Comparison of capture and storage methods for aqueous
macrobial eDNA using an optimized extraction protocol:
advantage of enclosed filter
Johan Spens
1,2
†, Alice R. Evans
1
†, David Halfmaerten
3
, Steen W. Knudsen
1
, Mita E. Sengupta
4
,
SarahS.T.Mak
1
, Eva E. Sigsgaard
1
and Micaela Hellstr
€
om
1,5
*
1
Centre for GeoGenetics, Natural History Museum of Denmark, Øster Voldgade 5-7, 1350 Copenhagen K, Denmark;
2
Wildlife,
Fish and Environmental Studies, Swedish University of Agricultural Sciences, Skogsmarksgra
¨
nd, 90183 Ume
a, Sweden;
3
Research Institute for Nature and Forest, Gaverstraat 4, 9500 Geraardsbergen, Belgium;
4
Department of Veterinary Disease
Biology Parasitology and Aquatic Diseases, Dyrlægevej 100, 1870 Frederiksberg C, Copenhagen, Denmark; and
5
Department
of Ecology, Environment and Plant Sciences, Stockholm University, 10691 Stockholm, Sweden
Summary
1. Aqueous environmental DNA (eDNA) is an emerging efficient non-invasive tool for species inventory studies.
To maximize performance of downstream quantitative PCR (qPCR) and next-generation sequencing (NGS)
applications, quality and quantity of the starting material is crucial, calling for optimized capture, storage and
extraction techniques of eDNA. Previous comparative studies for eDNA capture/storage have tested precipita-
tion and ‘open’ filters. However, practical ‘enclosed’ filters which reduce unnecessary handling have not been
included. Here, we fill this gap by comparing a filter capsule (Sterivex-GP polyethersulfone, pore size 022 lm,
hereafter called SX) with commonly used methods.
2. Our experimental set-up, covering altogether 41 treatments combining capture by precipitation or filtration
with different preservation techniques and storage times, sampled one single lake (and a fish-free control pond).
We selected documented capture methods that have successfully targeted a wide range of fauna. The eDNA was
extracted using an optimized protocol modified from the DNeasy
Ò
Blood & Tissue kit (Qiagen). We measured
total eDNA concentrations and Cq-values (cycles used for DNA quan tification by qPCR) to target specific
mtDNA cytochrome b (cyt b) sequences in two local keystone fish species.
3. SX yielded higher amounts of total eDNA along with lower Cq-values than polycarbonate track-etc hed filters
(PCTE), glass fibre filters (GF) or ethanol precipitation (EP). SX also generated lower Cq-values than cellulose
nitrate filters (CN) for one of the target species. DNA integrity of SX samples did not decrease significantly after
2 weeks of storage in contrast to GF and PCTE. Adding preservative before storage improved SX results.
4. In conclusion, we recommend SX filters (originally designed for filtering micro-organisms) as an efficient cap-
ture method for sampling macrobial eDNA. Ethanol or Longmire’s buffer preservation of SX immediately after
filtration is recommended. Preserved SX capsules may be stored at room temperature for at least 2 weeks with-
out significant degradation. Reduced handling and less exposure to outside stress compared with other filters
may contribute to better eDNA results. SX capsules are easily transported and enable eDNA sampling in remote
and harsh field conditions as samples can be filtered/preserved on site.
Key-words: capsule, eDNA capture, environmental DNA, extraction, filter, monitoring, quantita-
tive PCR, species-specific detection, water sampling method
Introduction
The realization that DNA from macrobiota can be obtained
from environmental samples (environmental DNA, eDNA)
started with excrements (H
€
oss et al. 1992) and sediments
(Willerslev et al. 2003). Over the last decade, the potential of
aqueous eDNA to identify a wide range of plants and animals
from a small volume of water has been realized (Martellini,
Payment & Villemur 2005; Thomsen et al. 2012; Rees et al.
2014). Aqueous eDNA is an emerging increasingly sensitive
technique for revealing species distributions (e.g. Jane et al.
2015; Valentini et al. 2016), early detection of invasive species
(e.g. Smart et al. 2015; Simmons et al. 2016) and monitoring
rare and/or threatened species for conservation (e.g. Zhan
et al. 2013; McKee et al. 2015). Aqueous eDNA monitoring
provides possibilities to upscale species distribution surveys
considerably, because much less effort in time and resources
are required compared to conventional methods (Dejean et al.
2012; Davy, Kidd & Wilson 2015). Based on literature
*Correspondence author. E-mail: micaela.hellstrom@su.se
†Joint first authors.
© 2016 The Authors. Methods in Ecology and Evolution © 2016 British Ecological Society
Methods in Ecology and Evolution 2016 doi: 10.1111/2041-210X.12683
searches, we catalogue 49 studies successfully applying eDNA
from water samples to detect macro-organisms in aquatic
ecosystems, published between January 2005 and March 2015
(when this study was initiated; Table S1, Supporting Informa-
tion). To our knowledge, 39 additional empirical studies were
published since then, indicating a rapid rise of interest in this
research area (Table S2).
The field of eDNA is still evolving, and a consensus of cap-
ture, storage and extraction methods has not yet been reached
(Goldberg, Strickler & Pilliod 2015; Tables S1 and S2). In fact,
thediversityofmethodsisalmostashighasthenumberof
research groups investigating this fairly new field of research.
To ensure reliable results of downstream applications such as
quantitative PCR (qPCR) and next-generation sequencing
(NGS), the quantity and quality of the starting material is cru-
cial. From our eDNA laboratory experience, we find that a
modified easy-to-follow extraction protocol resulting in high
yields is needed. Based on eDNA studies published so far
(Tables S1 and S2), we identify three pre-PCR key issues that
hold opportunities for improvement: (i) capturing sufficient
quantities of eDNA as quite a few studies report low amounts
of captured total eDNA, (ii) effectively preserving eDNA sam-
ples before extraction and (iii) lowering contamination risks
from collection to extraction of eDNA.
Comparative studies on aqueous eDNA capture and storage
techniques (i.e. optimal ways of preserving the eDNA captured
on the filters until extraction; e.g. Renshaw et al. 2015) were
basedontheso-called‘openfilters’ (requiring handling, a filter
funnel and a vacuum pump; e.g. Liang & Keeley 2013; Turner
et al. 2014b) and ethanol precipitation (EP; e.g. Piaggio et al.
2014; Deiner et al. 2015). However, no enclosed filters were
included in previous comparative assays.
The Sterivex-GP capsule filter (SX), with a polyethersulfone
membrane, is a standard method for characterizing microbial
communities (Chestnut et al. 2014) and for removing patho-
gens from water as the organisms are captured on the filter
membranes. To our knowledge, only two published aqueous
eDNA studies have used this filter to detect aquatic macro-
organisms (fish detection: Keskin 2014; Bergman et al. 2016),
and the technique has been successful to detect a wide range of
aquatic macro-organisms in Denmark and Belgium (M.
Hellstr
€
om, M.E. Sengupta, S.W. Knudsen, D. Halfmarten.
unpublished, S1). The SX filter is enclosed in a capsule, which
reduces handling. A water sample can easily be filtered in the
field, saving time and facilitating fixation of the eDNA imme-
diately after capture. Additionally, downstream DNA extrac-
tion takes place within the filter capsules with no need for the
membrane to be removed or handled. We therefore test the
performance of SX compared to other more frequently used
eDNA capture methods (Table S1), under different storage
conditions, in an effort to address issues 1–3 above. To date,
there are no studies comparing SX to other capture methods
and multiple storage treatments. We aim to fill this gap, with
an experimental study comparing SX with four other capture
methods in a set-up with five typical storage treatments and
three different storage times (up to 2 weeks). The tested open
filter materials polycarbonate, cellulose nitrate and glass fibre
(GF) and the range of tested pore sizes (02–06 lm) are typical
of previous studies (Tables S1 and S2). We used an optimized
extraction protocol based on a commercial kit to increase
eDNA yields. To evaluate the usefulness of the SX and preser-
vation buffers in comparison with typically used methods
(Tables S1 and S2), we test the following H
0
hypotheses:
H
0
1. CAPTURE METHOD: SX is equally effective as
other tested eDNA capturing techniques in regard to DNA
quantity and quality measured as the total extracted eDNA
concentration [eDNA
tot
] and as Cq-values (quantification
cycles, sensu Bustin et al. 2009) from two species-specific
qPCR assays.
H
0
2a. STORAGE PRESERVATIVE: Storing filters with
a preservation buffer does not affect qPCR amplification
compared to immediate extraction or freezing at 20 °C
(no buffer added).
H
0
2b. STORAGE TIME: There is no significant difference
in eDNA quality over time between SX and the other tested
capturing techniques.
H
0
3. CONTAMINATION: There is no significant differ-
ence between SX and the other tested capture techniques in
occurence of false positives.
To test these hypotheses, we use an experimental set-up with
subsampling a single large homogenous sample of water from
a Danish lake. Subsamples are subjected to different eDNA
capture methods within the same day followed by different
storage treatments. A control site (fish-free pond) is sampled
using the same set-up. Each capture and storage treatment is
assessed using concentration of total eDNA as well as species-
specific qPCR assays targeting pike Esox lucius L. and perch
Perca fluviatilis L. By testing H
0
hypotheses (1–3), the multiple
opportunities for optimization of eDNA surveys held by the
use of SX may be empirically evaluated. Based on the results,
we suggest recommendations for improved capture, storage
and extraction to use for aqueous eDNA, taking remote and
harsh field conditions into consideration.
Materials and methods
STUDY SITES
We chose Gent ofte Lake, Denmark (N557435°,E125348°), as the
study site and a fish-free pond in Copenhagen botanical garden as a
negative field control (N556875°,E125746°). Gentofte Lake (26 ha) is
an alkaline clear water (Appendix S2) harbouring a wide range of fish
species, including pike and perch.
WATER COLLECTION
We retrieved 130 L of water from Gentofte Lake on 17 March 2015.
The water (4 °C) was collected at c. 30 points along c. 100 m of shore-
line close to the outlet of the lake. Additionally, we collected 40 L of
water from the control pond on 21 March 2015. The water was
© 2016 The Authors. Methods in Ecology and Evolution © 2016 British Ecological Society, Methods in Ecology and Evolution
2 J. Spens et al.
collected in sterilized 5-L buckets which prior to sampling were soaked
in bleach (5%) for 10 min, and then rinsed with laboratory-grade etha-
nol (70%). The containers were soaked repeatedly in lake water at a
location away from the collection point. Nitrile gloves were used during
cleaning, collection and filtration.
CAPTURE AND STORAGE
We carried out 41 different treatment combinations of the water sample
intotal(Table 1,Fig.S1).Weusedfivecapturetechniques,fivestorage
methods and three time regimes. All treatments were performed in trip-
licate. Apart from an in-house modified SX procedure (see Fig. 1), the
capture and storage methods were based on published sources
(Table S1). The capture methods (hereafter referred to with their abbre-
viations in square brackets) were as follows: (i) ethanol precipitation
[EP] (Ficetola et al. 2008), (ii) mixed cellulose esters membrane filters
including cellulose nitrate and cellulose acetate [CN]; Advantec 47 mm
diameter 045 lm pore size (Toyo Roshi Kaisha, Ltd., Tokyo, Japan),
(iii) polycarbonate track-etched filters [PCTE]; Whatman Nucleopore
Membrane 47 mm diameter 02 lmporesize(MerckKGaA,Darm-
stadt, Germany)], (iv) glass fibre [GF] membrane filters; Advantec
GA-55 47 mm diameter 06 lm pore size (Toyo Roshi Kaisha, Ltd.,
Tokyo, Japan) and (v) sterivex-GP capsule filters [SX]; polyethersul-
fone 022 lm pore size with luer-lock outlet (Merck KGaA)]. Further
downstream, SX was divided into an extraction from the filter within
the capsule (SX
CAPSULE
), after removal of the storage buffer, and an
extraction from the removed preservation buffer within a centrifuge
tube (SX
TUBE
; see DNA extraction section below). The different stor-
age methods were as follows: (i) ethanol 99% 200 proof at room tem-
perature (RT), Molecular Biology Grade (Thermo Fisher Scientific
Inc.,Waltham,MA,USA);(ii)Longmire’s buffer at RT (Longmire’s;
Longmire, Maltbie & Baker 1997); (iii) RNAlater at RT (RNA Stabi-
lization Reagent; QIAGEN, Stockach, Germany); (iv) no buffer, fro-
zen at 20 °C; and (v) no buffer, refrigerated at 8–10 °C. The three
time regimes between filtration and extractions were (i) within 5 hours
(5 h), (ii) within 24 h and (iii) after 2 weeks. Each treatment (n = 41)
was performed in triplicate. For each filter replicate, 1 L of lake water
was processed (0015 L for EP). For each capture–storage treatment,
we included one negative control without lake water. Additionally, 1 L
tapwaterwasrunthrougheachfilter(0015 L for EP) as a control to
detect potential contamination from the filtration facilities. For the
control pond, one sample per capture–storage treatment was processed
(n = 23). We captured eDNA from 155 subsamples and negative con-
trols altogether. The water samples were filtered or ethanol-precipitated
by a team of 10 researchers and the replicates of each treatment started
at different times to avoid temporal bias of filtrations. Prior to DNA
capture, bench surfaces and all equipment were wiped with bleach
(5%) and laboratory-grade ethanol (70%). Prior to each collection of
subsamples, the water was mixed thoroughly in the 130-L container.
For the open membrane filter (GF, CN and PCTE), 1 L water samples
were vacuum-filtered (c. 15–30 min) using Nalgene 250-mL sterile dis-
posable test filter funnels (Thermo Fisher Scientific Inc. USA). The fil-
ters were removed from the funnel with forceps and then placed in 5-
mL DNA LoBind
Ò
centrifuge tubes (Eppendorf AG, Hamburg, Ger-
many) that were either empty (if the time regime was 5 h or the storage
method was freezing) or contained preservation buffer. For all treat-
ments and downstream applications, Eppendorf DNA LoBind
Ò
tubes
were used in order to avoid up to 50% retention of DNA by the plastic,
which is a documented problem especially for short DNA fragments
(Gaillard & Strauss 1998; Ellison et al. 2006). For the SX filters, 1 L of
water was slowly (c. 10 min to avoid tearing of filters, following manu-
facturer’s recommendations) pushed through each filter capsule using a
prepacked sterile 50-mL luer-lock syringe. Remaining water in the SX
was removed by pushing air through the filter until dry, also using the
syringe. The outlet ends of the filters were closed with MoBio outlet
caps (MOBIO Laboratories, QIAGEN) and 2 mL preservation buffer
was pipetted to the inlet end using filter tips. The inlet ends were closed
with inlet caps (MOBIO Laboratories, QIAGEN) and both ends were
sealed with parafilm whereafter the capsules were inverted vigorously.
The frozen samples and the (5 h) and (24 h) EP samples were placed at
20 °C until extraction, while the non-treated samples (5 h) were
placed in a refrigerator and extracted directly after the filtering session.
Samples containing buffers were stored at RT until processed. The
(2 weeks) EP samples were frozen for 24 h prior to extraction to allow
for precipitation. In total, we processed 96135 L of water from the lake
(32 treatments 9 3replicates9 1L+ 3EPtreatments9 3 repli-
cates 9 0015 L) and 20045 L of water from the control pond (20
treatments 9 1 replicate 9 1L+ 3EPtreatments9 1 repli-
cates 9 0015 L; Table 1).
MOLECULAR LABORATORY CONDITIONS
DNA extractions and qPCR assays took place in the laboratories at
the Centre for GeoGenetics, University of Copenhagen, Denmark.
The facilities are designed for handling environmental samples requir-
ing the most stringent precautions to avoid contamination. Pre-PCR,
extraction and PCR facilities are located in separate designated rooms
with positive air pressure. Laboratory coats are changed between
rooms. Prior to any work in the laboratory, all surfaces are washed
with 5% bleach and 70% ethanol. After completing extractions
Table 1. Outline of the number of samples processed per capture and storage treatment (negative control pond in parentheses)
Capture Sum
Storage
Refrigerated
Frozen Ethanol Longmire’s RNAlater Frozen Ethanol Longmire’s RNAlater
5h 24h 2weeks
SX
CAPSULE
27 (5) 3 (1) 3 (1) 3 (1) 3 (1) 3 (1) 3 3 3 3
SX
TUBE
18 (3) 3 (1) 3 (1) 3 (1) 3 3 3
Cellulose nitrate 15 (5) 3 (1) (1) (1) (1) (1) 3 3 3 3
Glass fibre 27 (5) 3 (1) 3 (1) 3 (1) 3 (1) 3 (1) 3 3 3 3
Polycarbonate 27 (5) 3 (1) 3 (1) 3 (1) 3 (1) 3 (1) 3 3 3 3
Precipitation 9 (3) 3 3 (3) 3
Total 123 (26)
Sterivex, eDNA extraction within capsule (SX
CAPSULE
); Sterivex, eDNA extraction from buffer in tube outside capsule (SX
TUBE
).
© 2016 The Authors. Methods in Ecology and Evolution © 2016 British Ecological Society, Methods in Ecology and Evolution
Testing enclosed filter for eDNA capture and storage 3
Fig. 1.
© 2016 The Authors. Methods in Ecology and Evolution © 2016 British Ecological Society, Methods in Ecology and Evolution
4 J. Spens et al.
involving guanidiumthiocyanate, surfaces are washed with 70% etha-
nol (to avoid reactions between chlorine in the bleach and guanidi-
umthiocyanate in two of the buffers provided with the Qiagen kit), 5%
bleach and then 70% ethanol. All extractions of eDNA took place in
laminar flow hoods which were UV-treated before and after extrac-
tions. Every night, the entire facilities are automatically UV-
treated for a 2-h period.
DNA EXTRACTION
We extracted the eDNA using the extraction protocol outlined in
Fig. 1 and Appendix S1. The SX filters containing preservation buffers
underwent two extractions, one extraction from the buffer and one
extraction within the filter capsule after it had been emptied of buffer
(hereafter referred to as SX
TUBE
and SX
CAPSULE
). Altogether, 179 (24
SX
TUBE
+ 155 (see ‘Capture and storage’ section above) samples from
the study lake and the control pond were extracted. We measured
[eDNA
tot
] in each extraction using a Qubit 1.0 fluorometer (Thermo
Fisher Scientific Inc.) applying the high-sensitivity assay for dsDNA
(Life Technologies, Carlsbad, CA, USA).
QUANTITATIVE PCR
For the qPCR assays (e.g. Wilcox et al. 2013), two species-specific Taq-
Man primers/probe sets were used targeting 84 and 89 base pair frag-
ments o f the mitochondrial cytochrome b (cyt b)geneinpikeand
perch, respectively (Table S3). Species specificity of the assays was
tested on extracted DNA from non-target species (Table S3) using the
qPCR set-up described below. These non-target species did not gener-
ate any amplification signals. The optimal ratio of probe: primer con-
centration was tested prior to the study. The final PCR set-up to detect
the target species was as follows: pike – 5 lLtemplateDNA,125 lL
TaqMan Environmental Master Mix 20 (Life Technologies), 3 lLfor-
ward primer (10 l
M), 2 lL reverse primer (10 lM)and3lL probe
(25 lM); and perch – 5 lLtemplateDNA,125 lL TaqMan Environ-
mental Master Mix 20 (Life Technolo gies), 05 lLforwardprimer
(10 l
M), 25 lL reverse primer (10 lM), 3 lL probe (25 lM)and
15 lL UV-treated laboratory-grade water. The TaqMan qPCRs were
performed on a Stratagene Mx3005P (Thermo Fisher Scientific Inc.)
using thermal cycling parameters of 50 °C(5 min),95 °C(10 min)fol-
lowedby50cyclesof95°C(30s)and60°C (1 min). For each plate,
no-template controls (NTCs) and positive/negative tissue extracts were
run alongside the samples. All filtering and extraction negatives were
included in the qPCR assays. Additional qPCR replicates were run in
order to detect effects of freezing and thawing of the samples. To check
for PCR inhibition in the lake, separate qPCR assays for both species
following the protocols above were performed in a dilution series
(1:1,1:2,1:10and1:20)ofextractedDNAonfoursamplesrepli-
cated twice plus two positive and two negative controls to determine
any deviation of the amplification curves. The dilution series did not
indicate inhibition.
DATA ANALYSIS
To compare detection probability (i.e. diagnostic sensitivity) between
eDNA capture methods, the proportion of positive qPCR replicates
was calculated for each target species. Positive samples were analysed
Fig. 1. Flow chart illustrating the modified environmental DNA (eDNA) extraction protocol based on DNeasy Blood & Tissue Kit (QIAGE N,
Carlsbad, CA, USA). *) Capture: SX, Sterivex-GP polyethersulfone capsule filters, Note that SX
CAPSULE
and SX
TUBE
are treated as separate sam-
ples from step 2. CN, cellulose nitrate; PCTE, polycarbonate track-etched; GF, glass fibre filters; EP, ethanol precipitation. Storage: Frozen at
20 °C, Refrigerated are samples stored at 8–10 °C and processed within 5 h. Steps 9–26 see Appendix S1.
© 2016 The Authors. Methods in Ecology and Evolution © 2016 British Ecological Society, Methods in Ecology and Evolution
Testing enclosed filter for eDNA capture and storage 5