension at 72°C
Common for 15 s, with a final extension at the same temperature for 5 min.
name Species name Order Family We performed triplicate first PCR, and those replicates were pooled
Brown fur Arctocephalus Carnivora Otariidae to mitigate the PCR dropouts. The pooled first PCR products were
seal pusillus purified using Exo-SAPIT (Affymetrix, Santa Clara, CA, USA). The
Black Diceros bicornis Perissodactyla Rhinocerotidae pooled, purified, and 10-fold diluted first PCR products were used as
rhinoceros
templates for the second PCR.
Asian Elephas maximus Proboscidea Elephantidae The second-round PCR (second PCR) was carried out with a 24-
elephant
ll reaction volume containing 12 ll of 2 9 KAPA HiFi HotStart
Japanese Macaca fuscata Primates Cercopithecidae
ReadyMix, 1.4 ll of each primer (5 lM primer F/R), 7.2 ll of steril-
macaque
ized distilled H2O and 2.0 ll of template. Different combinations of
Red Macropus rufus Diprotodontia Macropodidae
kangaroo
forward and reverse indices were used for different templates (sam-
ples) for massively parallel sequencing with MiSeq. The thermal cycle
Lion Panthera leo Carnivora Felidae
profile after an initial 3-min denaturation at 95°C was as follows (12
Tiger Panthera tigris Carnivora Felidae
cycles): denaturation at 98°C for 20 s; annealing and extension com-
Malayan Tapirus indicus Perissodactyla Tapiridae
tapir
bined at 65°C (shuttle PCR) for 15 s with the final extension at
72°C for 5 min.
Polar bear Ursus maritimus Carnivora Ursidae
The indexed second PCR products were pooled, and the pooled
Zebra Equus quagga Perissodactyla Equidae
libraries were purified using a GeneRead Size Selection Kit (Qiagen,
Giraffe Giraffa camelopardalis Artiodactyla Giraffidae
Hilden, Germany). The volume of each sample added to the library
Eland Taurotragus oryx Artiodactyla Bovidae
was determined depending on the concentration of each second
Cheetah Acinonyx jubatus Carnivora Felidae
PCR product (i.e., the volume added to the library was appropriately
reduced for a high concentration PCR product). A target size DNA
of the purified library (~370 bp) was excised using E-Gel SizeSelect
2.7 | Paired-end library preparation
(ThermoFisher Scientific, Waltham, MA, USA). The DNA size distri-
As a preliminary experiment suggested that the newly designed uni- bution of the library was estimated using an Agilent 2100 BioAna-
versal primer set could not effectively amplify elephant or bear lyzer (Agilent, Santa Clara, CA, USA), and the double-stranded DNA
sequences, two complementary primer sets for elephants and bears concentration of the library was quantified using a Qubit dsDNA HS
were designed (MiMammal-E-F/R, MiMammal-B-F/R). MiSeq assay kit and a Qubit fluorometer (ThermoFisher Scientific, Waltham,
sequencing primers and six random bases (N) were combined with MA, USA). The double-stranded DNA concentration of the library
MiMammal-U/E/B primers (hereafter, MiMammal-mix, see Table 1 was then adjusted to 2 nM using Milli-Q water, and the DNA library
for details), and subjected to multiplex PCR. The six random bases was applied to the MiSeq platform (Illumina, San Diego, CA, USA).
were used to enhance cluster separation on the flow cells during ini- The sequencing was performed using a MiSeq Reagent Kit Nano v2
tial base call calibrations on the MiSeq platform. for 2 9 150 bp PE (Illumina, San Diego, CA, USA). A 20% PhiX DNA
The work-space and equipment were sterilized prior to the spike-in control was added to improve the data quality of low diver-
library preparation, filter-tips were used for pipetting, and separation sity samples such as single PCR amplicons used in this study. Two
of pre- and post-PCR was carried out to safeguard against cross- negative controls were included in the PCR steps (throughout the
contamination. We used a two-step PCR protocol for the library first to second PCR and MiSeq sequencing), and generated negligible
preparation. We appended index sequences (for the identification of sequences after data processing (i.e., which were less than 0.04% of
different samples) at the second PCR step to reduce variations in total sequences, and which were identified as mouse sequences).
amplification across samples caused by different combinations of Note that, although our MiSeq run yielded approximately 1,000,000
index sequences (O’Donnell, Kelly, Lowell, & Port, 2016). We reads (an average number of reads for the MiSeq Reagent Kit Nano
employed two negative controls to monitor contamination during v2 for 2 9 150 bp PE), the total number of raw reads reported in
the experiments. this study is 618,767 and the rest of the reads correspond to those
The first PCR was carried out with a 12-ll reaction volume con- from other studies.
taining 6.0 ll of 2 9 KAPA HiFi HotStart ReadyMix (KAPA Biosys-
tems, Wilmington, WA, USA), 0.7 ll of MiMammal-mix primer (5 lM
2.8 | Data processing and taxonomic assignment
primer F/R), 2.6 ll of sterilized distilled H2O and 2.0 ll of template.
When the first PCR was multiplexed (i.e., when it was treated using The overall quality of the MiSeq reads was evaluated using the pro-
MiMammal-mix), the final concentration of each primer (MiMammal- grams FASTQC (available from http://www.bioinformatics.babraham.ac.
U/E/B-F/R) was 0.1 lM (0.6 lM total concentration of six primers, uk/projects/fastqc/) and SUGAR (Sato et al., 2014). After confirming
MiMammal-U/E/B-F/R). The thermal cycle profile after an initial a lack of technical errors in the MiSeq sequencing, low-quality tails
USHIO ET AL.
T A B L E 4 Binding capacity of MiMammal-U primers and frequency distributions of the interspecific edit distances of the primer set against
mammal sequences
Edit distance 0 1
Binding capacity of MiMammal-U primer set
MiMammal-U-F: GGGTTGGTAAATTTCGTGCCAGC
G/T pairs not accepted 331 255
G/T pairs accepted 516 171
MiMammal-U-R: CATAGTGGGGTATCTAATCCCAGTTTG
G/T pairs not accepted 664 58
G/T pairs accepted 676 58
Frequency distributions of the interspecific/genus edit distances of the insert sequence
Species 77 41
Genus 3 6
were trimmed from each read using DynamicTrim.pl from SOLEXAQA
software package (Cox et al., 2010) with a cut-off threshold set at a
Phred score of 10 (= 101 error rate). The tail-trimmed pair-end
reads were assembled using the software FLASH with a minimum
overlap of 10 bp. The assembled reads were further filtered using
custom Perl scripts to remove reads with ambiguous sites or with
unusual lengths compared to the expected size of the PCR ampli-
cons. Finally, the software TAGCLEANER (Schmieder, Lim, Rohwer, &
Edwards, 2010) was used to remove primer sequences with a maxi-
mum of three-base mismatches and to transform the FASTQ format
into FASTA. In general, the sequence quality produced by our proce-
dures was high (i.e., more than 95% of raw reads passed the filtering
process; Table S2).
The preprocessed reads from the above custom pipeline were
dereplicated using a “derep_fulllength” command in UCLUST (Edgar,
2010), with the number of identical reads added to the header line
of the FASTA formatted data file. Those sequences represented by
at least 10 identical reads were subjected to the downstream analy-
ses, and the remaining under-represented sequences (with fewer
than 10 identical reads) were subjected to pairwise alignment using
a “usearch_global” command in UCLUST. If the latter sequences
observed from fewer than 10 reads showed at least 99% identity
with one of the former reads (one or two nucleotide differences),
they were operationally considered as identical (owing to sequencing
or PCR errors and/or actual nucleotide variations in the populations)
and they were added to the group of sequences having at least 10
reads.
The processed reads were subjected to local BLASTN searches
(Camacho et al., 2009) against a custom-made database. The latter
was generated by downloading all whole mitogenome sequences
from Sarcopterygii deposited in NCBI Organelle Genome Resources
(http://www.ncbi.nlm.nih.gov/genomes/OrganelleResource.cgi?taxid=
8287). As of 15 March 2016, the database covers 1,881 species
across a wide range of families and genera. In addition, the custom
database was supplemented by all whole and partial fish mitogen-
ome sequences deposited in MitoFish (Iwasaki et al., 2013) to
cover unexpected fish detection using the MiMammal-mix primer
set.
| e69
2 3 4 ≥5 Total
75 50 28 2 741
43 11 0 0 741
15 3 0 1 741
6 0 0 1 741
94 101 119 273,738 274,170
12 28 33 82
The top BLAST hit with a sequence identity of at least 97% and
E-value threshold of 105 was applied for species assignments of
each representative sequence. The reliability of the species assign-
ments was evaluated based on the ratio of total alignment length
and number of mismatched bases between the query and reference
sequences. For example, if a query sequence was aligned to the top
BLAST hit sequence with an alignment length of 150 bp with one
mismatch present, the ratio was calculated as 150/(1 + 1). The value
one is added to the denominator to avoid zero-divisors. This value
(e.g., 150/(1 + 1)) was calculated for the top and second BLAST hit
species, and the ratio between these values was used as an indicator
of the relative support for the species assignment. Results from the
BLAST searches were automatically tabulated, with scientific names,
common names, total number of the reads and representative
sequences noted in an HTML format. The above bioinformatics pipe-
line from data preprocessing through taxonomic assignment is also
available in the supplemental information in a previous study (Miya
et al., 2015).
3 | RESULTS
3.1 | Tests of versatility of designed primers in
silico and using extracted DNA
First, the performance of MiMammal-U primers was tested in silico
(Table 4). When G/T pairs were accepted, MiMammal-U-F and
MiMammal-U-R perfectly matched 516 and 676 of 741 species,
respectively. Approximately 92% and 99% of the 741 species
showed less than 1 mismatch. In addition, interspecific/genus differ-
ences in the edit distance were calculated, and most of the species
pairs showed an edit distance larger than 5 (Table 4). Analyses per-
formed with the primerTree package also suggested that the primers
can amplify/distinguish a diverse group of mammalian species
(Fig. S1). The capacity to distinguish mammalian species (evaluated
by the length of the branch) is equal to, or greater than, that of
other commonly used mammalian primers targeting mitochondrial
16S rRNA (Boessenkool et al., 2012; Cannon et al., 2016). As for the
off-target amplification, MiMammal-U amplifies several groups of
e70 | USHIO
fish, bird, and reptile species, while the mitochondrial 16S rRNA pri-
mer set amplifies several groups of fish and amphibian species
(Fig. S1; Cannon et al., 2016).
Second, the performance of MiMammal-U primers was evaluated
using 25 extracted mammalian DNA samples. All of the extracted
DNAs were successfully amplified, and those sequences were depos-
ited in DDBJ/Emble/GenBank databases (Table 2).
3.2 | Primer testing with eDNA from zoo samples
As a result of MiSeq sequencing and data preprocessing, 10 samples
generated 509,497 sequences (Table 5). Among the 10 zoo cages
(13 species in total) examined, 10 species were successfully
detected. For brown fur seal (Arctocephalus pusillus), black rhinoceros
(Diceros bicornis), Asian elephant (Elephas maximus), red kangaroo
(Macrofus rufus), lion (Panthera leo), tiger (Panthela tigris) and Malayan
tapir (Tapirus indicus), large fractions of the total sequence counts
were of target species origin (at least > 30%, Figure 2). For the
Savanna cage sample, however, only 6.6% of total sequences were
of target species (zebra, giraffe and eland) origin. We were unable to
detect the sequences of cheetah (Acinonyx jubatus), Japanese maca-
que (Macaca fuscata) and polar bear (Ursus maritimus) in the first
trial.
Possible cause(s) of the nondetection of some mammal species
may have lain in the experimental conditions (e.g., PCR conditions).
To examine this possibility, we modified the annealing temperature
and primer concentrations of the first PCR. After several trial-and-
error attempts, we detected 1,400 reads of Japanese macaque
sequence (from 13,259 total reads) from the Japanese macaque
water sample and 187 reads of cheetah sequence (from 51,118 total
reads) from the savanna water sample under the conditions of 62°C
annealing temperature and a total 15 lM MiMammal-mix primer set
(each primer at 5 lM). We also detected 6,990 reads of polar bear
sequence (from 61,347 total reads) from the polar bear pool sample
under the conditions of annealing temperature 62°C and 5 lM
MiMammal-B primer set (these sequences were also deposited; see
Data Accessibility).
We also frequently detected the sequences of nontarget species.
In the zoo experiment, the most frequently detected nontarget spe-
cies were human being and mouse (Figure 2). In addition, feeds of
the target species were often detected (Figure 2; horse sequences
[Equus caballus] detected in the lion and tiger cages, and Japanese
quail sequences [Coturnix japonicus] detected in the Savanna cage),
which indicates that eDNA from dead organisms can also be
detected easily (see also Kelly, Port, Yamahara, & Crowder, 2014b).
On average, approximately 10% of the total number of reads
were considered to have originated from cross-contamination
(i.e., sequences were from unknown origin and/or other samples,
Figure 2 and Table S3). In addition, two negative controls (distilled
water) were taken to the field to monitor contaminations during the
sampling, transportation and experiments. One negative control gen-
erated some human and mouse sequences, which possibly contami-
nated this sample during the filtering and DNA extraction steps, but
ET AL.
did not generate any other sequences. The other negative control
generated rhinoceros sequences (948 sequences, 1.6% rhinoceros
sequences [among 59,842 sequences] detected in the rhinoceros-
positive sample) as well as human and mouse sequences, suggesting
the occurrence of cross-contamination.
3.3 | Detection of eDNA of terrestrial mammals
from natural forest water
Five samples from primary forest ponds generated 75,214 reads as a
result of MiSeq sequencing and data processing (Table 6). The per-
centage of forest mammalian sequences ranged from 15 to 89% in
the five samples (Table 6). In all of these samples, house mouse (Mus
musculus) sequences were frequently detected. From Pond 2 in
Teshio Forest, sequences of grey-sided vole (Myodes rufocanus), Nor-
way rat (Rattus norvegicus) and long-clawed shrew (Sorex unguicula-
tus), which are commonly observed mammals in this forest
(Table S4), were detected. From two ponds in Tomakomai forest,
Sika deer (Cervus nippon) and common raccoon (Procyon lotor)
sequences were detected, and these species are indeed common in
this area (Figure 1g and Table S4). Over 96% of “other sequences”
were human sequences (Table 6). We also detected sequences of
brown fur seal (1,545 reads) and black rhinoceros (22 reads) from
Teshio Pond 1, which indicates cross-contaminations from zoo sam-
ples. We did not detect any other cross-contamination in forest
pond samples. Two negative controls (distilled water) were taken to
the field (one each was taken to Teshio forest and to Tomakomai
forest) to monitor contamination. Human and mouse sequences
were detected from two negative controls, which were possibly con-
taminated in filtering and DNA extraction steps.
4 | DISCUSSION
4.1 | Tests of versatility of designed primers in
silico and using extracted DNA
In silico evaluation of the binding capacity of the primers and the
amplification of extracted DNA suggested that MiMammal-U primers
can amplify DNA of a diverse group of mammals (Tables 2 and 4). In
addition, in silico evaluations of the levels of interspecific variation of
the target region suggested that these primers designed here were
capable of distinguishing diverse mammalian species (Table 4). Pri-
merTree analysis also suggested that MiMammal-U primers can
amplify and distinguish DNA from a diverse group of mammals as
well as several fish and bird sequences (Fig. S1).
4.2 | Detection of target species from zoo cages
In the zoo experiment, we detected 10 of 13 mammalian species
(Table 5) in the first trial. Seven of the 10 detected mammals,
namely, brown fur seal, black rhinoceros, Asian elephant, red kanga-
roo, lion, tiger and Malayan tapir, frequently contacted the water in
their cages (for bathing or drinking), and perhaps not surprisingly,
 USHIO
ET AL.

T A B L E 5 Sequence reads of detected species from water samples collected in the zoo
Name of animal species living in a cage where a water sample was collected

Black Asian Japanese Water in

Brown fur seal rhinoceros elephanta macaque Red kangaroo Lion Tigerb Malayan tapir Polar bear Savanna cage
Arctocephalus Diceros Elephas Macaca Macropus Panthera Panthera Tapirus Ursus 4 spp. mixturec
Detected species Species name pusillus bicornis maximus fuscata rufus leo tigris indicus maritimus
Brown fur seal Arctocephalus pusillus 19,851 216 0 0 21 0 0 0 0 362
Black rhinoceros Diceros bicornis 0 59,842 0 0 0 0 0 0 0 0
Asian elephant Elephas maximus 0 0 1,182 0 0 0 34 0 0 0
Japanese macaque Macaca fuscata 0 0 0 0 0 0 0 0 0 0
Red kangaroo Macropus rufus 29 0 0 38 374,132 0 31 0 0 0
Lion Panthera leo 0 0 0 0 0 10,177 0 0 0 0
Tiger Panthera tigris 0 0 0 0 0 0 2,778 0 0 0
Malayan tapir Tapirus indicus 0 0 0 0 0 0 0 372 0 0
Polar bear Ursus maritimus 0 0 0 0 0 0 0 0 0 0
Zebra Equus quagga 0 0 0 0 0 0 0 0 0 18
Giraffe Giraffa camelopardalis 0 0 0 0 0 0 0 0 0 150
Eland Taurotragus oryx 0 0 0 0 0 0 0 0 0 166
Cheetah Acinonyx jubatus 0 0 0 0 0 0 0 0 0 0
Other sequences 0 2,326 406 1,210 24,485 318 5,776 470 359 4,748
Total sequence count 19,880 62,384 1,588 1,248 398,638 10,495 8,619 842 359 5,444
Volume of 2nd PCR product added to the 0.5 10 20 10 10 0.5 10 20 20 10
final library (ll)

Bold values indicate sequence reads of the target species.

a
Indialn elephant (subspecies of Asian elephant).
b
Sumatran tiger (subspecies of tiger).
c
Four species (zebra, giraffe, eland and cheetah) were present in the cage.
|
e71
e72 | USHIO ET AL.

100
Unknown source
Proportion of sequence sources (%)

Other sample
80 Mouse
Human
Feed
60 Target

40

20

0
F I G U R E 2 Proportions (%) of sequence
l
ros

nt

ue

o
lion

ger

ir
ear

r
sea

ate
aro

tap
pha

sources of the zoo samples. Different

caq
oce

n ti

ar b

aw
ian
ang
fur

an

colours indicate different sources, as

ele

ma

atra
rhin

Ind

Pol

ann
lay
wn

dk

described in the legend. Each label on the

ian

se

Sum

Ma
Bro

ck

Sav
Re

x-axis indicates a common name of a

Ind

ane
Bla

mammal living in an enclosure from where

Jap

the water sample was taken

T A B L E 6 Sequence counts of detected species from water samples collected in forest ponds in Hokkaido, Japan
Water samples from forest ponds

Teshio Teshio Teshio Tomakomai Tomakomai

Detected species (common name) Species name Pond 1 Pond 2 (rep1) Pond 2 (rep2) Pond 1 Pond 2
Sika deer Cervus nippon 0 0 0 19,452 1,306
House mouse Mus musculus 1,171 128 701 127 846
Grey red-backed vole Myodes rufocanus 0 497 393 0 0
Common raccoon Procyon lotor 0 0 0 741 1,831
Norway rat Rattus norvegicus 0 873 0 0 0
Long-clawed shrew Sorex unguiculatus 0 0 181 0 0
a
Other sequences 6,740 3,419 156 16,453 20,199
Percentage of other sequences 85% 70% 11% 45% 84%
Total sequence count 7,911 4,917 1,431 36,773 24,182
Volume of 2nd PCR product added to the final library (ll) 20 20 20 10 10
a
Over 96% of other sequences were identified as humans.

such behaviours are likely to have resulted in the high proportion/ compared to the other mammals (zebra, giraffe and eland) in the
number of the target species sequences detected (Table 5, Figure 2). cage. For Japanese macaque, the water sample was taken from a
For the Savanna cage sample, only 6.6% of total sequences were of moat surrounding the cage. This means that the Japanese macaques
target species (zebra, giraffe and eland) origin. The Savanna cage living in the cage cannot directly contact the water. Therefore, non-
occupies a relatively large area, and the mammals in the cage do not detection of the Japanese macaque sequences from the water sam-
frequently contact the water. The large cage area and the animals’ ple may not be surprising. For polar bear, the pool in the cage is
behaviour might have resulted in the low proportion of target spe- always sterilized with a low concentration of chloride (<0.05%) to
cies sequences from this cage. prevent harmful algal blooms, and this chloride might degrade DNA
In contrast to the above-mentioned target species, we could not and thus might have prevented the detection of polar bear
detect the sequences of cheetah (Acinonyx jubatus), Japanese maca- sequences.
que (Macaca fuscata) or polar bear (Ursus maritimus) in the first trial. To examine another possible cause of the nondetection of some
For cheetah, the reason for the nondetection seems to lie in the ani- mammal species, we modified the annealing temperature and primer
mal’s behaviour. The cheetah is often resting on a tree far from the concentrations of the first PCR in the second trial. The results
water, and it has much less opportunity to contact the water showed that a lower annealing temperature (62°C) and a higher
USHIO ET AL.
primer concentration (15 lM in total) enabled the detection of Japa-
nese macaque and cheetah, and that a lower annealing temperature
(62°C) and the use of MiMammal-B only enabled the detection of
polar bear. The number of detected species often increased if a
lower annealing temperature and/or a higher concentration of pri-
mers were chosen, but at the same time, it would increase the possi-
bility of false-positive detection. Therefore, optimal experimental
conditions should be carefully examined, and more conservative con-
ditions (i.e., a lower possibility of false-positive detection) would usu-
ally be recommended. Altogether, MiMammal primers successfully
detected all mammals examined in the zoo cages, but the detection
depended on the experimental conditions.
4.3 | Detection of nontarget species from zoo
cages
The most frequently detected nontarget species was human being
in the zoo experiment (Figure 2). Humans (including staff of the
zoo) are ubiquitous in the sampling area, and they are likely to
have frequent opportunities to contact the water directly/indi-
rectly. Thus, it is difficult to distinguish the origin of the
sequences, that is, whether they came from the sampling site and/
or contaminations during the experimental procedures, unless neg-
ative controls are appropriately included at every experimental step
(e.g., sampling, filtering, DNA extraction and PCR). More impor-
tantly, amplification and sequencing of human sequences (or other
nontarget sequences such as feed DNA; Figure 2) may cause the
suppression of amplification and sequencing of target sequences,
and result in false-negative detection of target species (Boessen-
kool et al., 2012). One solution to avoid amplification of nontarget
sequences is adding blocking primers. Boessenkool et al. (2012)
showed that the application of human-specific blocking primers to
an ancient DNA study improved the detection efficiency of target
species. Therefore, applying human (or other nontarget species)-
specific blocking primers to eDNA metabarcoding studies would
improve the detection efficiency, but we should also recognize
that blocking primers often influence the amplification efficiency of
sequences other than the target sequence (i.e., the sequence that
should be blocked).
As pointed out in a previous study (Miya et al., 2015), the most
serious pitfall of eDNA analysis is the risk of contamination. To avoid
this risk, we performed standard decontamination procedures of the
laboratory spaces and equipment (e.g., routine cleaning of experi-
ment benches using DNA remover, the use of filter-tips and physical
separation of DNA extraction and PCR rooms). Despite these efforts,
on average, approximately 10% of the total number of reads
were considered to have originated from cross-contamination
(i.e., sequences were from unknown origin and/or other samples,
Figure 2 and Table S3). This rate of cross-contamination is in a simi-
lar range as that of other studies using the metabarcoding technique
(Miya et al., 2015). Such contamination issues are among the most
challenging problems that still remain unsolved regarding experimen-
tal applications of the metabarcoding technique.
| e73
4.4 | Detection of mammals from forest ponds
From ponds in Teshio and Tomakomai forests, sequences of grey-
sided vole (Myodes rufocanus), Norway rat (Rattus norvegicus) and
long-clawed shrew (Sorex unguiculatus), and of Sika deer (Cervus nip-
pon) and common raccoon (Procyon lotor) were detected, respectively
(Table 6). Although we did not perform systematic visual/camera
censuses, these species commonly occur in this area (see Supple-
mentary methods and Table S4), indicating successful detection of
several terrestrial mammals inhabiting the sampling area. On the
other hand, several common mammals (e.g., brown bear and red fox
in Teshio forest) were not detected in any of the libraries. This is
not surprising because eDNA detection of terrestrial mammals relies
on contacts of mammals with forest ponds. Opportunities for mam-
mals to contact water in ponds would be spatially and temporally
stochastic to some extent, and thus, the limited number of water
samples collected during the short period of the study time may
have been insufficient to capture mammal diversity in the forests.
Although this study demonstrated the usefulness of terrestrial
water as a trapping medium for eDNA, several issues should be fur-
ther addressed in future studies. For example, the present study as
well as a previous study (Rodgers & Mock, 2015) used water sam-
ples to collect eDNA, but this might not be the most effective
approach. Artificial water that attracts mammals (e.g., water with a
high concentration of nutrients) might be a more effective trapping
medium for mammalian eDNA. In addition, the relationship between
the number of reads and the abundance of mammals in a system is
not necessarily reliable because the input of mammalian eDNA relies
on the contacts of mammals with terrestrial water. Furthermore,
issues such as cross-contaminations and false-positive detection
should be carefully examined, as discussed in previous studies (e.g.,
Boessenkool et al., 2012; Miya et al., 2015).
4.5 | Improvements for larger-scale studies
Unlike traditional monitoring approaches such as automated camera
traps, the eDNA metabarcoding approach does not require any spe-
cial equipment in the field. This will allow taking a “snapshot” of
mammal biodiversity more frequently on a larger spatial scale com-
pared with traditional approaches. A large-scale study (in terms of
spatial and temporal scales) will produce many more samples than
the present study, and several new techniques/tools will help in pro-
cessing large numbers of samples more efficiently and avoiding
cross-contaminations and amplifications of nontarget sequences.
First, degradation of eDNA at the sampling step (i.e., water collection
and transportation), which critically influences detection of rare spe-
cies, can be avoided by addition of cationic surfactant (Yamanaka
et al., 2017). The addition of cationic surfactant was shown to pre-
serve eDNA even at room temperature, and thus, this technique
allows collection of water samples without significant eDNA degra-
dation in a remote area where electricity (for refrigerator) is not
available. Second, using filter cartridges (e.g., SterivexTM Filter Units;
Merck Millipore, Darmstadt, Germany) will reduce the risk of
e74 | USHIO
cross-contamination during water filtration (Miya et al., 2016). Third,
as already discussed in the section above, applying non-target-speci-
fic blocking primers will reduce sequences of nontarget species,
which could improve the detection efficiency of target species
(Boessenkool et al., 2012). Finally, modifications and optimization of
PCR conditions may be required, as we have shown that the detec-
tion of some species depends on PCR conditions. Different PCR
approaches, such as touch-down PCR (Korbie & Mattick, 2008), may
reduce amplifications of nontarget sequences and contribute to
improvement of detection of target species.
5 | CONCLUSION
The present study demonstrated that eDNA combined with our new
primer sets and the MiSeq platform can be a useful tool for detecting
mammals living in a terrestrial ecosystem. Describing and monitoring
mammal diversity in a forest ecosystem is one of the critical steps in bio-
diversity conservation and management, but it is laborious and costly if
one relies on traditional survey methods such as direct visual census
and automated camera methods. The eDNA approach presented here
is noninvasive and efficient. Moreover, it does not require setting spe-
cial equipment in a field, and thus, there is no risk of equipment loss/de-
struction. Therefore, it would be possible to perform a larger-scale
study (in terms of spatial and temporal scales) to monitor mammal biodi-
versity. We propose that the eDNA metabarcoding approach could
serve as an effective tool for taking snapshots of mammal biodiversity
and tracking changes in the biodiversity in terrestrial ecosystems.
ACKNOWLEDGEMENTS
We would like to thank Mr. Noriya Saito, assistant manager of Yoko-
hama Zoological Gardens ZOORASIA, for help in sampling at the
zoo, and Mr. Sho Sakurai for assistance with experiments. This
research was supported by CREST (JPMJCR13A2) from Japan
Science and Technology Agency (JST). This study was approved by
Yokohama Zoological Gardens ZOORASIA. Field sampling in Hok-
kaido was approved by Hokkaido University.
DATA ACCESSIBILITY
DNA sequences: DDBJ Accession nos are DRA004882, PRJDB4990
(BioProject ID) and SAMD00056063–SAMD00056088 (BioSample
ID). The analytical pipeline for DNA sequences is available in Dryad
(https://doi.org/10.5061/dryad.54v2q.2).
AUTHORS’ CONTRIBUTIONS
MU and MM conceived and designed research; MU, HF, TI, OK, KS,
KM, MN and MM performed sampling; MU, HF, TI, TS and MM per-
formed experiments; MU, YS, MT and WI performed data analysis;
MU and MM wrote the early draft and completed it with significant
inputs from all authors.
ET AL.
REFERENCES
Andersen, K., Bird, K. L., Rasmussen, M., Haile, J., Breuning-Madsen, H.,
Kjaer, K. H., . . . Willerslev, E. (2012). Meta-barcoding of “dirt” DNA
from soil reflects vertebrate biodiversity. Molecular ecology, 21,
1966–1979.
Bista, I., Carvalho, G. R., Walsh, K., Seymour, M., Hajibabaei, M., Lallias,
D., . . . Creer, S. (2017). Annual time-series analysis of aqueous eDNA
reveals ecologically relevant dynamics of lake ecosystem biodiversity.
Nature Communications, 8, 14087.
Boessenkool, S., Epp, L. S., Haile, J., Bellemain, E., Edwards, M., Coissac,
E., . . . Brochmann, C. (2012). Blocking human contaminant DNA dur-
ing PCR allows amplification of rare mammal species from sedimen-
tary ancient DNA. Molecular Ecology, 21, 1806–1815.
Camacho, C., Coulouris, G., Avagyan, V., Ma, N., Papadopoulos, J., Bealer,
K., . . . Madden, T. L. (2009). BLAST+: Architecture and applications.
BMC Bioinformatics, 10, 421.
Cannon, M. V., Hester, J., Shalkhauser, A., Ma, N., Papadopoulos, J., Bealer,
K., . . . Madden, T. L. (2016). In silico assessment of primers for eDNA
studies using PrimerTree and application to characterize the biodiver-
sity surrounding the Cuyahoga River. Scientific Reports, 6, 22908.
Cox, M. P., Peterson, D. A., Biggs, P. J. (2010). SolexaQA: At-a-glance
quality assessment of Illumina second-generation sequencing data.
BMC Bioinformatics, 11, 485.
Deiner, K., Fronhofer, E. A., M€achler, E., Walser, J.-C., & Altermatt, F.
(2016). Environmental DNA reveals that rivers are conveyer belts of
biodiversity information. Nature Communications, 7, 12544.
Edgar, R. C. (2010). Search and clustering orders of magnitude faster
than BLAST. Bioinformatics, 26, 2460–2461.
Ficetola, G. F., Miaud, C., Pompanon, F., & Taberlet, P. (2008). Species
detection using environmental DNA from water samples. Biology let-
ters, 4, 423–425.
Ficetola, G. F., Pansu, J., Bonin, A., Coissac, E., Giguet-Covex, C., De
Barba, M., . . . Taberlet, P. (2015). Replication levels, false presences
and the estimation of the presence/absence from eDNA metabarcod-
ing data. Molecular Ecology Resources, 15, 543–556.
Fukumoto, S., Ushimaru, A., & Minamoto, T. (2015). A basin-scale appli-
cation of environmental DNA assessment for rare endemic species
and closely related exotic species in rivers: A case study of giant sala-
manders in Japan. Journal of Applied Ecology, 52, 358–365.
Goldberg, C. S., Turner, C. R., Deiner, K., Klymus, K. E., Thomsen, P.
F., Murphy, M. A., . . . Taberlet, P. (2016). Critical considerations
for the application of environmental DNA methods to detect aqua-
tic species. Methods in Ecology and Evolution, 7, 1299–1307.
Iwasaki, W., Fukunaga, T., Isagozawa, R., Yamada, K., Maeda, Y., . . .
Satoh, T. P. (2013). MitoFish and MitoAnnotator: A mitochondrial
genome database of fish with an accurate and automatic annotation
pipeline. Molecular biology and evolution, 30, 2531–2540.
Jerde, C. L., Chadderton, W. L., Mahon, A. R., Renshaw, M. A., Corush,
J., Budny, M. L., . . . Lodge, D. M. (2013). Detection of Asian carp
DNA as part of a Great Lakes basin-wide surveillance program.
Canadian Journal of Fisheries and Aquatic Sciences, 70, 522–526.
Kelly, R. P., Port, J. A., Yamahara, K. M., & Crowder, L. B. (2014b). Using
environmental DNA to census marine fishes in a large mesocosm.
PLoS ONE, 9, e86175.
Kelly, R. P., Port, J. A., Yamahara, K. M., Martone, R. G., Lowell, N.,
Thomsen, P. F., . . . Crowder, L. B. (2014a). Harnessing DNA to
improve environmental management. Science, 344, 1455–1456.
Kibbe, W. A. (2007). OligoCalc: An online oligonucleotide properties cal-
culator. Nucleic Acids Research, 35, W43–W46.
Korbie, D. J., & Mattick, J. S. (2008). Touchdown PCR for increased
specificity and sensitivity in PCR amplification. Nature Protocols, 3,
1452–1456.
Maddison, W. P., & Maddison, D. R. (2011). Mesquite: A modular system
for evolutionary analysis. http://mesquiteproject.org
USHIO ET AL.
Matsubayashi, H., Lagan, P., Majalap, N., Tangah, J., Sukor, J. R. A., &
Kitayama, K. (2006). Importance of natural licks for the mammals in
Bornean inland tropical rain forests. Ecological Research, 22, 742–
748.
Minamoto, T., Yamanaka, H., Takahara, T., Honjo, M. N., & Kawabata, Z.
(2011). Surveillance of fish species composition using environmental
DNA. Limnology, 13, 193–197.
Miya, M., Minamoto, T., Yamanaka, H., Oka, S., Sato, K., Yamamoto, S.,
. . . Doi, H. (2016). Use of a filter cartridge for filtration of water sam-
ples and extraction of environmental DNA. Journal of Visualized
Experiments, 117, e54741–e54741.
Miya, M., Sato, Y., Fukunaga, T., Sado, T., Poulsen, J. Y., Sato, K., . . .
Iwasaki, W. (2015). MiFish, a set of universal PCR primers for
metabarcoding environmental DNA from fishes: Detection of more
than 230 subtropical marine species. Royal Society open science, 2,
150088.
O’Donnell, J. L., Kelly, R. P., Lowell, N. C., & Port, J. A. (2016). Indexed
PCR primers induce template-specific bias in large-scale DNA
sequencing studies. PLoS ONE, 11, e0148698.
Palumbi, S. R. (1996). Nucleic acids II: The polymerase chain reaction. In
D. M. Hills, C. Moritz, & B. K. Mable (Eds.), Molecular systematics (pp.
205–247). Sunderland, Massachusetts: Sinauer.
R Core Team. (2016). R: A language and environment for statistical com-
puting. Vienna: Austria.
Rodgers, T. W., & Mock, K. E. (2015). Drinking water as a source of envi-
ronmental DNA for the detection of terrestrial wildlife species. Con-
servation Genetics Resources, 7, 693–696.
Sato, Y., Kojima, K., Nariai, N., Yamaguchi-Kabata, Y., Kawai, Y., Taka-
hashi, M., . . . Nagasaki, M. (2014). SUGAR: Graphical user interface-
based data refiner for high-throughput DNA sequencing. BMC geno-
mics, 15, 664.
Schmieder, R., Lim, Y. W., Rohwer, F., & Edwards, R. (2010). TagCleaner:
Identification and removal of tag sequences from genomic and
metagenomic datasets. BMC Bioinformatics, 11, 341.
Sigsgaard, E. E., Carl, H., Møller, P. R., & Thomsen, P. F. (2015). Monitor-
ing the near-extinct European weather loach in Denmark based on
| e75
environmental DNA from water samples. Biological Conservation, 183,
46–52.
Taberlet, P., Coissac, E., Pompanon, F., Brochmann, C., & Willerslev, E.
(2012). Towards next-generation biodiversity assessment using DNA
metabarcoding. Molecular ecology, 21, 2045–2050.
Takahara, T., Minamoto, T., & Doi, H. (2013). Using environmental DNA
to estimate the distribution of an invasive fish species in ponds. PLoS
ONE, 8, e56584.
Yamamoto, S., Masuda, R., Sato, Y., Sado, T., Araki, H., Kondoh, M., . . .
Miya, M. (2017). Environmental DNA metabarcoding reveals local fish
communities in a species-rich coastal sea. Scientific Reports, 7, 40368.
Yamamoto, S., Minami, K., Fukaya, K., Takahashi, K., Sawada, H., Mura-
kami, H., . . . Kondoh, M. (2016). Environmental DNA as a “Snapshot”
of fish distribution: A case study of japanese jack mackerel in maizuru
bay, sea of Japan. PLoS ONE, 11, e0149786.
Yamanaka, H., Minamoto, T., Matsuura, J., Sakurai, S., Tsuji, S., Moto-
zawa, H., . . . Kondo, A. (2017). A simple method for preserving
environmental DNA in water samples at ambient temperature by
addition of cationic surfactant. Limnology, 18, 233–241.
SUPPORTING INFORMATION
Additional Supporting Information may be found online in the
supporting information tab for this article.
How to cite this article: Ushio M, Fukuda H, Inoue T, et al.
Environmental DNA enables detection of terrestrial mammals
from forest pond water. Mol Ecol Resour. 2017;17:e63–e75.
https://doi.org/10.1111/1755-0998.12690