Genome-Wide Variant Analyses Reveals New Patterns of Admixture and Population Structure in Australian Dingoes

Canid sampling

We collected tissue, blood, or buccal samples from 307 wild and 84 captive dingoes from locations across Australia (Figure 1, Supplementary Information Table S1). A majority of the wild dingoes were sampled as part of programs to suppress dingo populations. Additional samples were provided by private landholders, government officers, found as roadkill or during carcass disposal, or sampled from puppies found in the wild. Unrelated captive dingo samples were contributed by dingo sanctuaries, zoos and private citizens. We included captive dingo samples to serve as controls in a comparison to wild dingoes, due to their known parentage. A subset of the wild and captive samples were from existing scientific collections and had been previously tested using a 23-marker microsatellite-based test developed for distinguishing dingoes from dingo x dog hybrids (Cairns et al., 2021a; Stephens et al., 2015). We also collected blood or buccal samples from 36 Australian domestic pet dogs and 111 North American domestic pet dogs. We chose dog samples from working, herding or mixed breeds as they represent the populations most likely to have hybridised with Australian dingoes. A set of five unrelated Vietnamese putative primitive breed samples were sourced from a private owner in Hồ Chí Minh City, Vietnam. DNA was extracted from blood, tissue or buccal samples using either a (1) Qiagen DNeasy Blood and Tissue (Qiagen Sciences, Germantown, USA) or (2) Monarch Genomic DNA Purification kits (New England Biolabs, Ipswich, USA or (3) a modified proteinase K, phenol-chloroform extraction method (Bell et al., 1981).

Axiom canine genotyping

After DNA quantification, the dingo, Australian dog and Vietnamese dog samples were genotyped on the Axiom Canine HD Genotyping array (Thermo Fisher Scientific Inc., Waltham, USA) at the Ramaciotti Centre for Genomics (University of New South Wales, Randwick, Australia). The 111 North American domestic pet dogs were genotyped on Axiom Canine Genotyping Array Set A and Set B (Thermo Fisher Scientific Inc., Waltham, USA) at the Thermo Scientific Microarray Research Services Laboratory (Santa Clara, CA, USA). We chose to use the >700,000 marker Axiom Canine HD Genotyping array (Thermo Fisher Scientific Inc., Waltham, USA) rather than the >170,000 marker Illumina CanineHD Whole-Genome Genotyping BeadChip (Illumina Inc., San Diego, USA) due to the increased number of informative markers (>130,000) contained in the Axiom Canine arrays. Genotypes were called using Axiom Analysis Suite (Thermo Fisher Scientific Inc., Waltham, USA) and quality control (QC) filtering was conducted using Plink v1.9 (Purcell et al., 2007). We trimmed the North American domestic dog data to include only the SNPs from the Axiom Canine HD Genotyping array (Thermo Fisher Scientific Inc., Waltham, USA), as the Axiom Canine Genotyping Set A and Set B arrays (Thermo Fisher Scientific Inc., Waltham, USA) contain probes for a larger set of SNP markers and were used in the development of the Axiom Canine HD Genotyping array (Thermo Fisher Scientific Inc., Waltham, USA). We excluded individuals or sites with ≥10% missing data using the commands --geno 0.1 and –missing. We did not perform minor allele frequency filtering because it has a negligible impact on ascertainment bias and may negatively impact genetic diversity or population structure assessments. Instead, we employed linkage-disequilibrium (LD) filtering (Malomane et al., 2018), by removing regions in high LD using the command --indep-pairwise 50 5 0.6 in Plink v1.9.

Dingo x dog ancestry and admixture analysis

We performed genetic ancestry analyses on the complete dataset of 307 wild dingoes, 84 captive dingoes and 152 domestic breed dogs using FastStructure (Raj et al., 2014). We employed FastStructure, a Bayesian population clustering method similar to STRUCTURE (Pritchard et al., 2000), due to its capacity to handle large genomic datasets and integrated method of choosing model complexity (K-values). Wild dingoes belonged to one of six populations according to geographic sampling location: Western Australia, Central Australia, Northern Australia, Eastern Australia, Southern Australia and the Big Desert (Figure 1). Captive-bred dingoes were identified as such for the analysis and domestic dogs were identified as either a specific breed or a mixed breed group based on owner reporting. The specific domestic dog breed groups were Vietnamese dog (VIET), Mixed breed (MIXED), Kelpie (KELPIE), Australian Cattle Dog (ACD), Australian Stumpy Tail Cattle Dog (ASTCD), Border Collie (BORD), Golden Retriever (GOLD), German Shepherd (GSD), Labrador (LAB), Patagonian sheepdog (PGOD), Tervuren Shepherd (TURV) and Belgian Shepherd (BELS). FastStructure analyses were ‘unsupervised’ meaning that we did not use a priori defined populations to inform the Bayesian modelling and admixture detection (Raj et al., 2014).

FastStructure analyses were run with the StructureThreader pipeline (Pina-Martins et al., 2017), modelling K=1–20 with a random seed and simple prior. A simple prior was suitable for our study because we are primarily interested in strong population structure between dingoes and domestic dogs rather than substructure between domestic dog breeds. FastStructure reports the optimal K values to maximise marginal likelihood and explain structure in the data (Raj et al., 2014). Ten replicates were completed for each modelling scenario that best fit the data (K=8–10), calculating the average proportion of dingo ancestry for each individual along with the standard error. Confidence intervals were calculated by multiplying standard error values by 1.96 (McFarlane et al. (2020). The proportion of estimated dingo ancestry for each individual with 95% confidence intervals was plotted in R using the ggplot package (Wickham, 2016). An ancestry component bar chart based on the average FastStructure q-values was created in R using the BITE package (Milanesi et al., 2017).

FastStructure ancestry component q-values were used to categorise each animal as a dingo, historical or recent dingo backcross, dingo x dog hybrid, recent or historical dog backcross or domestic dog (Table 1). Rather than using a strict threshold, we assigned animals to a specific category using the calculated 95% confidence intervals (McFarlane & Pemberton, 2019). We considered that the occurrence of different backcross classes in wild dingo populations may inform conservation and management policy directions, with some previous studies considering animals with >93% dingo ancestry as pure (Allen et al., 2017). Therefore, we categorised backcrosses as being either: (1) historical backcrosses (>93% parental ancestry) that were likely to be third-generation or higher, or (2) recent backcrosses (55–93% parental ancestry) which described animals that were likely to have had a dingo x dog hybrid ancestor in the last one to three generations (Dziech, 2021; Harmoinen et al., 2021; McFarlane & Pemberton, 2019). A map depicting the FastStructure ancestry assignment of each wild origin sample was created using the QGIS v3.12 (QGIS, 2020).

Principal coordinates analyses (PCoA) were carried out using the R package SambaR function ‘find_structure’. In the SambaR analyses, populations were defined according to the FastStructure results with ancestry categorised as in Table 1. However, Vietnamese dogs were allocated to their own category, given that dogs from South Asia and Dingoes may share ancestry (Cairns, 2021). F_ST values between animals categorized as dingoes, (historical and recent) dingo backcrosses, dingo x dog hybrids, (historical and recent) dog backcrosses, domestic dogs and Vietnamese dogs calculated in SambaR using the ‘calcdistance’ function.

Dingo population structure and biogeography

To explore patterns of biogeographic variation in dingoes across Australia, the SambaR ‘find_structure’ function was used to run PCoA using only animals with an ancestry category of dingo or historical dingo backcross, excluding all animals with greater than 7% ancestry deriving from domestic dogs. Analyses were run with and without dingoes of captive origin. A map of wild origin dingo samples and their primary FastStructure population cluster ancestry identity was plotted using QGIS v3.12 (QGIS, 2020).

Inbreeding coefficients (F_het) were calculated in Plink v 1.9 using the command --het based on the equation F_het = (O − E) / (N − E) where O is the number of observed homozygous markers of the individual, E is the expected number of homozygous markers under the Hardy-Weinberg equilibrium calculated from the allele frequencies estimated on the dingo dataset and N is the total number of markers (Gazal et al., 2014; Purcell et al., 2007). F_het is an individual-based inbreeding estimator allowing us to compare differences in homozygous excess resulting from inbreeding between the identified FastStructure dingo populations, historical dingo backcrosses, recent dingo backcrosses and dingo x dog hybrids. All samples identified as domestic dogs, Vietnamese dogs or (recent or historical) dog backcrosses were excluded from this analysis. Animals were assigned to subpopulations based on the primary FastStructure dingo population cluster and boxplots visualising the range of computed F_het values in the different dingo sub-populations were created in R using the ggplot package (Wickham, 2016).

Comparisons between microsatellite and SNP ancestry estimates

We compared the modelled ancestry q-values for a set of 113 animals that were tested using both the 195,434-marker Axiom SNP array method and the 23-marker microsatellite method of Cairns et al. (2021a) and Stephens et al. (2015). We calculated the difference between our modelled SNP ancestry q-values and previously reported values based on microsatellite DNA testing (Cairns et al., 2021a; Stephens et al., 2015). We then calculated the average difference between SNP and microsatellite q-values for individuals within the five dingo subpopulations. We also assessed the number of individuals which were categorised by microsatellite DNA testing as recent backcrosses or dingo x dog hybrids, but were pure or historical backcrosses based on SNP analysis.

Axiom canine genotyping

Genotyping of the 543 canid samples on the Axiom Canine Genotyping arrays (Thermo Fisher Scientific Inc., Waltham, USA) resulted in raw genotypes at 627,324 variable loci (Cairns et al., 2022). After QC and LD filtering using Plink v1.9 a dataset of 195,474 SNPs was available for dingo x dog admixture and biogeographic clustering analysis in FastStructure and SambaR (Cairns et al., 2022).

Dingo x dog ancestry and admixture analysis

FastStructure ancestry modelling identified K = 8 as the best fit for the data whilst K = 10 maximized model complexity. We observed the presence of five biogeographic dingo population clusters which we termed: Big Desert, West, East, South and Captive (Figure 2). Dingoes from central and northern Australia carried a mixture of West and East population ancestry and did not form distinct population clusters. Some dingoes from eastern Australia carried a mixture of East and West population ancestry. Domestic dog breeds did not cluster discretely from one another, however three main clusters were observed: (1) Border Collie, (2) Belgian Shepherd/Turvian Shepherd and (3) all other dogs. The Vietnamese primitive breed dogs did not form their own population cluster but showed ancestry from both domestic dogs and dingoes.

Each sample was assigned using the Table 1 SNP testing criteria to an ancestry category after assessing calculated 95% confidence interval dingo q-values (Figure 3). In circumstances where the 95% confidence interval spanned a category threshold, we used the lower bound of the 95% confidence interval to classify individuals. This allowed us to conservatively assign animals and minimise the number of Type II errors where a backcross animal is assigned as a parental type. Within our dataset of 307 putative dingoes from across Australia 70.0% were pure dingoes; 15.0% were historical dingo backcrosses and 15.0% were recent dingo backcrosses (Table 2). No wild dingo x dog hybrids or feral domestic dogs were identified. All seven of the dingo x dog hybrids in our dataset were born and kept in captivity. Recent dingo backcrosses were more common in QLD (26.5%) and NSW (20.4%), whilst there were few recent dingo backcrosses observed in VIC (6.5%). Indeed, the four recent dingo backcrosses from VIC all carried >85% dingo ancestry. There was minimal evidence of dog ancestry in the dingoes sampled from SA or WA with only a single recent dingo backcross observed. From the NT there were two recent dingo backcrosses collected near towns, and the rest (n=14) were pure dingoes or historical dingo backcrosses.

PcoA provided additional evidence of both differentiation between dingoes and domestic dogs, while highlighting the limited amount of dog ancestry in dingoes (Figure 4). Admixed dingoes clustered closer to pure dingoes than to domestic dogs. Dingo x dog hybrids were distributed roughly equidistant between the pure dingoes and domestic dog clusters. Interestingly, the Vietnamese dogs behaved similarly to dingo x dog hybrids, corroborating the FastStructure analysis (Figure 2 and 3), revealing both dingo and dog ancestry. K=10 modelling suggests there may be further regional substructure within dingo populations, particularly the East population cluster, as we observed five animals from the NSW Southern Highlands near Sydney forming their own cluster.

F_ST values indicate that dingoes are distinct from domestic dogs, F_ST=0.288, domestic dog backcrosses, F_ST=0.293–0.324, and Vietnamese dogs, F_ST=0.301 (Table 3). There was a moderate level of differentiation between dingoes and dingo x dog hybrids with F_ST=0.126. Dingo backcrosses while not distinct from pure dingoes (F_ST=0.008–0.028) were distinct from both domestic dogs (F_ST=0.157–0.259) and Vietnamese dogs (F_ST>0.191). Dingo backcrosses were only mildly differentiated (F_ST=0.039–0.088) from dingo x dog hybrids (Table 3). Calculated Inbreeding Coefficient (F_het) values indicate that dingoes, both pure and backcrossed, had elevated homozygosity levels compared to dingo x dog hybrids (Figure 7). Captive dingoes had similar F_het values to wild dingo populations.

Dingo population structure and biogeography

Mapping of the FastStructure modelled primary population cluster identity for the 307 wild and 84 captive dingoes depicts a pattern of subdivision between the five population clusters: West, South, East, Big Desert and Captive (Figure 5). Dingoes in northern and central Australia carry a mixture of East and West population ancestry with the proportion of West dingo ancestry increasing north along the QLD coast or inland away from the east coast. Dingoes that originate from captive breeding largely form their own population, but there were three instances of putatively wild dingoes carrying the captive population cluster ancestry.

The population substructure observed in FastStructure is corroborated by the PcoA with four wild dingo populations: Big Desert, South, East and West (Figure 6). The Captive dingo population forms its own cluster. The large variation within population clusters (Figure 6) may corroborate the observation of additional substructure in the K=10 FastStructure modelling (Figure 2).

Calculated F_ST values indicate that there has been some geneflow between the West, East and South dingo populations (Table 4). Indeed, dingoes in northern and central Australia carry mixed East and West ancestry whilst some South dingoes carry some West dingo ancestry (Figure 2). The Big Desert population is the most distinct dingo population with limited evidence of gene flow between it and any of the other dingo populations. The Captive population cluster also had elevated F_ST values suggesting limited geneflow between captive and wild dingo populations.

Comparison of median F_het values between the dingo population clusters indicates that all dingoes have higher homozygosity levels than recent dingo backcrosses and dingo x dog hybrids (Figure 7), which is consistent with whole genome sequencing data that also reported higher levels of homozygosity in dingoes compared to domestic dogs (Kumar et al., 2023; Zhang et al., 2020). The Big Desert population has the highest median F_het values, nearly 30% higher than the other dingo populations, indicating the population is extremely homozygous and likely inbred. We did not compare homozygosity between dingoes and domestic dogs due to potential ascertainment bias associated with the design of the Axiom Canine HD Genotyping array (Thermo Fisher Scientific Inc., Waltham, USA).

Comparisons between microsatellite and SNP ancestry estimates

Nuclear SNP-based approaches are now the preferred method for ancestry assessment of mammals throughout the tree of life (McFarlane et al., 2020; Steyer et al., 2018; Stroupe et al., 2022; Szatmári et al., 2021; Väli et al., 2010; Zimmerman et al., 2020). However, as microsatellites have historically been the method of choice to date, we conducted a direct comparison between SNP and microsatellite-calculated dingo ancestry, observing that the proportion of dingo ancestry and occurrence of dingo x dog hybridisation across Australia differs depending on the method used (Figure 9, Tables 2 and 5). The microsatellite DNA testing method severely overestimates the presence of dog ancestry in dingoes compared to the SNP DNA testing method (Figure 8). For animals from the Captive, Big Desert, East or South population clusters, microsatellite analysis results in the assignment of animals to the incorrect category 28–44% of the time (Table 5). Whilst microsatellite DNA testing demonstrates widespread dog admixture in southeastern Australia, SNP data shows limited domestic dog ancestry in the wild dingo population.

Biogeography and regional variation in the dingo

We observed the presence of four distinct geographically distributed dingo populations: West, East, South, Big Desert (Figure 2, 5 and 6). Captive dingoes formed their own population, distinct from wild dingo populations (Figures 1 and 5). The Big Desert dingo population was the most strongly differentiated, with F_ST values twice that of other populations (Table 4). Whilst there are several phenotypically described varieties of dingo, it is unclear if these align with the genetic populations (Corbett, 2001; Walters, 1995). Our biogeographic findings corroborate and extend the observation of geographic subdivision in dingoes from skull morphometrics, mitochondrial DNA, Y chromosome and nuclear DNA data (Cairns et al., 2017; Cairns et al., 2018; Cairns & Wilton, 2016; Greig et al., 2018; Koungoulos, 2020; Zhang et al., 2020). We hypothesise that there is additional regional substructure within the four geographic dingo populations, with preliminary evidence of a regionally distinct sub-type of the East dingo population around the NSW Southern Highlands, southwest of Sydney (Figure 2).

Geneflow between the distinct wild dingo populations may reflect differing dispersal and demographic patterns across Australia. Our analyses observed geneflow between the West and East dingo populations in central and northern Australia and in southern Australia between the South and East dingo populations (Figure 2, Table 4). However, there was limited evidence of geneflow into or out of the Big Desert dingo population (Figure 2). The pattern of strong genetic differentiation (F_ST = 0.267–0.343) and limited geneflow between the Big Desert and other dingo populations likely reflects the influence of historical and ongoing lethal control programs in the surrounding area. Lack of knowledge concerning historical and contemporary biogeographic patterns in dingoes warrants further investigation using DNA samples from ancient or colonial-era individuals, and comparison to Oceanic canids such as the New Guinea Singing Dog (Surbakti et al., 2020).

Dingoes from all four of the wild populations were observed to have increased levels of homozygosity (median F_het > 0.3) relative to recent dingo backcrosses and dingo x domestic dog hybrids (Figure 7). We observed elevated homozygosity levels (median F_het > 0.7) in the Big Desert compared to other dingo populations, suggesting the population is experiencing inbreeding (Figure 7). A recent microsatellite-based study found the Big Desert dingo population to have genetic diversity levels up to 50% lower than other dingo populations (Stephens et al., 2022). Whilst the absolute value of homozygosity measures calculated for dingoes in this study could be inflated due to ascertainment bias, whole genome data from a small number of captive and wild dingoes also reveal elevated homozygosity levels relative to domestic dogs (Field et al., 2022; Kumar et al., 2023; Zhang et al., 2020). Further research is needed to assess whether lethal management programs have had an impact on homozygosity and genetic diversity levels in regional dingo populations, such as the Big Desert population (Hohenlohe et al., 2021; Quevedo et al., 2019; Weeks et al., 2016). The increased levels of homozygosity (median F_het > 0.3) found in dingoes may also reflect ancestral bottlenecks, which could be investigated in the future with runs of homozygosity (RoH) analysis of genomic data.

Domestic dog admixture in Australia

We herein provide the most accurate data regarding the prevalence of domestic dog ancestry in the dingo population, particularly in south-eastern Australia. Our survey of 307 wild dingoes from across Australia revealed that the occurrence and extent of domestic dog ancestry within the dingo population was limited and no feral domestic dogs were observed (Figure 2 and 3). In Victoria, where previous studies using microsatellites suggest that the dingo population is small and highly admixed (Cairns et al., 2021a; Stephens et al., 2015), we observed that 87.1% (n=54) of animals tested were pure dingoes, and 6.5% (n=4) were historical dingo backcrosses, with >93% dingo ancestry (Table 2). Similarly, in New South Wales and Queensland (n=171), where dingo x domestic dog hybridisation had been considered pervasive (Cairns et al., 2021a; Stephens et al., 2015), we only observed two wild canids with <70% dingo ancestry. We found that most of the sampled New South Wales and Queensland animals were pure dingoes (n=101) or historical dingo backcrosses (n=31). In the Northern Territory, South Australia and Western Australia we found little evidence of domestic dog admixture in the dingo population.

Our genome-wide SNP data demonstrates that dingoes and domestic dogs are genetically distinct (Figure 3 and 4). F_ST values indicate contemporary and ongoing geneflow between admixed and non-admixed dingoes. Ongoing hybridisation between dingoes and dogs of different breeds would disrupt patterns of geographic population structure and decrease homozygosity. The geographic structure and elevated homozygosity levels observed in dingoes, particularly in southern and eastern Australia, provides additional evidence that there is limited admixture between dingoes and domestic dogs (Figures 4, 5 and 7). Dingoes from southern Australia exhibit greater phenotypic variation than those elsewhere, particularly in terms of pelage (Cairns et al., 2021b; Jones, 1990; Jones, 2009; Newsome & Corbett, 1985). Whilst variable pelage was previously hypothesized to be the result of domestic dog admixture (Newsome & Corbett, 1985), our finding of rare domestic dog admixture makes this unlikely, and suggests that diverse coat colours may represent standing ancestral variation or perhaps local adaption (Cairns et al., 2021b).

It is hypothesized that dingo x domestic dog admixture patterns in Australia fit a unimodal hybrid swarm with extensive genetic swamping (Allen et al., 2017; Claridge et al., 2014; Daniels & Corbett, 2003; Stephens et al., 2015). However, our data suggests that the pattern of genetic admixture observed in dingoes fits a bimodal hybrid zone, with a majority of the population being pure dingoes or dingo backcrosses with low levels of intermediate dingo x dog hybrids (Figure 3 and 4, Table 2). The low occurrence of both first-generation dingo x dog hybrids and low dingo ancestry backcrosses further supports this hypothesis as it suggests the presence of behavioural barriers or assortive mating preferences (Allendorf et al., 2001; Fitzpatrick et al., 2015; McFarlane & Pemberton, 2019). It has also been suggested that lethal management strategies such as aerial 1080 baiting, particularly during the dingo breeding season, increases the risk of dingo x dog hybridisation by fracturing social structures and reducing the availability of dingo mates (Cairns et al., 2021a; Cairns et al., 2019; Wallach et al., 2009). This might explain why dog ancestry is more commonly observed in dingo populations where intensive or aerial baiting strategies are widespread, such as NSW and QLD (Table 2). Further research is needed to identify whether behavioural barriers to hybridisation or assortive mating preferences occur in dingo populations and to understand the impact of lethal control on patterns of dog admixture in dingo populations.

Dog ancestry detection methodology

Our results demonstrate that microsatellite-based DNA methods consistently estimate higher levels of dog ancestry in dingoes and miscategorised pure dingoes as dingo backcrosses compared to analysis based on 195,474 SNPs (Figure 8). Interestingly, dingoes from the Big Desert, South, East or Captive population clusters were more often miscategorised by microsatellite-based methods than those from the West (Table 5). Our findings are consistent with reports from a wide range of other species demonstrating that genome wide SNP analysis outperforms microsatellites in terms of accurate identification of ancestry and admixture (Dziech, 2021; Mattucci et al., 2019; McFarlane et al., 2020; Steyer et al., 2018; Stroupe et al., 2022; Szatmári et al., 2021; Vaha & Primmer, 2006; Väli et al., 2010; Zimmerman et al., 2020).

While microsatellite-based markers remain useful for short evolutionary timeframes, or in taxa that are slow evolving or clonal (Putman & Carbone, 2014), they are much less useful for studies of population genetics and demography, hybrid detection or analysis of large pedigrees (McFarlane & Pemberton, 2019; Putman & Carbone, 2014; Szatmári et al., 2021; Telfer et al., 2015; Tokarska et al., 2009; VÄLi et al., 2008; von Thaden et al., 2017; Zimmerman et al., 2020). We suggest that microsatellite DNA testing underestimates dingo ancestry because (1) it relies on geographically restricted reference populations with strict thresholds, (2) uses a relatively small number of genetic markers that are not strictly diagnostic and (3) the Bayesian modelling is typically constrained to scenarios with K=2. The microsatellite markers used for ancestry assessment in dingoes were chosen by comparing allele frequencies between a small core set of mixed breed domestic dogs and dingoes from captivity or western and central Australia (Cairns et al., 2021a; Elledge et al., 2008; Stephens et al., 2015; Wilton, 2001). We hypothesize that the under representation of dingoes from eastern and southern Australia in reference populations during both marker selection and analysis has confounded microsatellite-based dingo ancestry analyses. In light of these data, we urge wildlife managers to interpret microsatellite-based DNA testing data cautiously, particularly of dingoes from southern and eastern Australia.

Genomic studies have identified historical or recent admixture in a wide range of species including bears (Cahill et al., 2018; Cahill et al., 2015; Kumar et al., 2017; Pongracz et al., 2017), hares (Jones et al., 2018), jackals (Galov et al., 2015), dolphins (Brown et al., 2014; Gridley et al., 2018), ducks (Stephens et al., 2020), and parrots (Qu et al., 2012). Hybridisation and genetic admixture has been an integral part of evolutionary adaption and speciation but it also challenges wildlife facing urbanisation, changing climates and habitat fragmentation (Ottenburghs, 2021; vonHoldt & Aardema, 2020). Genotyping-by-sequencing or the use of targeted SNP panels may allow researchers to collect data from non-invasive samples like scat or hair for landscape scale monitoring of dingo and other wildlife populations. A key benefit of array or sequencing-based SNP genotyping is the ability to perform simultaneous investigations regarding the genetics of biogeographic variation, phenotypic traits, inbreeding, kinship relationships and demography, and facilitating more complete studies of dingoes and other wildlife. SNP based genotyping with commercially available microarrays offers the advantage of expandable data collection, i.e., SNP data generated at different times and locations, but with the same microarray, can be combined for a reproducible dataset of high integrity.

In a rapidly changing field, we highlight the challenge for scientists and wildlife managers to accurately identify admixture, diagnose the processes driving hybridisation, assess the threat to a species or population and develop evidence-based conservation management plans to protect our wildlife. It is clear that as science advances, policy makers and wildlife managers need to rely on modern genome science and genome wide assessment methods to understand contemporary dog admixture in dingo populations. Further, current policy and management plans regarding dingo conservation and management driven by microsatellite-based surveys should be revised and updated.

The identity and management of dingoes

Whilst dingo x dog hybridisation is not a pervasive threat to most dingo populations, we acknowledge that in parts of Australia, where there is a long-history of suppressing dingo populations, levels of hybridisation are, by comparison high. Thus dingo x dog hybridisation may be a threat to specific regional dingo populations. There is ongoing debate concerning when dog admixture in dingo populations poses a threat to conservation and at what genetic threshold a dingo backcross becomes a dingo for purposes of conservation policy (Allen et al., 2017; Claridge et al., 2014; Crowther et al., 2020). We encourage policy makers to adopt a flexible framework such as that outlined by vonHoldt et al. (2018) which assesses local conservation goals, ecological function and a genomic synthesis of species concepts. At a minimum it would be appropriate to expand the definition of dingoes in conservation policy and legislation to include historical dingo backcrosses using a genetic threshold of 93% dingo ancestry, as suggested by Allen et al. (2017).

In Australia the term “wild dog” is widely used in policy and legislation based on the logic that the wild canine population is dominated by mixed ancestry dingoes or feral domestic dogs (Bird & Bowman, 2016; DEPI, 2013; Fleming et al., 2001; NWDAP, 2020). Given our demonstrated findings of limited domestic dog admixture in dingoes, future use of the terms “dingo” and “feral domestic dog” will more accurately reflect the identity of wild canids in Australia. Furthermore, such a shift in terminology aligns with calls from Australian Aboriginal people for government policy to better acknowledge and respect the value of dingoes as a native animal and culturally significant species (Costello et al., 2021).