Abstract.

Noninvasive fecal genotyping can be a useful tool for population monitoring of elusive species. We tested extraction protocols on scat samples from the threatened Mojave Desert tortoise, Gopherus agassizii, to evaluate whether scat-based mark–recapture and population genetic monitoring studies are feasible.We extracted DNA from G. agassizii scat samples collected in California and Nevada using several extraction protocols and evaluated the reliability of resulting genotypes using quality scores, maximum likelihood reliability estimates, and paired scat and blood genotypes from the same individuals. Finally, we assessed probabilities of identity and sibship, and locus amplification quality, and calculated genotyping error rates for 19 microsatellite loci to determine the best set of loci to use with G. agassizii scat extractions. We found that genotype quality depended more on the sample quality than on the extraction method, and that the Qiagen DNeasy Plant Mini extraction kit is an efficient method for extracting tortoise DNA from tortoise scat. We identified 6 G. agassizii microsatellite loci that can be used to generate a unique molecular tag for individual tortoises. We characterized the reliability of an additional 13 microsatellite loci for use in population genetic analyses where additional power at the expense of some increase in error may be advantageous. As proof of concept, with very low error rates, we matched 3 opportunistically collected scat samples to blood genotypes from animals captured during population surveys within the study area and discovered at least 3 new individuals, even after 2 yrs of extensive survey work. These results suggest that genotyping of field-collected scat can complement existing methods used in long-term demographic and movement studies of G. agassizii and other, closely related, tortoise species.

The Mojave Desert tortoise, Gopherus agassizii, occurs in the Mojave Desert and in parts of the Sonoran Desert of the United States (US Fish and Wildlife Service [USFWS] 2011; Edwards et al. 2015). Because of decreases in population densities, the Mojave population of G. agassizii was listed as threatened under the federal Endangered Species Act in 1990 and also receives state protection in California, Nevada, Arizona, and Utah (USFWS 1990). Numerous factors threaten G. agassizii including loss, degradation, and fragmentation of habitat due to an expanding human footprint throughout its range (USFWS 2011). Currently, populations are monitored using traditional methods such as plot surveys, radiotelemetry, and genetic samples (i.e., blood) collected when animals are handled. However, these elusive herbivores can be difficult to detect because of their cryptic appearance, behavior, and propensity to use burrows. This results in many animals being missed during surveys (Freilich et al. 2000; Anderson et al. 2001; Nussear and Tracy 2007; Nussear et al. 2008). Therefore, to complement current methods implemented for studying G. agassizii populations, we explored the potential of using scat samples, which can be opportunistically collected during other survey efforts, to yield genetic data for monitoring.

Noninvasive scat sampling and individual identification through genotyping have become useful tools for population monitoring of wildlife species (Schwartz et al. 2007). Scat-based genotyping can provide information on individual movement, home range, relatedness, abundance, food habits, parasite load, and sex ratios (Waits and Paetkau 2005; Schwartz et al. 2007; Beja-Pereira et al. 2009; Luikart et al. 2010). Using scat-based genetic sampling without the need to handle individual animals is appealing for studies of elusive, rare, or endangered species (Piggott et al. 2008; Giambattista and Gentile 2018). While monitoring mammals using scat-based genetic methods has become fairly common (e.g., Piggott et al. 2008; Mitelberg and Vandergast 2016; Arandjelovic and Vigilant 2018), these methods have not been applied as often to herpetofauna (but see Pearson et al. 2015; Giambattista and Gentile 2018).

Scat-based genotyping must account for higher rates of genotyping errors such as allelic dropout (ADO), which is the failure to amplify one of the two true gene copies present at a locus, and false alleles (FA), which can arise through the amplification of spurious polymerase chain reaction (PCR) fragments (Taberlet et al. 1999; Bonin et al. 2004; Pompanon et al. 2005). These errors can result from degradation due to environmental exposure (Murphy et al. 2007; Brinkman et al. 2010; Panasci et al. 2011) and the presence of inhibitory compounds in scat (i.e., complex polysaccharides, bile salts, lipids, urate) that can interfere with sample extraction and PCR (Schrader et al. 2012). Genotyping errors have the potential to bias the conclusions of individual-based analyses, such as population size estimates (Waits and Leberg 2000) and parentage analyses (Gagneux et al. 1997), and can also influence populationlevel analyses (Pompanon et al. 2005). Therefore, it is important to conduct a pilot study to determine if reliable and informative DNA fingerprints can be obtained from G. agassizii scat.

In this study, we evaluated whether genetic monitoring using scat was feasible for G. agassizii. Specifically, we addressed 3 main questions. First, could a reliable multilocus genotype that is adequate for individual identification (a molecular tag) be obtained from DNA extracted from G. agassizii scat? Second, of the extraction protocols evaluated in the study, which was the most reliable and cost/time efficient? And third, which microsatellite loci were the most informative for population genetic analyses while limiting genotyping error? To address these questions, we tested 6 extraction protocols, evaluated the quality of genotypes derived from these extractions using several quality metrics, and estimated the cost of these methods. The extraction protocols tested in this study can be applied in long-term demographic and movement studies of G. agassizii and could extend to other, closely related, tortoise species.

METHODS

Collection and Storage. — We used scat and blood samples obtained from G. agassizii individuals, the majority of which were collected within the Ivanpah Valley in California and Nevada between March 2016 and October 2018 (Fig. 1). Paired scats and blood were collected directly from animals handled in the field for radio-tracking, marking, and health assessments. Additionally, to test the utility of genotypes obtained from fieldcollected tortoise scat, 11 fresh scats (recently deposited with sheath shiny, smooth, and intact) from unknown animals were collected opportunistically from the ground during the same sampling period.

Figure 1.

Study area in the Mojave Desert at the border of California and Nevada, USA. Locations of paired Gopherus agassizii scat and blood samples are shown as dots and unknown, opportunistically collected scats as triangles. Locations of all captured and genotyped tortoises used as reference samples are shown as black crosshatches.

Figure 1.

Study area in the Mojave Desert at the border of California and Nevada, USA. Locations of paired Gopherus agassizii scat and blood samples are shown as dots and unknown, opportunistically collected scats as triangles. Locations of all captured and genotyped tortoises used as reference samples are shown as black crosshatches.

Blood samples were stored at room temperature on Whatman cards. Scat samples were collected either in closed, 50-ml tubes with desiccating beads or in individual, non–air-tight plastic boxes or paper lunch bags and allowed to dry outdoors overnight (samples in boxes were dried with lids open). Increased air contact appeared to inhibit fungal development. A similar collection method has worked well in studies genotyping mule deer scat (Bohonak and Mitelberg 2014; Mitelberg and Vandergast 2016). To simulate the approximate age and quality of samples that would be collected in the field, a subset of 3 fresh samples were additionally exposed to 7–9 d of full sun. Scat samples were stored dry in the laboratory at room temperature and extracted within 3 wks of collection.

Blood Extractions. — All blood extractions were performed with the Qiagen DNeasy Blood and Tissue extraction kit, according to the manufacturer-provided protocol, with these minor modifications to improve yields: following addition of Buffer AL, samples were incubated at 70°C for 10 min; elution volume was halved to 100 µl, and the elution step was performed twice, for a total of 200 µl final elution volume.

Scat Extractions. — We experimented with 6 scat extraction protocols consisting of combinations of different pre-extraction surface washes, extraction kits, and postextraction clean-up kits to determine if any of these improved quality (Table 1). Five of these extraction protocols used the Qiagen QIAmp Fast DNA Stool Mini extraction kit (hereafter stool kit) on surface-washed cells, and one protocol used the Qiagen DNeasy Plant Mini extraction kit (hereafter plant kit) on whole, homogenized scat. Scats were split in half prior to extraction to increase sample sizes across extraction protocols and to provide paired samples for protocol comparisons. To minimize opportunities for contamination, we used new disposable gloves for each scat sample and performed all extraction steps (with the exception of centrifugation) under a PCR hood.

Table 1.

Extraction protocols including pre-extraction surface wash method, extraction kit, and clean-up kit (if used), number of Gopherus agassizii scat samples extracted, approximate time, cost, and number of steps per protocol.

Extraction protocols including pre-extraction surface wash method, extraction kit, and clean-up kit (if used), number of Gopherus agassizii scat samples extracted, approximate time, cost, and number of steps per protocol.
Extraction protocols including pre-extraction surface wash method, extraction kit, and clean-up kit (if used), number of Gopherus agassizii scat samples extracted, approximate time, cost, and number of steps per protocol.

Stool Kit with Surface Wash and Clean-Up Steps. — Scat surface washes utilized Inhibitex buffer (included in the stool kit) by either 1) placing the scat segment into a 50-ml centrifuge tube with buffer and agitating on a nutating rocker for 10–15 min (“tube surface wash”; Table 1), or 2) placing the scat in a weighing dish with buffer and leaving the dish on the nutating rocker for 10–15 min (“dish surface wash”; Table 1). Following washes, remaining buffer and epithelial cells were transferred by pipette to 2-ml centrifuge tubes and extracted with the stool kit following the manufacturer's protocol. As an additional measure for removing PCR inhibitors from extracted DNA, an aliquot of each extraction was processed through either the OneStep PCR Inhibitor Removal kit (Zymo) or the Genomic DNA Clean & Concentrator-10 kit (Zymo). We followed manufacturer's protocols for the OneStep kit and made the following modifications to the Clean & Concentrator-10 kit: a 1-min spin step was added following the use of Wash Buffer (Zymo) to facilitate removal of all ethanol, and two sequential elution steps with 10-min incubation periods were performed using 20 µl warmed (60°C–70°C) Elution Buffer (Zymo).

Plant Kit Extractions. — The DNeasy Plant Mini kit (plant kit) includes steps to shear cell walls and reduce PCR inhibitors often present in plant tissues, which could be useful given the herbivorous diet of the desert tortoise. We cut off a small piece (< 300 mg) of each scat sample (making sure to include material from the surface and interior of the scat) and extracted the piece using the manufacturer's protocol. For a paired comparison of the plant and stool kit protocols, the remainder of each of these scat samples was extracted with stool kit protocols 4 or 5 (Table 1).

Amplifications. — Blood and scat DNA extractions were amplified at either 6 or 19 microsatellite loci available for G. agassizii (Edwards et al. 2003; Schwartz et al. 2003; Hagerty et al. 2008). We assessed each locus individually in single-locus, 10-µl PCR reactions containing 5 µl 2×Multiplex PCR Plus cocktail (Qiagen), 1 µl 10 µM primer, 2.5 µl water, and 1.5 µl DNA (blood extractions were standardized to 4 ng/µl). Thermocycler conditions were as follows: enzyme activation for 5 min at 95°C followed by 30 and 40 cycles for blood and scat, respectively, of 30 sec at 95°C, and 3 min at 56°C, 45 sec at 72°C, with a final elongation at 68°C for 30 min. One microliter of PCR product was aliquoted into 10.5 µl Hi-Die formamide (Thermofisher) with 0.5 µl GeneScane 500 LIZt size standard (Thermofisher) and was submitted for genotyping to Eton Bioscience (San Diego, CA). GeneMarker v.1.90 was used to score chromatographs. Negative controls were included with each round of PCR to monitor contamination and all PCR reactions were prepared under a PCR hood.

Deriving Consensus Genotypes. — To determine whether a reliable molecular tag could be obtained from G. agassizii scat, we amplified all extractions at a subset of 6 microsatellite loci yielding the shortest PCR products (referred to as screening loci from this point on). Loci with shorter amplicons generally have higher amplification and lower error rates (Frantzen et al. 1998; Broquet et al. 2006). Additionally, successful genotyping of at least 5 of these 6 loci satisfies the recommended minimum threshold for probability of identity (PSIB ≤ 0.01; Waits et al. 2001). Each scat extraction was genotyped 3 times at each locus, with extractions from the same scat sample before and after postextraction clean-up being treated as separate extractions. Because each scat sample was split in half and subjected to 2 different extraction treatments, this yielded a total of 6 replicate PCRs for each scat sample.

To arrive at an “extraction consensus genotype”, we used the following guidelines of Frantz et al. (2003): 1) for heterozygotes, each allele must be present at least twice among replicate PCR reactions; and 2) for homozygotes, the allele must be present at least 3 independent times and all 3 times as a homozygote (to eliminate the possibility of ADO). In addition to a consensus genotype for each extraction, we were able to determine a “sample consensus genotype” for each scat sample using all 6 replicate amplifications. To arrive at the sample consensus genotype, we used the same rules as for the extraction consensus genotype.

Assessing Genotype Quality. — We used three approaches to estimate genotype quality across extractions, samples, and loci. First, we calculated genotype quality (Q-score) following Miquel et al. (2006). At each locus, genotypes were compared with the consensus genotype for that sample and a score of 1 was assigned for that replicate in the case of a match; all other genotypes (ADO, FA, PCR failure [FAIL], and lack of consensus genotype [UNK]) were assigned a score of 0. Q-scores were calculated for each extraction and sample, at each locus, by averaging scores for replicate genotypes. Second, we used the maximum likelihood software RELIOTYPE (Miller et al. 2002), which uses repeated amplifications to determine the probability that samples met certain reliability thresholds. We used default settings (95% reliability, 200 bootstrap replicates) and allele frequencies estimated from a larger sample of 159 tortoises collected throughout the Ivanpah Valley in 2016–2017 and genotyped from blood (Dutcher et al., in press). We applied the reliability criteria to the entire data set, limiting the incidence of false acceptances to less than 5% with 95% probability, and retained all instances of FA, as the focus of this study was to document potential issues with genotyping because of poor DNA quality. Samples accepted without further PCR replicates were deemed as suitable quality samples (SQS). Finally, for a subset of samples for which we had paired scat and blood samples available, we directly compared sample consensus genotypes of scat samples to their respective blood genotypes.

Evaluation of Scat Extraction Methods. — To evaluate extraction protocols, we compared Q-scores, cost, time, and steps involved in each extraction protocol. We used box plots to visualize Q-scores for the 6 extraction protocols we evaluated. We used a paired Wilcoxon signed-rank test to determine if there was a significant difference in Q-scores between samples extracted with the stool and plant kits. We limited this comparison to 16 samples that were divided and extracted with both kits (protocols 4, 5, and 6; Table 1).

Locus Suitability for Genotyping G. agassizii Scat. — To explore additional loci that may be useful for genotyping G. agassizii scat samples, we further genotyped 8 scat and 3 blood extractions at 19 loci, with 5 PCR replicates per locus. For scat, we limited this experiment to samples that were extracted with the plant kit (protocol 6; Table 1) and were deemed SQS samples when assessed at the 6 screening loci as described above. We calculated Qscores and analyzed reliability with RELIOTYPE. We calculated error rates for all 19 loci, following Broquet and Petit (2004), based on consensus or blood genotypes when available (in the case of 3 of 8 scat samples). We visualized average Q-scores for loci using box plots. To assess our power to reliably identify individual tortoises from scat samples, probabilities of identity (PID) and probabilities of sibship (PSIB) were calculated for these 19 microsatellite loci using allele frequencies estimated from the larger (159 sample) data set (Dutcher et al., in press).

Identifying Opportunistically Collected Scat. — To further test the utility of scat-derived genotypes, we conducted an identity analysis in CERVUS (v. 3.0.7; Kalinowski et al. 2007). We tested whether any of the opportunistically collected scat samples with SQS genotypes matched any tortoises genotyped from blood samples in Ivanpah Valley to date (309 individuals; Dutcher et al., in press). For this analysis, we used allele frequencies from this larger data set, allowing up to 1 mismatching locus to avoid excluding matches due to genotyping error.

RESULTS

Quality of G. agassizii Scat Sample Genotypes at Screening Loci. — Overall, we evaluated 11 blood samples and 25 scat samples (in 50 extractions). Of 25 total scat samples, RELIOTYPE identified 11 (44%) as SQS samples (Table 2). Nine of 11 SQS samples qualified with all 6 loci genotyped. Two samples (2017S_01 and 2017F_09A) qualified with a 5-locus genotype. For these samples, locus GOAG4 was not evaluated by RELIOTYPE due to the presence of aberrant (previously unrecorded) alleles in at least one of the replicates. Of the 16 samples which were either field collected (with unknown identity) or for which field collection was simulated by exposure to field conditions (Table 2; samples identified with “E” in sample name), 5 (31%) were deemed SQS samples. Of the 9 fresh samples collected directly from animals being handled (and without exposure to field conditions), 6 (67%) were found to be SQS samples. SQS samples had average Q-scores that ranged from 0.64 to 1.

Table 2.

Quality of 25 Gopherus agassizii scat samples genotyped at 6 screening loci using 6 replicate PCR amplifications. Scat samples marked with E were exposed to 7–9 days of full sun. Quality metrics (averaged across loci) including Q-scores and reliability (per RELIOTYPE) are shown. Whether or not the sample met SQS (suitable quality sample) conditions, percent loci where a consensus genotype could be determined, and percent loci that matched paired blood genotypes are included. NA = not applicable for scat samples with no matching blood samples. Eleven samples had average Q-scores above 0.6 and met SQS standards without further replication.

Quality of 25 Gopherus agassizii scat samples genotyped at 6 screening loci using 6 replicate PCR amplifications. Scat samples marked with E were exposed to 7–9 days of full sun. Quality metrics (averaged across loci) including Q-scores and reliability (per RELIOTYPE) are shown. Whether or not the sample met SQS (suitable quality sample) conditions, percent loci where a consensus genotype could be determined, and percent loci that matched paired blood genotypes are included. NA = not applicable for scat samples with no matching blood samples. Eleven samples had average Q-scores above 0.6 and met SQS standards without further replication.
Quality of 25 Gopherus agassizii scat samples genotyped at 6 screening loci using 6 replicate PCR amplifications. Scat samples marked with E were exposed to 7–9 days of full sun. Quality metrics (averaged across loci) including Q-scores and reliability (per RELIOTYPE) are shown. Whether or not the sample met SQS (suitable quality sample) conditions, percent loci where a consensus genotype could be determined, and percent loci that matched paired blood genotypes are included. NA = not applicable for scat samples with no matching blood samples. Eleven samples had average Q-scores above 0.6 and met SQS standards without further replication.

All SQS samples with average Q-scores ≥ 0.6 matched their blood genotypes at all loci, suggesting that samples that meet these quality criteria should have accurate genotypes (Table 2). In 3 samples that failed to meet the SQS and Q-score criteria above, we found 6 mismatches between the sample consensus genotype and the blood genotypes (Table 3). Five of these errors were identified as false alleles and one was a case of large allelic dropout (where the larger of two fails to amplify in a heterozygous individual).

Table 3.

Mismatches between Gopherus agassizii scat and blood-derived genotypes (presented as allele sizes) and inferred error type. LADO = large allele dropout; FA = false allele.

Mismatches between Gopherus agassizii scat and blood-derived genotypes (presented as allele sizes) and inferred error type. LADO = large allele dropout; FA = false allele.
Mismatches between Gopherus agassizii scat and blood-derived genotypes (presented as allele sizes) and inferred error type. LADO = large allele dropout; FA = false allele.

Evaluation of Scat Extraction Methods. — Thirtyfour scat extractions were performed using the stool kit (with 5 variations) and 16 with the plant kit. All extraction methods had overlapping distributions in average Qscores, with protocol 1 performing slightly worse than the other protocols (Fig. 2). In addition, we found no significant difference in Q-scores between paired stool kit and plant kit extractions (W15 = 58.5; p ≤ 0.3815). The cost and time investment for all stool kit protocols is higher than that of the plant kit protocol. The plant kit has fewer pipetting steps (limiting potential for contamination or operator error), consumes just a small portion of the scat sample (so multiple extractions could be performed to increase DNA yield), and extractions may also be used in the future for G. agassizii diet analyses.

Figure 2.

Box plots of the distributions of average Q-scores for the 6 scat extraction protocols tested on G. agassizii scat samples. Lines within boxes are placed at the median and the boxes extend between the 25% and 75% quartiles. Whiskers extend to the lowest and highest values within 1.5* interquartile range (IQR).

Figure 2.

Box plots of the distributions of average Q-scores for the 6 scat extraction protocols tested on G. agassizii scat samples. Lines within boxes are placed at the median and the boxes extend between the 25% and 75% quartiles. Whiskers extend to the lowest and highest values within 1.5* interquartile range (IQR).

Locus Suitability for Genotyping G. agassizii Scat Extractions. — For blood extractions genotyped multiple times, we encountered no PCR failures or erroneous genotypes at any of 19 loci (i.e., Q-score = 1.0). Of 8 scat extractions, each with 5 replicate PCR reactions, RELIOTYPE accepted the multilocus genotype of 4 extractions without additional PCR replicates (Supplemental Table S1; supplemental material is available at https://doi.org/10.2744/CCB-1394.1.s1). One of these samples qualified with a 14-locus genotype, 2 samples qualified with 18-locus genotypes, and 1 sample qualified with a 19-locus genotype. Error rates ranged from 0% to 47.4% for ADO and 0% to 5.9% for FA. Average Q-scores ranged from 0.23 for Locus GP81 to 1.0 for Locus GOA1 (Fig. 3; Supplemental Table S1). The 3 loci with the lowest average Q-scores (< 0.54), also had either exceptionally low successful allele call rates or low genotype accuracy rates. The cumulative probabilities of identity (PIDcum and PSIBcum) for the remaining 16 loci were 5.8 × 10-22 and 9.7 × 10-8, respectively.

Figure 3.

Box plots of Q-scores for 19 loci genotyped for 8 SQS (suitable quality samples) extracted with the Qiagen DNeasy Plant Mini kit. Five replicate PCR amplifications were performed for each G. agassizii scat sample and locus. Lines within boxes are placed at the median and the boxes extend between the 25% and 75% quartiles. Whiskers extend to the lowest and highest values within 1.5* IQR.

Figure 3.

Box plots of Q-scores for 19 loci genotyped for 8 SQS (suitable quality samples) extracted with the Qiagen DNeasy Plant Mini kit. Five replicate PCR amplifications were performed for each G. agassizii scat sample and locus. Lines within boxes are placed at the median and the boxes extend between the 25% and 75% quartiles. Whiskers extend to the lowest and highest values within 1.5* IQR.

Identifying Opportunistically Collected Scat. — Three of the 7 unknown scat samples that met SQS standards were matched unambiguously at all 6 screening loci to previously captured tortoises. All three scats were found within 400 m of where the matching animal was captured. Additionally, we found 2 sets of samples with fuzzy matches (mismatched at 1 locus). One fuzzy match occurred between 2 blood samples taken approximately 8 km apart in different plots in the Ivanpah Valley data set, and so likely indicates these 2 individuals are close relatives. The other fuzzy match occurred between a blood and scat sample located within the same sampling site within 500 m. The 3 remaining high quality unknown scat samples did not match any individuals in our blood samples and likely represented previously unsampled individuals.

DISCUSSION

In this pilot study, we evaluated the feasibility of using G. agassizii scat to complement current methods used to monitor this species. First, we found that a reliable molecular tag can be obtained from DNA extracted from G. agassizii scat. Notably, scat samples that met SQS criteria and had average Q-scores > 0.6 also exactly matched blood samples from the same individuals, suggesting that these 2 quality criteria are appropriate for retaining accurate tortoise genotypes from scat. Second, of the 6 extraction protocols evaluated in the study, we detected no differences in extraction quality, but we found the Qiagen DNeasy Plant Mini kit to be the most cost- and time-efficient method. Finally, we found that 16 of 19 loci tested could be genotyped from scat with relatively high confidence for population genetic analyses while limiting genotyping error.

Reliability of G. agassizii Scat Sample Genotypes. — As many as 7 amplifications per locus have been recommended to obtain reliable genotypes in noninvasive studies (Taberlet et al. 1999). Statistical approaches such as those employed by RELIOTYPE are designed to reduce replication number, without compromising accuracy, by taking into account data already available (i.e., allele frequencies and data from a limited number of initial replicates), inferring the reliability of the consensus genotype derived from those data and suggesting an appropriate number of additional replicates to arrive at a predetermined confidence level for the multilocus genotype. In our study, we found that the default settings of RELIOTYPE were stringent enough to accept without further replicates only those scat genotypes that matched their respective blood genotypes. We also found that approximately 31% and 67% of scat samples collected from the ground and directly from animals, respectively, yielded DNA of sufficient quality to obtain reliable genotypes. A multilocus genotype of at least 5 loci will yield a molecular tag with PSIB below the recommended threshold of 0.01 (Waits et al. 2001); therefore, these multilocus genotypes can be used as molecular tags for purposes of mark–recapture studies or for population size estimates. As proof of concept, we were able to match several unknown scat samples to genotyped animals in the same plots with low PID and PSIB.

Evaluation of Extraction Methods. — We observed high variability in genotype quality among samples regardless of extraction protocols. Sample quality is likely the most critical factor in determining whether a reliable genotype can be obtained from G. agassizii scat. Edwards et al. (submitted) evaluated a different scat extraction protocol on G. agassizii and Gopherus morafkai scat; they used epithelial cells swabbed from the surface of scats and reported similar success rates to our study (avg. 35% across field collected tortoise scats). Although we detected no difference in sample quality related to extraction protocol, the plant extraction protocol has advantages in terms of cost, time, and yield efficiencies. Pearson et al. (2015) found that the Qiagen DNeasy Plant Mini kit performed well for individual genotyping from field-collected lizard scat when alternative extraction protocols optimized on captive lizard scat failed on wild samples. The authors attributed the failure to the diet of wild lizards, which was richer in vegetation than that of captive lizards. In our study, we did not find a significant difference between genotype quality for the subset of samples that were extracted with both the stool and plant kits. This may be because all stool kit extractions were performed on the surface washes, avoiding most of the plant material inside the scat (unlike alternative protocols in Pearson et al. 2015). We also note that, while pelleted herbivore scat lends itself well to surface washes (Flagstad et al. 1999; Wedrowicz et al. 2013), this was not the case with tortoise scat because 1) it is not pelleted, 2) size is variable, and 3) tortoise scat segments are larger and less compacted. At times, scat segments partially crumbled during the wash and absorbed significant amounts of buffer during the wash procedure (up to 4 ml of Inhibitex Buffer per sample). Overall, the plant kit protocol was less expensive and less labor-intensive, took less time, and had fewer steps (potentially reducing contamination risk). Finally, plant kit extractions could also be used to amplify and analyze the tortoise's herbivorous diet, the quality of which has been shown in controlled experiments to be directly linked tortoise health (Drake et al. 2016).

Locus Suitability for Genotyping G. agassizii Scat Extractions. — There is a trade-off in the amount and accuracy of information that can be obtained from noninvasively collected samples (Waits and Leberg 2000). For purposes of molecular tagging for estimates of census population size using capture–recapture methods, if the number of loci is too low some individuals will have the same molecular tag, resulting in an underestimate of population size. Conversely, introducing more loci increases the probability for error, which in turn creates false individuals resulting in population size overestimation. Thus, highly polymorphic loci with low genotyping error rates are recommended for studies using molecular tags to obtain estimates of population size (Waits and Leberg 2000). We found the initial 6 screening loci to be a reliable set of loci for molecular tagging, yielding a molecular tag with low probabilities of misidentifying 2 random individuals (PID) or 2 siblings (PSIB) as the same individual.

In contrast to estimating census population size and paternity, genotyping error at low frequencies may not be as problematic in the estimation of population genetic diversity, differentiation, and effective population size, and the inclusion of more loci can increase precision in these estimates. In simulations, Smith and Wang (2014) found that, with genotyping error rates below 0.2, reasonable estimates of genetic variation and population subdivision could be obtained. Based on our tests of 19 loci for G. agassizii, we found that 16 loci could be genotyped with relatively high confidence.

Conclusions. — While sample sizes in this pilot study are small, the results are encouraging enough to warrant pursuing scat genotyping as a viable source of data for G. agassizii population monitoring. In particular, detection of 3 “recaptures” of previously genotyped animals indicates that scat genotyping could be used for recapture in regions with a good set of reference DNA samples. Finally, the developed scat extraction and genotyping protocol can be extended across the species range and possibly to other tortoise species for which these microsatellite loci cross-amplify. This reliable, scatbased genotyping method can complement ongoing efforts characterizing desert tortoise populations throughout the Mojave Desert.

Acknowledgments

We thank K. Drake, F. Chen, B. Gottsacker, and Ironwood Consulting for sample collection efforts. Research was supported by the US Bureau of Land Management (BLM), the USFWS, the National Fish and Wildlife Foundation (NFWF), and the US Geological Survey (USGS), Ecosystems Mission Area Energy and Wildlife Program. We are grateful to A. Fesnock (BLM) and M. Slaughter (BLM) for their support of our desert tortoise research program. All tortoises were handled in accordance with USGS, Western Ecological Research Center, and UC Davis IACUC, USFWS Permit TE030659-11, Nevada Department of Wildlife Scientific Collection Permit 317351, and a Memorandum of Understanding with the California Department of Fish and Wildlife held by T.C.E. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government. Data are available as a USGS Data Release (https://doi.org/10.5066/P9SNWMJY).

LITERATURE CITED

LITERATURE CITED
Anderson,
D.R.,
Burnham,
K.P.,
Lubow,
B.C.,
Thomas,
L.,
Corn,
P.S.,
Medica,
P.A.,
Marlow,
R.W.
2001
.
Field trials of line transect methods applied to estimation of desert tortoise abundance.
Journal of Wildlife Management
65
:
583
597
.
Arandjelovic,
M.
Vigilant,
L.
2018
.
Non-invasive genetic censusing and monitoring of primate populations.
American Journal of Primatology
80
:
e22743
.
Beja-Pereira,
A.,
Oliveira,
R.,
Alves,
P.C.,
Schwartz,
M.K.,
Luikart,
G.
2009
.
Advancing ecological understandings through technological transformations in noninvasive genetics.
Molecular Ecology Resources
9
:
1279
1301
.
Bohonak,
A.J.
Mitelberg,
A.
2014
.
Final report: Social structure and genetic connectivity in the southern mule deer: implications for management.
Prepared for California Department of Fish and Wildlife
.
Bonin,
A.,
Bellemain,
E.,
Bronken Eidesen,
P.,
Pompanon,
F.,
Brochmann,
C.,
Taberlet,
P.
2004
.
How to track and assess genotyping errors in population genetics studies: tracking and assessing genotyping errors.
Molecular Ecology
13
:
3261
3273
.
Brinkman,
T.J.,
Schwartz,
M.K.,
Person,
D.K.,
Pilgrim,
K.L.,
Hundertmark,
K.J.
2010
.
Effects of time and rainfall on PCR success using DNA extracted from deer fecal pellets.
Conservation Genetics
11
:
1547
1552
.
Broquet,
T.,
Ménard,
N.,
Petit,
E.
2006
.
Noninvasive population genetics: a review of sample source, diet, fragment length and microsatellite motif effects on amplification success and genotyping error rates.
Conservation Genetics
8
:
249
260
.
Broquet,
T.
Petit,
E.
2004
.
Quantifying genotyping errors in noninvasive population genetics.
Molecular Ecology
13
:
3601
3608
.
Drake,
K.K.,
Bowen,
L.,
Nussear,
K.E.,
Esque,
T.C.,
Berger,
N.A.,
Custer,
A.,
Waters,
S.C.,
Johnson,
J.D.,
Miles,
A.K.,
Lewison,
R.L.
2016
.
Negative impacts of invasive plants on conservation of sensitive desert wildlife.
Ecosphere
7
:
e01531
.
Dutcher,
K.E.,
Vandergast,
A.G.,
Esque,
T.C.,
Mitelberg,
A.,
Matocq,
M.D.,
Heaton,
J.S.,
Nussear,
K.E.
Genes in space: what Mojave Desert tortoise genetics can tell us about landscape connectivity. Conservation Genetics (in press).
Edwards,
T.,
Berry,
K.H.,
Nussear,
K.E.,
Grandmaison,
D.D.,
Culver,
M.,
Jones,
C.A.,
Inman,
R.D.,
Esque,
T.C.
2015
.
Testing taxon tenacity of tortoises: evidence for a geographical-selection gradient at a secondary contact zone.
Ecology and Evolution
5
:
2095
2114
.
Edwards,
T.,
Clemons,
J.,
Manning,
J.A.,
Leavitt,
D.J.,
Culver,
M.
Efficacy of DNA sampling from herbivorous reptile scat. Submitted for publication.
Edwards,
T.,
Goldberg,
C.S.,
Kaplan,
M.E.,
Schwalbe,
C.R.,
Swann,
D.E.
2003
.
PCR primers for microsatellite loci in the desert tortoise (Gopherus agassizii, Testudinidae).
Molecular Ecology Notes
3
:
589
591
.
Flagstad,
O.,
Røed,
K.,
Stacy,
J.E.,
Jakobsen,
K.S.
1999
.
Reliable noninvasive genotyping based on excremental PCR of nuclear DNA purified with a magnetic bead protocol.
Molecular Ecology
8
:
879
883
.
Frantz,
A.C.,
Pope,
L.C.,
Carpenter,
P.J.,
Roper,
T.J.,
Wilson,
G.J.,
Delahay,
R.J.,
Burke,
T.
2003
.
Reliable microsatellite genotyping of the Eurasian badger (Meles meles) using faecal DNA.
Molecular Ecology
12
:
1649
1661
.
Frantzen,
M.A.J.,
Silk,
J.B.,
Ferguson,
J.W.H.,
Wayne,
R.K.,
Kohn,
M.H.
1998
.
Empirical evaluation of preservation methods for faecal DNA.
Molecular Ecology
7
:
1423
1428
.
Freilich,
J.E.,
Burnham,
K.P.,
Collins,
C.M.,
Garry,
C.A.
2000
.
Factors affecting population assessments of desert tortoises.
Conservation Biology
14
:
1479
1489
.
Gagneux,
P.,
Boesch,
C.,
Woodruff,
D.S.
1997
.
Microsatellite scoring errors associated with noninvasive genotyping based on nuclear DNA amplified from shed hair.
Molecular Ecology
6
:
861
868
.
Giambattista,
L.D.
Gentile,
G.
2018
.
A molecular protocol to distinguish syntopic Galápagos land iguanas (Conolophus marthae and C. subcristatus) from faecal samples.
Herpetology Notes
11
:
97
100
.
Hagerty,
B.E.,
Peacock,
M.M.,
Kirchoff,
V.S.,
Tracy,
C.R.
2008
.
Polymorphic microsatellite markers for the Mojave Desert tortoise, Gopherus agassizii.
Molecular Ecology Resources
8
:
1149
1151
.
Kalinowski,
S.T.,
Taper,
M.L.,
Marshall,
T.C.
2007
.
Revising how the computer program Cervus accommodates genotyping error increases success in paternity assignment: Cervus likelihood model.
Molecular Ecology
16
:
1099
1106
.
Luikart,
G.,
Ryman,
N.,
Tallmon,
D.A.,
Schwartz,
M.K.,
Allendorf,
F.W.
2010
.
Estimation of census and effective population sizes: the increasing usefulness of DNA-based approaches.
Conservation Genetics
11
:
355
373
.
Miller,
C.R.,
Joyce,
P.,
Waits,
L.P.
2002
.
Assessing allelic dropout and genotype reliability using maximum likelihood.
Genetics
160
:
357
366
.
Miquel,
C.,
Bellemain,
E.,
Poillot,
C.,
Bessière,
J.,
Durand,
A.,
Taberlet,
P.
2006
.
Quality indexes to assess the reliability of genotypes in studies using noninvasive sampling and multiple-tube approach.
Molecular Ecology Notes
6
:
985
988
.
Mitelberg,
A.
Vandergast,
A.G.
2016
.
Non-invasive genetic sampling of southern mule deer (Odocoileus hemionus fuliginatus) reveals limited movement across California State Route 67 in San Diego County.
Western Wildlife
3
:
8
18
.
Murphy,
M.A.,
Kendall,
K.C.,
Robinson,
A.,
Waits,
L.P.
2007
.
The impact of time and field conditions on brown bear (Ursus arctos) faecal DNA amplification.
Conservation Genetics
8
:
1219
1224
.
Nussear,
K.E.,
Esque,
T.C.,
Heaton,
J.S.,
Cablk,
M.E.,
Drake,
K.K.,
Valentin,
C.,
Yee,
J.L.,
Medica,
P.A.
2008
.
Are wildlife detector dogs or people better at finding desert tortoises (Gopherus agassizii)?
Herpetological Conservation and Biology
3
:
103
115
.
Nussear,
K.E.
Tracy,
C.R.
2007
.
Can modeling improve estimation of desert tortoise population densities?
Ecological Applications
17
:
579
586
.
Panasci,
M.,
Ballard,
W.B.,
Breck,
S.,
Rodriguez,
D.,
Densmore,
L.D.,
III,
Wester,
D.B.,
Baker,
R.J.
2011
.
Evaluation of fecal DNA preservation techniques and effects of sample age and diet on genotyping success.
Journal of Wildlife Management
75
:
1616
1624
.
Pearson,
S.K.,
Tobe,
S.S.,
Fusco,
D.A.,
Bull,
C.M.,
Gardner,
M.G.
2015
.
Piles of scats for piles of DNA: deriving DNA of lizards from their faeces.
Australian Journal of Zoology
62
:
507
514
.
Piggott,
M.P.,
Wilson,
R.,
Banks,
S.C.,
Marks,
C.A.,
Gigliotti,
F.,
Taylor,
A.C.
2008
.
Evaluating exotic predator control programs using non-invasive genetic tagging.
Wildlife Research
35
:
617
.
Pompanon,
F.,
Bonin,
A.,
Bellemain,
E.,
Taberlet,
P.
2005
.
Genotyping errors: causes, consequences and solutions.
Nature Reviews Genetics
6
:
847
859
.
Schrader,
C.,
Schielke,
A.,
Ellerbroek,
L.,
Johne,
R.
2012
.
PCR inhibitors—occurrence, properties and removal.
Journal of Applied Microbiology
113
:
1014
1026
.
Schwartz,
M.,
Luikart,
G.,
Waples,
R.
2007
.
Genetic monitoring as a promising tool for conservation and management.
Trends in Ecology & Evolution
22
:
25
33
.
Schwartz,
T.S.,
Osentoski,
M.,
Lamb,
T.,
Karl,
S.A.
2003
.
Microsatellite loci for the North American tortoises (genus Gopherus) and their applicability to other turtle species.
Molecular Ecology Notes
3
:
283
286
.
Smith,
O.
Wang,
J.
2014
.
When can noninvasive samples provide sufficient information in conservation genetics studies?
Molecular Ecology Resources
14
:
1011
1023
.
Taberlet,
P.,
Waits,
L.P.,
Luikart,
G.
1999
.
Noninvasive genetic sampling: look before you leap.
Trends in Ecology & Evolution
14
:
323
327
.
Us Fish and Wildlife Service (USFWS).
1990
.
Endangered and threatened wildlife and plants; determination of threatened status for the Mojave population of the Desert Tortoise.
Federal Register
55
:
12178
12191
.
Us Fish and Wildlife Service (USFWS).
2011
.
Revised recovery plan for the Mojave population of the desert tortoise (Gopherus agassizii).
US Fish and Wildlife Service, Pacific Southwest Region
,
Sacramento, CA
.
Waits,
J.L.
Leberg,
P.L.
2000
.
Biases associated with population estimation using molecular tagging.
Animal Conservation
3
:
191
199
.
Waits,
L.P.,
Luikart,
G.,
Taberlet,
P.
2001
.
Estimating the probability of identity among genotypes in natural populations: cautions and guidelines.
Molecular Ecology
10
:
249
256
.
Waits,
L.P.
Paetkau,
D.
2005
.
Noninvasive genetic sampling tools for wildlife biologists: a review of applications and recommendations for accurate data collection.
Journal of Wildlife Management
69
:
1419
1433
.
Wedrowicz,
F.,
Karsa,
M.,
Mosse,
J.,
Hogan,
F.E.
2013
.
Reliable genotyping of the koala (Phascolarctos cinereus) using DNA isolated from a single faecal pellet.
Molecular Ecology Resources
13
:
634
641
.

Author notes

Handling Editor: Jeffrey E. Lovich