Figure 3: Sankey diagram summarising water preservation,
capture, filter preservation, extraction (lysis and inhibitor removal),
and detection methods used by studies included in our literature review.
Most assays (n = 100) targeted fish (49.0%), followed by
crustaceans (17.0%) (Figure S3a). The majority of assays (n =
123) were used for quantification (39.8%) or detection (35.8%) (Figure
S3b), and used in the Palearctic (45.3%) or Nearctic (40.7%) (Figure
S3d). Assays were usually deployed in natural environments (n =
84), specifically lentic (20.7%) or lotic (26.5%) freshwater systems
(Figure. S3c). Where assays were used in experimental systems (classed
as ‘other’ for environment), the volume of these artificial water
systems ranged from 900 ml to 336,000 L (median = 20 L). Sample sizes
(i.e., number of sampling sites) ranged from 1 to 197 (median = 4), with
between 1 and 120 biological replicates (median = 5) taken, and between
15 ml and 6 L of water (median = 500 ml) collected per biological
replicate.
The broad workflows used by assays included in our literature review are
summarised in Figure 3. Most assays used cooling (n = 25) after
‘other’ (n = 58, typically centrifugation or resin beads) for
water sample preservation, followed by filtration (n = 126;
Figures 3, S4) for eDNA capture. Ethanol/sodium acetate (n = 5)
for water sample preservation followed by precipitation (n = 31)
for eDNA capture was less popular but constituted a second major
analytical workflow (Figures 3, S4). Of those assays using filtration,
glass fibre filter membranes (0.7 μm pore size) were most commonly used,
followed by polycarbonate track-etched, cellulose nitrate, nylon and
‘other’ membrane types, including cellulose acetate and polyethersulfone
(Figure S5). Filters were typically frozen at -20℃ for preservation of
DNA in the retentate (Figure 3), but storage times were often not
reported. A full breakdown of precipitation and filtration methods can
be found in Supporting Information (Figures S6, S7).
The vast majority of assays (n = 125) used commercial extraction
kits (82.0%) as opposed to unbranded protocols (18.0%) (Figure S8),
with the Qiagen DNeasy Blood and Tissue Kit being the most commonly used
(47.76%; n = 110) (Figure S9). Mechanical disruption with
chemicals and chemicals only were secondary to an enzyme with chemicals
for cell lysis (Figure 3). Commercial kits typically employed an enzyme
with temperature to induce cell lysis and lacked an inhibitor removal
step, yet post-extraction inhibitor removal was uncommon (Figure S10).
Where post-extraction inhibitor removal was performed, this was either
done by phase separation or chemical flocculation (Figure 3) using
methods such as the Zymo One Step PCR Inhibitor Removal Kit, the Promega
Wizard Genomic DNA Purification Kit, chloroform or dilution.
Most assays (n = 136) targeted mitochondrial genes using
quantitative PCR (69.9%) (Figures 3, S11). Typically, three technical
replicates were performed in 20 μl reactions using 2 μl of template DNA.
The majority of assays (80.0%) did not use an internal positive control
to test for inhibition and did not determine the Limit of Detection
(54.4%), Limit of Quantification (71.0%) or effects of the
environmental matrix (78.0%) (Figure S12). Most assays used commercial
master mixes (Figure S13), such as Applied Biosystems TaqMan
Environmental Master Mix 2.0 and TaqMan Gene Expression Master Mix
(Figure S14), as opposed to custom master mixes. Where custom master
mixes were used, MgCl2 concentration ranged from 1.5 to
2.5 (median = 2) and dNTP concentration ranged from 0.05 to 0.25 (median
= 0.20). Promega 5x Colorless GoTaq Flexi Reaction Buffer and Promega
GoTaq Flexi DNA Polymerase were the most commonly used buffer type and
enzyme type respectively. Enhancers were not often added to PCR
reactions, but Bovine Serum Albumin (BSA) was most common where
enhancers were used (21.3%) (Figure S15).
Crucially, most assays (n = 145) did not measure, record or
report environmental parameters that are expected to affect the
distribution of DNA among states and determine stability of DNA (Figure
S16). Parameters that were recorded and reported included temperature
(50.3%), pH (22.0%), UV exposure (9.0%), season (68.0%), canopy
cover (3.0%), conductivity/salinity (22.0%), geology of catchment
(12.0%) and dissolved oxygen (15.0%). One third of papers would
require the authors to be contacted to clarify their analytical workflow
or ascertain if they collected environmental data but did not report it
(Figure S17).
Taken all together, our meta-analysis suggests that most single-species
studies employ methods that analyse a similar and potentially restricted
state of eDNA. The majority of studies use filtration at pore sizes
through which most dissolved eDNA may pass if clogging does not occur.
After filtration, the eDNA on the filter is isolated with similar lysis
methods and purification buffers provided with commercial extraction
kits that fundamentally employ similar chemistry (see Table S2). Most of
these commercial kits likely do not promote particle bound DNA to
desorb. To be certain, the constituents of the buffers would need to be
determined which was not feasible since most of these are trade secrets.
If these commercial kit buffers do not have competitive binders and do
not reach a pH high enough to promote desorption, it is likely that DNA
adsorbed onto particles was not isolated. If these assumptions are true
(i.e. dissolved DNA flows through filters and extraction kit buffers do
not promote desportion), then eDNA detections from the studies reviewed
here may originate from only inter-cellular or organellar DNA. However,
extensive research comparing whether specific molecular methods
co-purify eDNA states would be needed to verify this claim.