Individual genotypes from environmental DNA: Fingerprinting snow tracks of three large carnivore species

Continued advancements in environmental DNA (eDNA) research have made it possible to access intraspecific variation from eDNA samples, opening new opportunities to expand non‐invasive genetic studies of wildlife populations. However, the use of eDNA samples for individual genotyping, as typically performed in non‐invasive genetics, still remains elusive. We present successful individual genotyping of eDNA obtained from snow tracks of three large carnivores: brown bear (Ursus arctos), European lynx (Lynx lynx) and wolf (Canis lupus). DNA was extracted using a protocol for isolating water eDNA and genotyped using amplicon sequencing of short tandem repeats (STR), and for brown bear a sex marker, on a high‐throughput sequencing platform. Individual genotypes were obtained for all species, but genotyping performance differed among samples and species. The proportion of samples genotyped to individuals was higher for brown bear (5/7) and wolf (7/10) than for lynx (4/9), and locus genotyping success was greater for brown bear (0.88). The sex marker was typed in six out of seven brown bear samples. Results for three species show that reliable individual genotyping, including sex identification, is now possible from eDNA in snow tracks, underlining its vast potential to complement the non‐invasive genetic methods used for wildlife. To fully leverage the application of snow track eDNA, improved understanding of the ideal species‐ and site‐specific sampling conditions, as well as laboratory methods promoting genotyping success, is needed. This will also inform efforts to retrieve and type nuclear DNA from other eDNA samples, thereby advancing eDNA‐based individual and population‐level studies.


| INTRODUC TI ON
Environmental DNA (eDNA) sampling and analysis, using organismal DNA extracted from environmental samples (Taberlet et al., 2012), are revolutionizing the way we assess biodiversity, enhancing the scope of ecological investigations and conservation studies (Beng & Corlett, 2020;Cristescu & Hebert, 2018;Deiner et al., 2017Deiner et al., , 2021;;Taberlet et al., 2018).So far, eDNA applications have primarily focused on species detection and ecosystem-level diversity (Beng & Corlett, 2020), but continued advancements within eDNA research have resulted in increased effectiveness of approaches for recovering eDNA potentially suitable also for addressing intraspecific diversity and population-level questions (Adams et al., 2019;Sigsgaard et al., 2020).
In the context of wildlife studies of macroorganisms, the ability to access intraspecific genetic variation from various eDNA sources represents an advancement in non-invasive genetic methods typically based on the collection of scats, hair, feathers, urine, etc. (Andrews et al., 2018;Waits & Paetkau, 2005).DNA traces in the environment, in fact, offer new opportunities to noninvasively genetically sample animals in their natural setting, without handling or even observing them (Adams et al., 2019).One main challenge is that environmental samples comprised DNA of several species and individuals of the same species, all diluted in the sample matrix and contributing unequal amounts of DNA to the eDNA mixture (Barnes & Turner, 2016;Sigsgaard et al., 2020).
However, through targeted eDNA sampling aimed at maximizing DNA retrieval of the target species and sometimes individuals, researchers have been able to assess mitochondrial DNA (mtDNA) haplotype diversity, frequency and distribution and even compile mitogenomes (Dugal et al., 2022;Farrell et al., 2022;Parsons et al., 2018;Sigsgaard et al., 2016;Székely et al., 2021).Further, studies are now showing real potential for calling nuclear variants in eDNA samples for use in a population genetic framework (e.g.Andres et al., 2021;Jensen et al., 2021).
Reliable analysis of nuclear DNA (nDNA) is key in enabling eDNA-based population studies because of the higher information content and resolution of nDNA compared to mtDNA (Adams et al., 2019;Sigsgaard et al., 2020).Typing of nDNA will also allow for individual identification, which is the basis of wildlife noninvasive genetic surveys, genetic monitoring programs and forensics (Kelly et al., 2012;Ogden et al., 2009;Schwartz et al., 2007).
However, individual genetic profiling from eDNA sources remains elusive.Retrieving nDNA of a target species from an environmental mixture in sufficient quantity and quality is more difficult compared to mtDNA because nDNA is present in significantly lower copy number (except for the multi-copy regions) and it degrades faster due to the absence of organellar membranes protection (Parsons et al., 2018;Sigsgaard et al., 2020).
Snow tracks, that is, footprints left by animals while walking in the snow, are an ideal setting for targeted eDNA sampling in population-level wildlife studies.eDNA from animal tracks originates from cells present on the animal paw and deposited on the snow surface due to friction against the ground.Therefore, a first advantage of snow track eDNA sampling in terrestrial ecosystems is that an animal's DNA is found in a well-delimited area as opposed to samples from aquatic environments where eDNA dilution and mixing from multiple sources is greater (Dalén et al., 2007;Franklin et al., 2019;Howell et al., 2021).This feature also increases the chances of collecting DNA from single individuals of the target species.Secondly, snow limits DNA degradation by acting as a natural freezer (Dalén et al., 2007;Howell et al., 2021) and hence facilitates the preservation of nDNA.Finally, snow tracks of terrestrial animals are commonly found in winter in snowy ecosystems (Kinoshita et al., 2019), potentially allowing for adequate sample sizes in population studies.
Snow track eDNA has already been used for species detection of several predators through mtDNA analysis (Barber-Meyer et al., 2020;Dalén et al., 2007;Franklin et al., 2019;Kinoshita et al., 2019).A number of published studies have attempted to analyse nDNA with varying results for lynx (Hellström et al., 2019), wolf (Barber-Meyer et al., 2020, 2022) and polar bear (Von Duyke et al., 2023) with only this latter recent study being successful in achieving reliable multilocus genotyping for individual identification in a single species.However, individual genotyping from snow track eDNA as a wildlife non-invasive genetic method still remains elusive.Several reasons have been called into play for the earlier failures, spanning from field conditions and collection methods to laboratory protocols.All these previous works evaluated the amplification and genotyping performance of existing microsatellite loci (i.e.short tandem repeats -STR) either on agarose gel or by capillary electrophoresis.
In this study, we present the first successful individual genotyping from eDNA in snow tracks of three large carnivore species in temperate ecosystems: brown bear (Ursus arctos), wolf (Canis lupus) and Eurasian lynx (Lynx lynx).We sampled snow tracks in the field and used an extraction protocol for water eDNA samples and a genotyping approach based on amplicon sequencing of STRs and a sex marker on a high-throughput sequencing (HTS) platform (Figure 1a).
We discuss genotyping success in relation to field conditions, the ecology of eDNA (Barnes & Turner, 2016) of the three species and laboratory protocols with implications for advancing the use of eDNA approaches for population-level wildlife studies (Wilcox & Jensen, 2022).

| Snow track eDNA sampling
Snow tracks were collected opportunistically during winter in 2019, 2020 and 2022 in the Slovenian Alps and Dinaric Mountains (seven brown bear samples and nine lynx samples) and in the French Alps (10 wolf samples) (Table 1).Field personnel including volunteers, field biologists and park/forest rangers performed the sampling in areas known for the stable presence of the species.
Samples were collected upon discovery of trails of snow tracks visually attributed to the target species.Brown bear is the only ursid in southern Europe and it occurs at high density in the study area.Footprints of adult brown bears are readily distinguishable from other wildlife.Lynx and wolves are closely monitored as part of ongoing projects.Therefore, to locate trails on snow for these species, we took advantage of available fine-scale information on presence and movement from GPS-telemetry and camera trapping for individual lynx and wolves within previously identified packs.
A sterilized spoon was used to scrape the surface of a snow track and place the snow in a sterile plastic bag (Fisherbrand Sterile Polyethylene Sampling Bags, 10″ × 12″).Multiple bags were used when larger volumes of snow were collected for a sample.The number of tracks collected for a given sample ranged from 1 to 17. Sampling location, sample characteristics and environmental conditions at the sampling site were recorded by field operators (Table 1, Table S1).Plastic bags containing the snow were labelled and transported frozen to the genetic laboratory, where they were kept at −20°C until DNA extraction.

| Snow track eDNA extraction
We extracted DNA from snow track samples using the DNeasy PowerWater Sterivex Kit (Qiagen, Germany) following manufac- Snow samples in plastic bags were completely thawed at room temperature (this took up to 24 h, depending on the amount of snow).The following day, melted snow was left to settle until large forest debris was deposited on the bottom of the bag.For each sample, the resulting water was filtered through a Sterivex filter (Millipore cat.no.SVGPL10RC) using a 60-mL volume syringe (Omnifix Luer Lock Solo 50 mL).We measured the amount of water filtered by collecting it in a graduated container.For two brown bear samples, we performed two extractions for each sample using two filters because the first filter clogged before filtering all the available water (this resulted in a total of nine DNA extractions analysed for the brown bear) (Table 1).Once all samples were filtered (this step took up to a full working day), filters were stored in a freezer at −20°C until the next morning.DNA extraction was completed following the kit protocol, omitting the incubation step at 90°C and the steps with the PowerBead Tubes as recommended for samples containing easy-to-lyse organisms or where less DNA shearing is desired.The centrifuge was used instead of the vacuum manifold with kit handbook settings and collection tubes provided with the kit.DNA was eluted in 100-μL volume.An extraction negative control was included with all sets of extractions to monitor contamination and was processed with the snow samples in all subsequent stages of the analysis.
DNA extraction and the following PCR set-up were carried out in a room dedicated to low-quantity/quality DNA samples.

| STR amplicon sequencing
We performed individual profiling using genotyping by HTS of STR amplicons (De Barba et al., 2017;Fordyce et al., 2011).For each species, we used a set of STR markers designed for optimal multiplex amplification and HTS genotyping.The brown bear set includes 13 STR recently described and used for individual profiling from faecal DNA (De Barba et al., 2017), with the addition of a sex-specific marker (Pagès et al., 2009).For wolf and lynx, we used 13 new STRs (Table S2) developed following criteria outlined in De Barba et al. (2017).
For each species, STRs (and a sex-specific marker for brown bears) were co-amplified in a single multiplex PCR.Reactions were carried out in a 20-μL volume and contained 1× concentrated Platinum Multiplex PCR Master Mix, 1% GC enhancer (brown bear) or 0.0032 mg of BSA (lynx, wolf), 0.035-0.1 μM of each primer (Table S2) and 2-μL DNA template.The thermocycling profile had an initial denaturation step of 2 min at 95°C, followed by 50 cycles of 30 s at 95°C, 90 s at 57°C (brown bear)/60 s at 55°C (wolf, lynx), 60 s at 72°C and a final elongation step of 10 min at 72°C.Amplifications were performed in eight replicates per sample, following a full multitube approach (Taberlet et al., 1996).
Tagged primers, modified by the addition of molecular identifiers on the 5′ end, were used in each PCR to uniquely label any given PCR product for retrieving the respective sequence data in postsequencing bioinformatic analysis.Tags consisted of eight nucleotides enabling a minimum of five mismatches between any pair of tags (Coissac, 2012).An additional 1-2 specified nucleotides were added to the tags 5′ end to increase complexity for cluster detection on the flow cell.PCR negative (water) and positive (a non-invasive DNA sample previously successfully genotyped) controls and "tagging system" controls (corresponding to unused tag combinations) were included in the PCR set-up to facilitate the detection of potential contamination, false positive caused by tagjumps (Schnell et al., 2015), and monitor the performance of the amplification and the sequencing process (De Barba et al., 2014;Zinger et al., 2019).
For each species, PCR products were pooled equivolume, pu-

| Bioinformatics analysis of the sequence data
DNA sequence data analysis was performed using a modified version of the pipeline published in De Barba et al. (2017) (Figure 1a), implemented using in-house Python and R scripts, on a standard desktop computer running Linux or MacOSX (pipeline description available at https:// github.com/ Pazhe nkova EA/ ngs_ pipel ines.py).
Initially, Illumina reads were processed using the OBITools3 (Boyer et al., 2016) to assemble paired-end reads, filter out unaligned sequences, demultiplex sequences by markers and samples discarding sequences without a perfect tag match and at least three primer mismatches.STR alleles were inferred from the observed sequences and relative read counts in each PCR product following the process already described in De Barba et al. (2017).In summary, alleles were defined as the most abundant sequences containing the STR motif of the locus and associated with their relative stutter sequence.If a sequence had no stutter and a lower number of reads than the userdefined threshold (default 100 reads), it was discarded.Consensus genotypes at each locus for a sample were determined based on STR sequence alleles observed across the eight PCR replicates, requiring that an allele be observed at least twice for heterozygotes and three times for homozygotes.Similarly, with the sex marker, males were scored by the detection of the homologous X and Y sexual chromosome sequences in at least two replicate PCRs, while females were scored by the detection of the X chromosome sequence in at least three replicate PCRs.

| Genotyping performance and individual identification
For each sample, we estimated i. amplification success (AS), as the proportion of positive PCR replicates at each STR locus, that is, replicates yielding reads assigned to at least one allele sequence, averaged across loci; ii.rate of allelic dropout (ADO) and iii.rate of false alleles (FA) averaged across loci following formulas in Broquet and Petit (2004) using data for each PCR replicate compared to the consensus; iv.locus genotyping success (GS), as the proportion of loci analysed for which a consensus genotype was obtained and vi. the quality index (QI), as the proportion of PCR replicates at each locus in which the consensus genotype was observed, averaged across loci (Miquel et al., 2006).
For each species, we calculated overall multilocus genotyping (1-3 MM) were scrutinized to determine whether mismatches could have been caused by genotyping errors, assuming that samples with no mismatches (0 MM) were left by the same individual.However, with fewer alleles observed (i.e.1-2 at several loci) and error-prone samples (i.e.QI <0.5), we adopted more stringent criteria for individual assignment, as, under these premises, genotyping errors could be difficult to distinguish from true genotypic differences.In these cases, we specifically checked if mismatches between pairs of genotypes involved different alleles at some of the loci (i.e.MM not compatible with ADO/FA) before assigning samples to different individuals.In addition, we used field notes (Table S1) and available monitoring data, that is, about the presence or transit of single/multiple individuals at the sampling site, to ascertain dubious individual assignment and the ability for accurate individual genotyping.For lynx, we also disposed of genotyping data previously obtained at the same markers from samples collected from collared animals that were compared with snow track genotypes.
Genotypes were organized in a custom Microsoft Access database.All calculations were performed in R (v.4.2.1) and Microsoft Excel.

| RE SULTS
Sequencing of the snow track samples generated 4,818,564 reads assigned to markers and samples, 3,667,315 for brown bear, 224,061 for wolf and 927,188 for lynx, with an average of 1529 (bear), 228 (wolf), 876 (lynx) reads/marker/PCR that were used for genotyping.
The average proportion of reads cumulatively attributed to alleles for all loci multiplexed in an amplification reaction was 59% (14%-82%) across all samples.Remaining sequences included stutter sequences and a variable number of less abundant sequences originating from PCR and sequencing errors.The level of reads observed in the negative and tagging system controls was very low in general, and negligible in the samples prepared with the Tagsteady protocol (Appendix S1).
Short tandem repeats genotyping performance differed among samples analysed and for the three species (Tables 1 and 2).Brown bear samples showed generally higher genotyping success, resulting in a consensus genotype for 6-13 of 13 loci (GS = 0.46-1).However, among the five lynx samples that had non-zero GS, three samples had 12 out of 13 loci genotyped (GS ≥0.92), and among the eight wolf samples that had non-zero GS, seven samples had at least 11 out of 13 loci genotyped (GS ≥0.85).Number of alleles per locus in the samples analysed was 2-5 for brown bear, and 1-3 for both lynx and wolf (Tables S2 and S3).
A consensus genotype at ≥7 loci was reached for eight of the brown bear DNA extracts corresponding to six snow track samples (Table 2).One sample genotype (CX.113E) had >2 alleles at three  2).The sex marker was typed also for the mixed sample, but sex ID could not be ascertained in this case (Table 2).
For the lynx, four samples were genotyped at ≥7 loci (Table 2).
Despite low QI values for most samples (Table 1) and low allelic diversity (Tables S2 and S3), three could be reliably assigned to three  S1), were left by an adult lynx and a younger individual possibly stepping on the same tracks, initially raising concerns on the ability of distinguishing their genotypes.However, these two genotypes matched those previously determined from buccal swabs collected from a GPS-collared female lynx monitored in the area and from her kitten, supporting reliable individual identification.
For the wolf, seven samples were genotyped at ≥7 loci (Table 2).
These samples were collected from the area occupied by a single pack (Table S1) and presented low allelic diversity, that is, 1-2 alleles at most loci (Tables S2 and S3).In addition, they had low QI values (Table 1).After accounting for genotyping errors and consulting field notes, sample genotypes could be assigned to at least two individuals detected in two and five samples, respectively, resulting in 70% MGS (7/10 samples).Specifically, the two sample genotypes assigned to one individual (W1) matched at all genotyped loci except two (Cl285 and Cl291), with allelic differences compatible with ADO/FA.In addition, they had, respectively, 3-6 (sample Neige-2,1) and 5-7 (sample Neige-2,3) locus mismatches with sample genotypes assigned to the other individual, with mismatches involving different alleles.The other five samples were all conservatively assigned to a second individual (W2).Their sample genotypes differed at six loci (1-6 mismatches between pairs of sample genotypes) with mismatches compatible with ADO (loci Cl233, Cl285, Cl291, Cl308, Cl527) and FA (locus Cl375).However, field notes reported the possible presence of two individuals in some of the samples (Table S1).
Therefore, we could not exclude that mismatches are actually true genotypic differences or that the DNA profile obtained from some samples resulted from DNA mixing within a track of related individuals with highly similar genotypes.Consequently, W2 genotype remains to be validated and the wolf snow tracks analysed can only indicate the detection of at least two individuals.
Figure 1b provides a schematic illustration of the subsequent decision-making steps described above for assigning sample genotypes to individuals.

| DISCUSS ION
In this study, we successfully performed individual genotyping of STRs for three large carnivore species, and of a sex marker for one of these species, using snow track eDNA.Multilocus genotyping success rates for individual identification were in the range of those reported for the species using non-invasive genetic sampling   Highly variable per sample AS, GS and QI, as well as genotyping error rates similar to non-invasive genetic studies (e.g.0.016-0.41ADO and 0.002-0.08De Barba & Waits, 2010;Dufresnes et al., 2019;Sindičić et al., 2013;Skrbinšek et al., 2019), suggest that, under certain conditions, snow track eDNA can be preserved and recovered in suitable quantities and quality to allow reliable individual genotyping of nuclear loci.We were able to genotype samples stored in the freezer for over 2 years, including samples taken from a single snow track.Compared to other eDNA sources, targeted sampling at snow tracks is expected to facilitate the detection of individual nDNA, thanks to favourable preservation on the snow substrate and limited DNA mixing (Franklin et al., 2019;Howell et al., 2021).Environmental conditions of the sampling sites and sample characteristics (i.e.number of tracks, track age, size, etc.) varied considerably in our study, and limited sample sizes prevented us from identifying clear patterns and applying statistical testing about factors driving genotyping performance.Still, most samples were collected within 1 day or even a few hours, and all within 3 days since the estimated time of animal passage.Furthermore, collection of several tracks or filtering of larger volumes did not systematically result in higher genotyping success, suggesting that interactions among sample and environmental variables or factors other than those recorded in the field are also key determinants of genotyping success.
We reported differing results between the species considered.
Brown bear tracks had higher genotyping performance, and in particular considerably higher overall GS and lower ADO rates, than wolf and lynx tracks.However, most of the samples that were amplified, regardless of the species, resulted in high GS.While we cannot rule out that these differences are due to sampling conditions or laboratory methods in the different laboratories (see below), it may also suggest that the ecology of eDNA of a species could play a role in determining genotyping success.The ecology of eDNA refers to the combination of factors and processes influencing DNA production, state, transport and degradation in a given environment (Barnes & Turner, 2016).For snow track eDNA, this is relevant because the amount and state of DNA shed by each species may differ due to biological and behavioural differences between them, playing a role in eDNA preservation and retrieval.For example, brown bears have larger paws and are heavier, perhaps resulting in more skin cells Note: Different grey shadings indicate locus data for sets of sample genotypes that were assigned to the same individual but that differed at that locus, with differences compatible with genotyping errors (ADO, FA).Allele sizes followed by underscore indicate allele variants of the same base pair length as another detected allele, but differing by sequence polymorphism (Table S3).Beside the sampling conditions discussed above, laboratory protocols, from DNA extraction and amplification to the genotyping approach, differed compared to previous snow track genotyping studies (Barber-Meyer et al., 2020, 2022;Hellström et al., 2019;Von Duyke et al., 2023) and may have contributed to genotyping success.A major difference was the adoption of an HTS approach for amplicon sequencing of STRs.Markers analysed are short (<120 bp) tetranucleotides, selected for optimal multiplexing, to facilitate am-  (Kelly et al., 2012;Mumma et al., 2015), will also  tions (Barber-Meyer et al., 2020, 2022;Hellström et al., 2019;Howell et al., 2021).We further recommend that these effects be assessed for various target species in their ecosystem in order to evaluate species-and site-specific differences in eDNA deposition and deg- turer's instructions (DNeasy PowerWater Sterivex Kit Handbook F I G U R E 1 Workflow of snow track eDNA genotyping.(a) Components of the workflow from eDNA sampling to individual identification, with main steps of the data analysis outlined.(b) Flowchart of the matching and validation of individual assignment process for pairs of sample genotypes, detailing how brown bear, lynx and wolf snow track samples were assigned to individuals.In (b) blue text indicates the sample genotypes for each specific case described, while grey dashed arrows indicate cases not represented in the sample genotypes analysed.ADO, allelic dropout; FA, false allele; MM, mismatches; QI, quality index.Snow tracks photo credits: Miha Krofel.05/2019) with slight modifications as described below.We processed 10-12 samples at a time, with each set of extractions taking 3 working days: the first day for snow melting, the second day for water filtering and the third day to complete the DNA extraction.
success (MGS), as the proportion of samples that were identified to individual.Sample individual assignment was a multistep process that considered all genotypic, field and ecological information available for the analysed samples and the species in the study area(Figure 1a,b).We first required that samples had a consensus genotype obtained at >50% of the STR loci analysed and excluded samples with more than two alleles detected at several loci.Then, to reliably assign samples to different individuals, we evaluated sample genotype similarity by calculating the number of locus mismatches between pairs of sample genotypes(Paetkau, 2003) using a custom R script (provided in Supplementary Information).With moderate/ high allelic diversity (i.e.>2 alleles at most loci for the genotypes compared) and sample QI ≥0.5, sample genotypes with ≥4 mismatches (4 MM) were considered as originating from different individuals.Pairs of similar genotypes presenting 1-3 locus mismatches loci (UA06, UA16 and UA51) indicating a possible mixed sample containing DNA from multiple individuals.Each of the genotypes of the remaining five samples had at least four-locus mismatches with genotypes of other samples and was assigned to an individual, resulting in MGS = 71.4% (5/7 samples) for individual identification.The genotype identified from the brown bear tracks extracted using two filters matched between duplicate extractions, except for one allele difference at one locus, due to ADO or FA (locus UA14 and UA64, respectively, in each of the duplicate extraction sets).Sex was successfully identified from all five (one female and four males) of the six brown bear samples for which an individual genotype was obtained, and was concordant among duplicate DNA extracts (Table different individuals L1, L2 and L3 (≥4 MM, including differing alleles).A fourth sample (CX.1158), collected in the same area and day of one of the unique genotypes (L1), was considered having originated from the same individual after accounting for possible ADO at three loci (LL0043, LL0044 and LL0125) and given differing alleles at two loci from the other unique genotypes.This resulted in MGS = 44.4% (4/9 samples genotyped to individual).L1 genotype matched that of a lynx sampled the same day from a hair tuft collected in the area (lynx monitoring data not shown).The other two unique genotypes (L2 and L3) were identified from samples that, based on field notes (Table B L E 2 Snow track genotypes and individual and sex identification at the STR loci analysed for brown bear (a), lynx (b) and wolf (c) and a sex marker for brown bear (a).
of snow tracks collection (wolf hair, saliva, scat, regurgitate and urine: 22%-60%Dufresnes et al., 2019; lynx hair, scat and   urine: 9.4% Sindičić et al., 2013; brown bear scats: 88% Skrbinšek   et al., 2019).In addition, the detection of individuals genotyped independently from other DNA sources, multiple observations of a multilocus genotype within the samples analysed and genotyping concordance from duplicate DNA extractions support the ability for accurate profiling.While the current study is a proof of concept, results for three different species show that reliable individual genotyping, including sex determination, from snow track eDNA of wild animals is possible, underlining its great potential for complementing wildlife non-invasive genetic sampling methods, with exciting prospects to expand ecological and conservation studies.

a
Heterozygote 104 104_2: male; homozygote 104 104: female.b First set of duplicate DNA extractions.c Possible mixed sample with >2 alleles at three loci, no individual assigned.d Second set of duplicate DNA extractions.TA B L E 2 (Continued)being deposited on the snow.In addition, brown bears are known to exhibit pedal marking behaviour, actively twisting their feet on the ground(Sergiel et al., 2017).The amount of DNA left on snow by individuals of a species could be affected by other behaviours, such as animals licking their paws for self-grooming.Beside the amount of DNA, animal behaviour could also affect the accuracy of individual genotyping.For example, it is not unusual for some species, including large carnivores, to step on tracks left by other individuals(Liberg et al., 2011;Sergiel et al., 2017), potentially resulting in eDNA sampling from multiple individuals (mixed samples).If not considered during sample collection or in the sampling design of a study, such instances can imperil individual genotyping efforts and bias results.While using STR genotyping in an outbred, genetically diverse population, mixed DNA profiles would typically be revealed through the presence of >2 alleles at several loci (as in one of the brown bear samples in this study), their detection could be subtler when related individuals are involved.This was a concern with the samples from a parent-offspring lynx pair and from a wolf pack in our study, as high genotype similarity could have resulted in the detection of an erroneous profile resembling that of a single individual.In such cases, field information about track characteristics and knowledge of the study system (i.e.presence of individuals of the target species), as available in our study, could be very important for ascertaining genotyping data and assessing if accuracy can be ensured.Under some circumstances, mixed samples could be resolved at the individual level (e.g. with animals known to be in the area and whose genotype has already been determined).Nonetheless, even if individual identification is prevented, detection and reporting of mixed samples will benefit data accuracy in population studies and support wildlife forensics, management and conservation, for example, indicating the presence of >1 individual at the sampling site, or informing on the efficiency of a sampling method for detecting individuals.
profit from genotype data collected through snow tracking.Snow track eDNA can complement other genetic sampling methods, by increasing individual detection and sample sizes, that is, for all age/ sex classes or for the winter season, supporting more effective population monitoring and identification of targeted individuals for management purposes(Barber-Meyer et al., 2020).These systems, where ecological information is already available for the study species, are also those that would allow the most robust use of snow track eDNA for reliable individual identification.Here, the genotypes obtained from snow tracks can be used in association with other available field or genotype data to compensate for possible bias associated with snow track sampling, specifically high ADO rate and eDNA sampling of multiple individuals.To fully leverage the potential of snow track eDNA genotyping, future studies should work on aspects relating to both sampling in the field and laboratory analysis.In the field, efforts should be directed towards a thorough understanding of the optimal conditions for snow track sampling, investigating factors affecting genotyping success and accuracy related to the sampling site and methods and considering the eDNA ecology of target species.Previous studies have already stressed the importance of understanding the effect of track age, number and conditions of tracks sampled, temperature and UV exposure, equipment utilized for sampling and storage condi- radation on snow tracks, and ideal sampling conditions for detection of individuals.In the laboratory, we emphasize the importance of DNA extraction protocols maximizing the amount and the quality of DNA retrieved from snow tracks, as well as investigating how sample treatments, e.g. the effect of thawing snow at room temperature for several hours, may affect DNA degradation and observed genotyping performance.We also recommend using highly discriminating individual profiling approaches optimized for accurate detection of low-level allele signals to increase genotyping sensitivity and inform about mixed samples.This includes the employment of library preparation protocols specifically developed for minimizing the occurrence of spurious sequences and therefore the noise-to-allele ratio.The acquisition of comprehensive knowledge of the multiple factors affecting genotyping success and accuracy is paramount to inform the implementation of cost-effective snow track eDNA sampling efforts for large-scale wildlife surveys, monitoring and population studies in terrestrial ecosystems.Additionally, understanding of the drivers of genotyping success in the simplified snow track system would also inform efforts of nDNA retrieval and typing in more complex eDNA samples, such as water and soil samples.AUTH O R CO NTR I B UTI O N S P.T. conceived the idea.M.D.B., L.F., T.S. and P.T. designed the study.F.B., M.B., M.D.B., M.K., C.M., E.P., N.R., C.S. and T.S. conducted laboratory work and data analyses.M.D.B. wrote the paper with input from all co-authors.
(Von Duyke et al., 2023)mp reads is a concern because it could reduce allele detection and also lead to inaccurate genotyping.The low level of reads observed in the controls indicates that spurious sequences, including tag-jumps, were not a problem with both protocols and were actually negligible with the Tagsteady protocol used for lynx and wolf samples.Sequencing coverage was, on average, higher for brown than for Janečka, 2013) and the polar bear (Ursus maritimus)(Von Duyke et al., 2023)among other species for which population data are lacking.In addition, species commonly monitored through noninvasive genetic sampling, such as wolves, brown and black bears and mesocarnivores (Carøe & Bohmann, 2020)tions differed between brown bear samples and wolf/lynx samples, and a higher number of sequence reads was available for genotyping brown bear samples.Libraries for wolf and lynx samples were prepared using a protocol especially developed for minimizing tag-jumps that can form at different steps of the library preparation(Carøe & Bohmann, 2020), while a proprietary protocol was used for preparing the brown bear samples.