Effects of filtration methods and water volume on the quantification of brown trout ( Salmo trutta ) and Arctic char ( Salvelinus alpinus ) eDNA concentrations via droplet digital PCR

The quantification of the abundance of aquatic organisms via the use of environmen‐ tal DNA (eDNA) molecules present in water is potentially a useful tool for efficient and noninvasive population monitoring. However, questions remain about the reli‐ ability of molecular methods. Among the factors that can hamper the reliability of the eDNA quantification, we investigated the influence of five filtration methods (filter pore size, filter type) and filtered water volume (1 and 2 L) on the total eDNA and the fish eDNA concentrations of two species, brown trout ( Salmo trutta ) and Arctic char ( Salvelinus alpinus ) from tanks with known number of individuals and biomass. We applied a droplet digital PCR (ddPCR) approach to DNA extracted from water samples collected from two cultivation tanks (each of them containing one of the targeted species). Results showed that the quantification of fish eDNA concentra‐ tions of both species varies with filtration methods. More specifically, the 0.45‐µm Sterivex enclosed filters were identified to recover the highest eDNA concentra‐ tions. Difficulties to filter 2 L water samples were present for small pore size filters (≤0.45 µm) and likely caused by filter clogging. To overcome issues related to filter clogging, common in studies aiming to quantify fish eDNA molecules from water samples, we recommend a procedure involving filtration of multiple 1 L water sam‐ ples with 0.45‐µm enclosed filters, to recover both high quality and high concentra‐ tions of eDNA from targeted species, and subsequent processing of independent DNA extracts with the ddPCR method.

Among the numerous factors than can explain failures in studies aiming to detect and quantify fish species with the eDNA-based approach, many are methodological including sampling strategy, filtered volume, filtration method, DNA extraction, and quantitative molecular methods (Hansen et al., 2018;Wilcox et al., 2018).The volume of water to filter is, for instance, a major issue notably because eDNA detection rates can vary between systems related to ratio of fish biomass to lake size.For example, Wilcox et al. (2018) showed that studies failed to quantify fish populations due to the use of too small water volumes (15-75 ml).While most studies are based on the collection of 1 L (or less) water samples, recent works suggested that larger water volumes should be collected to ensure reliable detection rate of targeted species (>10-100 L; Valentini et al., 2016;Civade et al., 2016;Pont et al., 2018;Riaz, Wittwer, Nowak, & Cocchiararo, 2018).Another factor that may impact the eDNA detection probability is the filtration method, where both pore size and filter type may strongly influence the results (Eichmiller et al., 2016a;Fujii et al., 2019;Li, Lawson Handley, Read, & Hänfling, 2018;Miya et al., 2016;Spens et al., 2017;Takahara et al., 2012).If pore size is small (e.g., ≤0.45 µm), the filtration may allow to catch more free or particle-bound eDNA molecules from the targeted species.
However, it has been demonstrated that if pore size is too small, filters may be clogged, for example, in systems turbid, eutrophic, or with high concentration of dissolved organic carbon (DOC) (Li et al., 2018), causing less eDNA molecules from the targeted fish to be captured.
To study the potential methodological biases for fish detection and quantification, we applied a comparative approach of filtration methods and water volume coupled with the use of the ddPCR, a highly sensitive quantitative PCR method.The ddPCR has been suggested to be more accurate than qPCR to quantify the amount of eDNA molecules but is still rarely used in studies of fish eDNA (with exceptions of Doi et al., 2015b;Doi et al., 2015a;Hunter, Ferrante, Meigs-Friend, & Ulmer, 2019;Nathan, Simmons, Wegleitner, Jerde, & Mahon, 2014).To perform this analysis, we designed species-specific molecular tools (primers and probes) for brown trout (Salmo trutta) and Arctic char (Salvelinus alpinus).The studied species are highly valued as food and targeted both by recreational and smallscale local commercial fisheries but on the same time sensitive to overharvesting (Eriksson et al., 2006;Roux, Tallman, & Lewis, 2011).
As such, future fishery management should benefit from well-developed methods to detect and quantify trout and char population abundances from eDNA water samples.We collected water samples from cultivation tanks, one with a brown trout population, and one with an Arctic char population, with relatively high biomass of both species (0.88 m 3 , approx.45 kg of fish).One liter and two liter water samples were filtered with five filtration methods (0.2-1.2 µm pore size filters), the ddPCR outputs revealing the best strategy to use to recover both high total eDNA concentrations and fish eDNA concentrations from both species.

| Study site and collection of water samples
Water samples were collected from two tanks containing 1-year-old fish of brown trout (mean weight 13.5 g) and Arctic char (mean weight 42.6 g), on 24 April 2018 at a fish cultivation station in Lycksele (Sweden).Tanks were characterized by a water volume of 0.88 m 3 , a water temperature of 1.65°C, and a fish biomass of approximately 45 kg.Prior to sampling, the flow through river water circulation was closed for 90 min.Approximately 40 L of water was collected from each tank, then stored in sterilized 20-L plastic containers (two from each tank).We assumed that fish eDNA was equally distributed in the water of each tank.Sampling controls consisted of two 20-L containers fill up with MilliQ water placed at 50 cm from each tank during water collection (approx.20 min) to control for contamination of the two sample containers by DNA molecules that may have been released by fish in the cultivation station.In complement, two types of environmental controls were sampled: 2 L water samples collected from the inlet water of each tank and a sample of the food given to fish populations (biomar inicio 917) to check for potential contamination with fish DNA from the food source.

| Filtration of water samples
The 1 and 2 L water samples were filtered with a peristaltic pump  1).The filtered volume was measured for the filtration with [0.45MCE] and [0.45PVDF] filters when clogged (Table 1), and filtration was arbitrarily stopped after 1 hr.Note that only 1 L of water was filtered using the [0.22GP]-type filters because of the impossibility to filter 2 L water samples using this type of filter.In complement, the filtration of one 2 L water sample with [0.45MCE] filter failed from brown trout tank due to loss of water during filtration.
Inlet water samples were filtered only using the filtration method [1.2GF + 0.45MCE] (2 L filtered in duplicates).Sampling controls (i.e., SC) consisted of 1 or 2 L water samples filtered with the five different filtration methods.All filters were stored at −20°C with 1.8 ml of storage buffer (50 mM Tris-HCl, 40 mM EDTA, 0.75 M sucrose, pH = 8) until further analyses (DNA extractions within the following 2-3 weeks).All filtration equipment was sterilized by soaking for 1 day in 5% bleach and rinsing with 70% ethanol and MilliQ water before and between each filtration, respectively.

| Design of species-specific primers and probes for brown trout and Arctic char
Sampling and methods of sacrifices of fish used in this study comply with the current laws of Sweden and were approved by the local ethics committee of the Swedish National Board for Laboratory Animals in Umeå (CFN, license no.A20-14 to Pär Byström).
DNA was extracted from approximately 500 mg of tissue samples, preserved in absolute ethanol, from seven individuals brown trout (Salmo trutta) and seven individuals of Arctic char (Salvelinus alpinus) that originated from Swedish mountain lakes (Jämtland and Västerbotten counties) using the DNeasy Blood & Tissue Kit (Qiagen) and following the manufacturer protocol.The cytB and COI mitochondrial genes were selected to design species-specific primer and probes, this work resulting in the selection of the cytB genic region as the best candidate for both species.
1,330 bp of the COI genic region.Each PCR was performed in a total volume of 25 µl including 12.5 µl of 2*Qiagen Multiplex PCR Master Mix, 7 µl of ultrapure water, 4 µl of DNA template and 1.5 µl of a mix of both primers (300 nM).For both target, touchdown PCR protocol was applied.For PCR mixtures aiming to amplify COI genic region for trout and char tissue's DNA extracts from both species and cytB genic region for DNA extracts from trout tissue only, the PCR protocol includes an initial denaturation at 95°C for 15 min followed by 7 cycles of 30 s at 94°C, 90 s of annealing at 62°C (lowered by 0.5°C compared to each previous cycle), and 120 s at 72°C and 25 cycles of 30 s at 94°C, 90 s at 58°C, and 120 s at 72°C.The amplicons were then subjected to a final 5-min extension at 72°C.For PCR mixtures aiming to amplify cytB genic region for char tissue DNA extracts, the PCR protocol includes an initial denaturation at 95°C for 15 min followed by 10 cycles of 30 s at 94°C, 90 s of annealing 65°C (lowered by 0.5°C compared to each previous cycle), and 120 s at 72°C and 25 cycles of 30 s at 94°C, 90 s at 60°C, and 120 s at 72°C.The amplicons were then subjected to a final 5-min extension at 72°C.Sanger sequencing was applied to PCR amplicons using a 3730 DNA Analyzer (Applied Biosystems).Forward and reverse reads were then cleaned and merged using the software BioEdit (Hall, 1999) and MEGA7 version 7.0.26(Kumar, Stecher, & Tamura, 2016).
Consensus sequences are provided in Table S1.For both targeted regions, DNA sequences from Salmo and Salvelinus species were downloaded from GenBank (date: 07/07/2017) combining the genus name with the search terms "COI," or "cytB."Those sequences were aligned with obtained sequences.The software Primer3Plus (Untergasser et al., 2007) was used to design primers and probes fitting the following criteria: amplicon length around 50-150 bp, primers length around 15-30 bp, the total number of Gs and Cs in the last five nucleotides at the 3' end of the primer should not exceed two (GC-clamp), and GC content between 30% and 80%.The criteria that did not follow the recommendation were the ideal primer melting temperature (T m ) that should be around 58-60°C and the probe T m around 10°C higher than primer T m .For both targets, for primer sets and probes, the calculated T m values were around 54-56°C and 58-60°C, respectively.However, the calculation of primers/probes T m depending on the thermodynamic parameters used, and here, we used the salt correction formula SantaLucia 1998 and chose to try the primers in situ to verify their efficiency.
The species specificity of both primer sets and probes was verified using the software Primer-BLAST with default settings (Ye et al., 2012).Results show that only online DNA sequences from the same target species have 100% matches.Furthermore, both primer TA B L E 1 Volume of water filtered with the five filtration methods for the two volume of water for both fish species.A, B, and C correspond to the three filtrations done for each filtration method (biological replicates).SC samples correspond to sampling controls obtained from containers fill up with MilliQ water.The entire volume of 1 L was successfully filtered for the five filtration methods for replicates and SC.For the samples with a foreseen volume of water to filter at 2 L, the filtered volumes were written.Dash marks were displayed when filtrations (a) failed for the 2 L sample from brown trout tank with filtration method [0.45PVDF](b) were not performed possible for 2 L sample using [0.22GP] filters and probe sequences were designed to have a least 2 mismatches with nontarget species more particularly between the studied two species of the present study (Figure S1).In situ tests of specificities were also performed between primer and probe sets applying ddPCR method to DNA extracts from fish tissues from Arctic char and brown trout and showed no cross-amplifications.The specificity of primers to amplify the desired target was verified in DNA extracts from a part of water samples from cultivation tanks by applying a cloning-sequencing approach as following: PCR mixtures were performed in a total volume of 10 µl following the protocol described above.The PCR protocol includes an initial denaturation at 95°C for 15 min followed by 40 cycles of 30 s at 95°C and 1 min at 62°C.
The amplicons were then subjected to 5 min at 4°C and 5 min at 90°C.PCR amplicons were cloned using CloneJet PCR cloning kit (Thermo Scientific), followed by purification and Sanger sequencing (Eurofins).Sequencing results confirmed the specificity of each primer set.
The nucleotidic sequences of primers and probes designed in this study from the mitochondrial gene cytB are presented in Table 2.

| DNA extraction from filters
DNA extraction was performed from the filters using the DNeasy Blood & Tissue Kit (Qiagen).All filters were placed in 2-ml tube, and 720 µl ATL + 80 µl proteinase K was added.In the case of

| Droplet digital PCR assays
The ddPCR assays were performed independently for both species using the designed primers and probes (see cytB_Sa1 and tissues diluted at 1/1,000 and 1/100, respectively-were used to define a range of fluorescence to consider positive results and to check for repeatability between ddPCR assays.A lower threshold for a positive signal was arbitrarily defined to increase the stringency level: Any droplet beyond the fluorescence threshold was counted as a positive event (2,800 for brown trout and 850 for Arctic char).The ddPCR mixture with less than accepted 8,000 droplets was discarded from the analysis.To apply a stringent procedure, only ddPCR assays with more than 2 droplets were considered positive in the analyses, and only DNA extracts from which positive droplets were found in a least two of the three technical replicates were used to calculate the mean values of eDNA concentrations for each DNA extract (Table S2).
TA B L E 2 Nucleotidic sequences of primers and probes used for the ddPCR assay (cytB_St1 for brown trout eDNA quantification and cytB_Sa1 for Arctic char eDNA quantification).The TaqMan® probes were composed of FAM and VIC dyes (for brown trout and Arctic char detection, respectively), and the selected nucleotide sequence and MGB (minor groove binder)

| Data analysis
For each sample, we calculated fish eDNA concentrations (in copy number) for each DNA extract as follow: the mean eDNA concentration (described above) was divided by the volume of DNA extract used in the ddPCR mixtures (exactly 1.8 µl because only 20 µl out of the 22 µl of the starting mix is used for each assay), divided by the used dilution factors (none, 1:2, 1:10, or 1:100), and multiply by the total volume of the DNA extract (50 µl).The dilution factors were chosen carefully for each DNA extract in order to be able to discriminate, in the outputs of the ddPCR assays, the positive droplets from the background of negative droplets accordingly to the recommendations from the manufacturer (Bio-Rad).The outputs of this analysis are provided in Table S3.Part of sampling and environmental controls showed positive droplets for Arctic char.We considered that this was caused by an aerial contamination due to spreading of Arctic char DNA in the building where samples were collected.
The detection of Arctic char eDNA molecules in inlet water can be explained by the potential presence of Arctic char in water from upstream cultivation stations or natural occurrence in streams.
However, it did not hamper the reliability of the outputs of this work because this aerial contamination was very low compared to the number of copies found in water samples from cultivations tanks (number of Arctic char DNA copies always superior to 30 000 copies in tank water samples, while 1,500 copies were found at the maximum in one replicate from a control sample).Overall, DNA extraction and ddPCR controls showed no positive amplification (applying the stringent procedure described above) except for one DNA extraction control for each species but with also very limited number of copies per DNA extract that could not have impact the reliability of the outputs of the present work: 65 and 28 copies for the DNA extraction controls tested for ddPCR amplifications of brown trout and Arctic char DNA, respectively, Table S3).Pearson's product-moment correlations were calculated between total eDNA concentrations and fish eDNA concentrations obtained for each species using the cor.test function (method = 'pearson').Relationships between filtration methods, total eDNA concentrations (in ng/µl), and fish eDNA concentrations (in number of copies per DNA extract) were analyzed using a one-way ANOVA with interactions with the functions aov and summary from R software (version 3.6.0).

| RE SULTS
The filtration of all 1 L water samples was successfully performed with the five different filtration methods for samples from brown trout and Arctic char tanks and sampling controls (Table 1).
However, the filtration of 2 L water samples was not possible to complete with [0.45MCE] and [0.45PVDF] filters.No filtration with [0.22GP] filters was performed from 2 L water samples based on previous knowledge of clogging of these filters.The estimated total eDNA concentrations across all samples (filters and volume) ranged from 5.2 to 225.6 ng/µl in water samples from the brown trout's tank and from 2.9 to 129.7 ng/µl from Arctic char's tank (Figure 1).
As shown by the results of the one-way ANOVA, the choice of filtration methods had an effect on the total eDNA concentrations of the obtained DNA extracts (Table 3).[1.2GF] filters extracted very low total eDNA concentrations for both species from either 1 or 2 L water samples (mean values 6.4 ng/µl ± 1.1 and 4.1 ng/ µl ± 1.5 for 1 L samples from brown trout and Arctic char tanks, respectively), while relatively higher total eDNA concentrations were measured from [0.45MCE] filters (Table 3).The filters that allowed to extract the highest quantity of total eDNA were the The number of obtained fish eDNA copies obtained from 1 L water samples differed between filtration methods (Figure 1, Table 3).While the number of fish eDNA copies from brown trout was found significantly correlated to total eDNA concentrations (Pearson's product-moment correlation: 0.70, p < .001),such relationship was not detect for Arctic char DNA extracts (Pearson's product-moment correlation: 0.08, p < .78).For Arctic char, the lack of relationship was due to the high total eDNA concentration estimates obtained for the [1.2GF + 0.45MCE] filters compared to estimated total eDNA concentrations (Figure 2).A similar pattern was also present for brown trout.It was also observed that the filtration of higher volume of water (up to 2 L) led to higher total and fish eDNA concentrations (Figure 1).However, the filtration of 2 L was not possible in the case of filters [0.45MCE] and [0.45PVDF] (as well as [0.22GP] for which no filtration was performed) for both species, but higher water volumes were filtered in the case of Arctic char's tank (compared to brown trout's tank).Interestingly, while the DNA extracts obtained from [1.2GF + 0.45MCE] filtration methods were found to have high total eDNA concentrations, the number of fish eDNA copies amplified from them was lower compared to the [0.45PVDF] and [0.22GP] filters (Figures 1 and 2).
Overall, the combination that recovered the highest rate of fish eDNA copies was the filtration of 2 L water samples using either [0.45PVDF] and [0.45MCE] filters.However, the filtration of 1 L samples with these two filtration methods showed less variation between biological replicates (in the number of copies per DNA extract, Figure 1).

| D ISCUSS I ON
Our present study highlights the effects of the use of different filtration methods to reach desired water volume and to retrieve high concentrations of both total eDNA and targeted fish eDNA molecules.
As previously shown by many studies (Li et al., 2018;Miya et al., 2016;Spens et al., 2017;Takahara et al., 2012), the choice of filtration method has a strong impact on the estimation of targeted eDNA concentrations.The [1.2GF] filters recovered very low total eDNA as well as fish eDNA concentrations, suggesting that the large pore size of these filters may explain the poor efficiency to retain particles binding fish eDNA molecules compared with other filter sizes used (0.2 and 0.45 µm pore sires; Figure 1), such results being in line with some previous findings (e.g., Eichmiller et al., 2016a;Miya et al., 2016;Turner et al., 2014).One the other hand, studies have shown that large pore size filters-and even larger-are sufficient to catch eDNA molecules to detect or quantify fish populations and sometimes even better than smaller pore size filters (Lacoursière-Roussel, Rosabal, & Bernatchez, 2016a;Takahara et al., 2012).Li et al. (2018) recommended the use of 0.8-µm filters as optimal filters for turbid, eutrophic, and high fish density ponds since they reached a good balance between filtration efficiency and probabilities of species detection.
Filter type (not size) may also influence the efficiency to catch total DNA and targeted fish eDNA.Indeed, we found that [0.45PVDF] filters TA B L E 3 Results of the one-way ANOVA performed from outputs obtained from 1 L samples for each species.Models include as factors the filtration methods, and as response variables the total eDNA concentrations and fish eDNA concentrations.df values correspond to the degree of freedom.F corresponds to the ratio of the two variables divided by their respective degrees of freedom  confirming that filter cartridges (both 0.45 and 0.22 µm) may be at least equally or more sufficient than others to recover fish eDNA molecules (Miya et al., 2016;Spens et al., 2017).In contrast, Djurhuus et al. (2017) revealed that the use of different filter membranes had no impact on richness and community composition assessment (i.e., 0.2 µm GFF, NC, shedding material from the fishes.These findings are also lined with studies highlighting that filters with pore size smaller than 0.45 µm may be clogged easily by suspended solid (Fujii et al., 2019).
The use of qPCR may be useful to quantify the number of gene copies of the specific species from water samples and already revealed its potential to estimate fish abundance in comparison with classical technics (Eichmiller et al., 2014;Eichmiller et al., 2016a;Eichmiller et al., 2016b;Lacoursière-Roussel, Rosabal, et al., 2016a;Lacoursière-Roussel, Côté, Leclerc, Bernatchez, & Cadotte, 2016b;Takahara et al., 2012;Takahara, Minamoto, & Doi, 2013;Wilcox et al., 2013); however, both types of approach may be prone to inefficiency related to PCR inhibitors.In our study, the presence of PCR inhibitors, suspected in [1.2GF + 0.45MCE] DNA extracts, appeared to have no effects on the success of PCR highlighting the strength of the ddPCR method to reduce potential PCR inhibitions by partitioning of humic substances in droplets (alongside with the partitioning of DNA molecules) and thus reducing PCR inhibition (Doi et al., 2015b;Doi et al., 2015a).In complement, the ddPCR method provides an absolute quantification of the number of copies of a specific genic region and, thus, is a promising tool for fish monitoring using eDNA-based approaches.
To conclude, in line with Miya et al. (2016) and Hunter et al. (2019) recommendations, we recommend the filtration of multiple 1 L samples using [0.45PVDF] filters followed by the application of the ddPCR method that reduce the potential effects of PCR inhibitors, an important aspect of the eDNA-based approach more particularly when studying fish populations from turbid or high DOC aquatic systems.(2018).Monitoring riverine fish communities through eDNA metabarcoding: determining optimal sampling strategies along an altitudinal and biodiversity gradient.Metabarcoding and Metagenomics, 2, e30457.https ://doi.org/10.3897/mbmg.2.30457

[ 1 .
2GF + 0.45MCE] filtration method, both filters were put together in the same tube during the lysis step.Then, the next steps were performed following manufacturer protocols resulting to elution of DNA into a 50 µl volume.DNA was extracted from 500 mg of the food sample using the same procedure.DNA extraction controls were performed alongside with environmental samples to evaluate potential cross-contamination.The eDNA concentration (ng/ µl) of each DNA extract was estimated using a Nanodrop ND-1000 Spectrophotometer (Thermo Scientific) with triplicates measurements for each sample.
cytB_St1 in Table2).Each ddPCR mixture contained 2 μl of DNA template (diluted at 1, 2, 10, or 100 depending on DNA extracts), 400 and 200 nM of primers and TaqMan MGB probe for brown trout and Arctic char, respectively, 10 µl of 1× Bio-Rad Supermix for Probes (Bio-Rad) with ultrapure sterilized water up to a total volume of 22 µl.From this 22 µl reaction mix, 20 µl (note that it includes then 1.8 µl of DNA template) was mixed with Bio-Rad droplet generator oil and partitioned into up to 20,000 droplets using the Bio-Rad QX-200 droplet generator (Bio-Rad).PCR mixtures were performed in sealed 96-well plates (Bio-Rad) with the following conditions: 5 min at 95°C, 40 cycles of denaturation for 30 s at 95°C, and extension for 60 s at 62°C, followed by 5 min at 4°C, 5 min at 95°C, and a hold at 4°C.After PCR amplification, plates were transferred to a Bio-Rad QX-200 droplet reader (Bio-Rad).PCR optimizations were previously performed to select suitable primers concentration and extension temperature for the amplification of both target.The ddPCR assays were run in triplicates (technical replicates) for a total number of 27 and 26 DNA extracts for the water samples from brown trout and Arctic char tanks respectively (Table1)as well as for DNA extracts from sampling controls (n = 16, 8 for each species), environmental controls (n = 3, 2 inlet water samples + 1 food), DNA extraction controls (n = 4), and for ddPCR controls (i.e., ultrapure water instead of DNA template).The Bio-Rad's QuantaSoft software version 1.7.4.0917 was used to quantify the number of copies of target DNA by μl of DNA extract.For each assay, the ddPCR mixture is partitioned into up to 20,000 droplets in which individual PCR occurred.When the PCR of a droplet is successful, the Bio-Rad QX200 droplet reader (Bio-Rad) give an estimation of the fluorescence measured in each droplet.Thus, the fluorescence amplitude is displayed by the Bio-Rad's QuantaSoft software to differentiate between the droplets that yielded positive and negative results.Positive controls-DNA extracts from brown trout (98 ng/µl) and Arctic char (71.4 ng/µl)

F
I G U R E 1 Total eDNA concentrations (in ng/µl) and fish eDNA concentrations (in number of copies per DNA extract) for each treatment (filtration methods and foreseen water volume) for both species | 157 CAPO et Al.[0.45PVDF] and [0.22GP] filters as well as the [1.2GF + 0.45MCE] filter's combination with values higher than 50 and 35 ng/µl for DNA extracts from brown trout and Arctic char's tanks, respectively (Figure 1).

F
I G U R E 2 Plot showing the relationships between the values of total eDNA concentrations and fish eDNA concentrations for each DNA extract.Filtration methods and filtered volume are displayed in each figure by color and shape-based categories, respectively.Each set of samples with the same color and shape corresponds to each biological replicate (filter) analyzed in this study.Regression lines showed the relationships between the two sets of values.Pearson's product-moment correlation values are presented in the text allow to recover more total DNA and fish eDNA than [0.45MCE] filters 2-and 0.45-µm filters (i.e., [1.2GF + 0.45MCE]) and the co-extraction of DNA from both filters were less efficient to retrieve fish eDNA molecules compared to the [0.45PVDF]-and [0.22GP]-based approaches.We suggest that the co-extraction of the eDNA molecules from both filters, notably from [1.2GF] filters, may have negatively influenced the extraction of DNA molecules from [0.45MCE] filters.While 1 and 2 L water samples werefully filtered with this serial filtration method, less fish eDNA concentrations were measured especially when compared to the measured total eDNA concentrations, while the relationships were more consistent for the other filtration methods (Figure2).We therefore suspect an overestimation of the total eDNA concentration by the Nanodrop analysis (i.e., absorbance measurements) for the filtration method [1.2GF + 0.45MCE].Therefore, this high estimate of total eDNA concentration implies the presence of co-extracted compounds such as humic substances (e.g., phenols) that may act as PCR inhibitors(McKee, Spear, & Pierson, 2015).Interestingly, those co-extracted compounds were not detected when extracting DNA from [1.2GF] filters only, suggesting that the co-extraction of DNA from [GF] and [MCE] may have increased the probability of co-extraction of PCR inhibitors.
This work was funded by the foundation Carl Tryggers Stiftelse för Vetenskaplig Forskning (https ://www.carltrygge rssti ftelse.se/,grant numbers CTS16:84 and CTS18:812).We are very grateful to the help from Johan Andersson at Rural Economy and Agricultural Society´s fish cultivation station at Lycksele that allowed us to perform this study.We would also like to thank Mikael Lindberg (Protein expertise platform, Umea University) who provided services for cloning-sequencing of PCR products.Thanks also to Sonia Brugel and Guillaume Grobois for help to design filtration setup, to Collin Duinmeijer and Marine Vandewalle-Capo for help during water filtration, and to Fredrik Olajos for fruitful discussion on ddPCR analysis., GS, and HK designed the study.EC performed the laboratory work and data analysis and interpretation.EC, PB, GS, and HK contributed to the writing of the manuscript.All authors contributed to the revision of the manuscript, read, and approved the submitted version.orcid.org/0000-0001-9143-7061R E FE R E N C E S Bylemans, J., Gleeson, D. M., Lintermans, M., Hardy, C. M., Beitzel, M., Gilligan, D. M., & Furlan, E. M.