Standardization and validation of a panel of cross-species microsatellites to individually identify the Asiatic wild dog (Cuon alpinus)

Introduction

The Asiatic wild dog or dhole (Cuon alpinus) is a highly elusive, endangered, social canid distributed in south and southeast Asia (Johnsingh, 1982; Durbin et al., 2004) occupying a range of habitat types including alpine, temperate, sub-tropical and tropical forests (Durbin et al., 2004). Driven by habitat loss, prey depletion, disease transmission from domestic dog, human persecution and interspecific competition (Hayward, Lyngdoh & Habib, 2014; Kamler et al., 2015), dholes are currently found in about 75% of their historical global range (Durbin et al., 2004; Kamler et al., 2015). Global dhole population is roughly estimated to be about 4,500–10,500 with only 949–2,215 mature individuals, but accurate estimates and population trends are not available from any part of its range (Kamler et al., 2015). They are considered as ‘Endangered’ by IUCN under criteria C2a(i) and included in Appendix II of the Convention on International Trade in Endangered Species. The Indian subcontinent currently retains majority of the remaining dhole populations (Kamler et al., 2015) in the Western Ghats and central Indian forests (Karanth et al., 2009), along with smaller populations in the Eastern Ghats (Karanth et al., 2009), northeast India (Gopi, Lyngdoh & Selvan, 2010; Lyngdoh et al., 2014) and Himalayan region (Bashir et al., 2014). The species has faced about 60% decline in their historical distribution in the subcontinent (Karanth et al., 2010).

Given the current anthropogenic disturbance scenario across its range, the future survival of this monotypic genus depends on integrated conservation measures involving detailed, accurate information on ecology, demography and genetics. However, generating reliable information for this elusive, forest-dwelling and pack-living canid at landscape scale is challenging. Traditional ecological techniques such as regular photographic capture approach are ineffective for dholes due to absence of unique coat patterns and their monomorphic forms, and physical tagging methods are impractical at landscape scales due to logistical difficulties, high costs and small numbers of captures possible. In this context, genetic tools have tremendous potential to generate critical information such as population size estimation (Mondol et al., 2009b), phylogeography (Luo et al., 2014; Waits et al., 1998), pack dynamics and reproductive fitness (Sillero-Zubiri, Gottelli & Macdonald, 1996; Girman et al., 1997), dispersal patterns (Epps et al., 2007; Gour et al., 2013) for elusive species conservation across large landscapes (Mondol, Bruford & Ramakrishnan, 2013). The ability to identify individuals from non-invasive samples collected over large space provides a feasible option to generate detailed information on elusive, forest-dwelling dholes as they cannot be identified using other approaches.

In this study, we addressed key methodological issues related to selection and standardisation of a set of molecular markers for individual identification of dholes. Subsequently, we tested these markers on field-collected dhole samples from five protected areas of the central Indian landscape in the state of Maharashtra, India for individual identification. In addition to the utilisation of these markers in dhole population estimation at landscape level, we believe that this approach has wider relevance in non-invasive, faecal DNA based population assessments of many other low density, elusive, wide-ranging species.

Methods

Field sampling

Dholes prefer dense forested habitats (Johnsingh, 1985) where the social groups defaecate in communal latrine sites mostly found on the junctions of roads/trails (Johnsingh, 1982). Their elusive nature and highly social behaviour present unique challenges in scat sampling for individual identification. In this study, sampling was conducted through intensive foot and vehicle surveys covering the entire study area. We sampled a total of 49 latrine sites covering five protected areas. Once a latrine site was found only fresh scats were targeted for collection. One bolus from each fresh scat was collected assuming it to derive from one individual. Separate gloves were used to collect each sample. All samples were collected directly in wax paper and stored in separate ziplock bags. Once brought to the field station, the sample containing ziplock bags were temporarily stored in a large box containing silica gel to minimise fungal growth in humid conditions. Samples were then shipped to the laboratory, where they were stored in a −20 °C freezer. GPS co-ordinates and other associated information (track marks, substrata etc.) were collected for each sample. Entire sampling was conducted once per site between January 2015 and June 2017 ensuring maximum coverage of the study area, covering PTR (257.3 km²), MTR (1,500.49 km²), NNTR (152.8 km²), TATR (627.5 km²) and UKWLS (189 km²), Maharashtra. A total of 249 samples were collected (PTR—92, MTR—76, NNTR—37, UKWLS—34, TATR—10, respectively) for this study. Details of all sample locations are given in Fig. 1.

For blood sampling, four animals (three males and one female) were remotely administered with reconstituted lyophilised mixture of Telatamine-Zolezepam (Zoletil 100; Virbac, Carros, France) at the dose rate of 8.6 mg/kg body weight using Dan-inject syringe projector (Model-IM) at a distance of 15–25 m. After ensuring sedation safe for handling, animal was approached, blindfolded and one ml of blood was collected through femoral-saphenous vein puncture. Blood was collected in EDTA vacutainers and preserved at −20 °C in the laboratory for genetic work.

Results

During initial standardisations we tested all 18 selected markers (see Table 1) with four wild-caught dhole blood DNA samples. Three of these tested markers (WD2159, CPH16, AHT136) did not show any amplification in the blood DNA samples and were removed from subsequent analyses. The remaining 15 markers were then amplified with 225 genetically confirmed dhole faecal samples. Following data validation through multiple repeats, amplification success rates and polymorphism for these loci were calculated. The results show that loci CXX608 and CXX140 were monomorphic in all amplified samples, and locus CPH6 has low amplification success rate (~35%) from faecal DNA and thus were removed from the panel. The remaining 12 markers were finally standardised as four multiplex panels (see Table 1) for dhole individual identification.

None of these final 12 loci showed any signatures of large-scale allelic dropouts. The mean allelic dropout rate was found to be 0.1, whereas mean FA frequency for all the 12 loci was 0.092. Overall frequency of NAs was calculated as 0.11, indicating this 12 loci panel has low genotyping error rates. Amplification success ranged between 57% and 88% from dhole faecal DNA. The loci showed relatively higher (WD2201-9 alleles, H_o = 0.63) to medium (CXX251-3 alleles, H_o = 0.23) levels of polymorphism (Table 1). Except locus WD2001, none of the other loci were found to deviate from the Hardy–Weinberg equilibrium and there were no evidences for strong linkage disequilibrium between any pair of loci. Summary statistics for various measures of polymorphism (H_o and expected heterozygosity, Na and allelic size range) for all loci in the final panel are presented in Table 1.

For individual identification, we only considered samples that produced good quality data for at least seven of the 12 panel loci. This cut-off value of average of seven loci was decided based on the statistical support (PID_sibs value of 1 in 500 siblings) produced by these loci. Given that any single largest dhole population is about 250–300 individuals (Kamler et al., 2015), this value is sufficient for individual identification at local scales in India. Out of the 225 field-collected dhole faecal samples amplified with the panel of 12 markers, 98 produced seven or more loci data. Overall, we generated genetic data from a total of 102 samples (four blood and 98 faecal samples) (Fig. 1). About 70% of these samples (n = 71) have successfully amplified for 10–12 loci. Following analyses with CERVUS, we identified 101 unique dhole individuals from the entire data, whereas one individual from NNTR was found to be a ‘genetic recapture’. Cumulative PID_sibs and PID_unbiased values were found to be 1.5 × 10⁻⁴ and 4.7 × 10⁻¹⁰, respectively, indicating a strong statistical support for unambiguous individual identification. The number of unique individuals from each sampled area was found to be: PTR—33, MTR—35, NNTR—16, UKLWS—9 and TATR—4 (Fig. 2). Molecular sexing showed a success rate of 67%, with a male:female sex ratio of 4:1 in all identified dhole individuals (n = 101).

Discussion

In this paper, we standardised protocols for individual identification of Asiatic wild dogs from poor quality DNA samples, and the final marker panel could unambiguously identify individual dholes in our field-based pilot study from five protected areas of Maharashtra, India. The systematic protocols followed here offer some advantages over earlier efforts on dhole individual identification from faecal samples by Iyengar et al. (2005). Firstly, use of a large panel of 37 microsatellite loci for preliminary assessment of marker suitability along with genomic mapping-based selection of final markers (n = 18) allowed us to ascertain a combination of loci for unambiguous individual identification with high statistical power. The rigorous testing of the loci with large number of DNA samples from different sources also allowed us to exclude loci that might be problematic due to low amplification success from non-invasive samples. The final panel consisting 12 markers were further standardised into four multiplex reactions to provide time and cost-effective options during data generation. We were very careful to initially select a large number of tetranucleotide markers as they are known to have low stutter peak problems and better allele characteristics from poor quality samples (Walsh, Fildes & Reynolds, 1996), while dinucleotide markers generally have higher amplification success (Broquet, Ménard & Petit, 2007). Thus, our final panel with a ratio of 2:1 tetra vs. dinucleotide microsatellites would provide the ideal combination in terms of high success rate and less technical issues in allele calling during dhole individual identification. The amplification success rate for all loci was >70% except locus PEZ5 (~60%), but it was found to be polymorphic and was included in the panel. The overall genotyping error frequency was found to be <0.2 from dhole faeces, which is within the recommended limits for non-invasive population genetic research (Smith & Wang, 2014).

Our motivation in this study was to develop effective protocols that could be applied for individual identification of Asiatic wild dogs as they are difficult to identify from physical characteristics (spots, marks, stripes etc.). Their elusive nature also makes it challenging to estimate population size using traditional techniques (photographic capture, field-based observations etc.) at landscape levels. For genetic estimation of population size Waits, Luikart & Taberlet (2001) recommended a threshold PID_sibs value that is at least double than the approximate number of animals in any given area. The cumulative PID_sibs value of 1.5 × 10⁻⁴ achieved in this study is better than Iyengar et al. (2005) (PID_sibs of 3.3 × 10⁻⁴) and should be sufficient to study dhole genetics and specifically population estimation across its range. Among all the dhole range countries India is considered to retain the highest (about 1,500–3,000) number of individuals (Kamler et al., 2015) and our misidentification rate achieved in this study (1 in 6,700 siblings) would provide strong statistical power in individual identification. The most recent assessment suggests that the largest dhole population in Western Ghats, India holds about 207–304 individuals (Kamler et al., 2015), thereby assuring that our seven loci cut-off (misidentification rate 1 in 500 siblings) to select samples and 12 loci panel would be useful in population estimation at local scale. However, it is important to point out that we generated individual level information from about 43.5% (98 out of 225 faeces) of the field-collected samples in this study. Similar patterns of low amplification success rate from field-collected faecal samples have been observed in earlier genetic study of dhole (Iyengar et al., 2005), leopard (Mondol et al., 2009a) and other species (Smith & Wang, 2014). Considering dhole cryptic nature, social behaviour and ecology in corroboration with low amplification success rate, we suggest an intensive faecal sampling effort for estimation of population size for this species. It is also noteworthy to point out that literature survey for dhole marker selection in this study was mostly based on available information on non-invasive canid population genetic research with specific information available on markers such as marker polymorphism (Na, H_o) and amplicon size etc. (Holmes et al., 1995; Ostrander, Sprague & Rine, 1993; Fredholm & Winterø, 1995; Ostrander et al., 1995; Francisco et al., 1996; Neff et al., 1999). However, future studies should also consider additional markers those are tested as part canid forensic studies (for example see Van Asch et al., 2009; Berger et al., 2014; Hellmann et al., 2006; Eichmann, Berger & Parson, 2004) on dholes. In addition, already available dhole genome information (Campana et al., 2016; Habib et al., 2018) also can be used to develop a suitable SNP panel for in depth analyses of dhole population and demographic parameters.

During individual identification we had identified only one genetic recapture from the field-collected faecal samples. This pattern of low dhole recapture could be attributed to our sampling strategy to cover large geographical area and maximise collection of faeces from potentially different individuals, as well as relatively low amplification success rates from faecal samples. We have sampled the entire study area only once and focused on collecting fresh samples, thereby probably missed recapturing the same individuals from same latrine sites. Further, low amplification success rate (101 genotypes from 225 fresh samples) might have resulted in getting lesser number of recaptures. We got a male biased sex-ratio (4:1) in this pilot study. While there is no conclusive information on dhole sex ratio across its range, our earlier study (Modi et al., 2018) in the same landscape has shown a male biased (M:F ratio of 3:1) sex ratio, and ecological study in southern India by Venkataraman (1998) suggested male biased packs. Future studies with extensive genetic sampling across this landscape would potentially provide more accurate sex ratio for dholes.

Conclusion

In the broader context of understanding dhole population dynamics at local or landscape scales, genetic sampling is possibly the only way to generate information with spatial and temporal coverage for this elusive, social carnivore as photographic sampling or conventional tagging cannot be employed due to lack of distinguishing natural marks and logistical difficulties of physical captures of large number of animals. Results from this study provide a robust tool to generate individual level information from field-collected faecal samples. In combination with a good sampling strategy, our methods can be used in a cost-effective way to investigate species biology (including patterns of genetic diversity, relatedness and population connectivity) as well as to estimate population abundance of dholes in the wild.

Supplemental Information