Widespread hybridization in the introduced hog deer population of Victoria, Australia, and its implications for conservation

Abstract In Australia, many species have been introduced that have since undergone drastic declines in their native range. One species of note is the hog deer (Axis porcinus) which was introduced in the 1860s to Victoria, Australia, and has since become endangered in its native range throughout South‐East Asia. There is increased interest in using non‐native populations as a source for genetic rescue; however, considerations need to be made of the genetic suitability of the non‐native population. Three mitochondrial markers and two nuclear markers were sequenced to assess the genetic variation of the Victorian population of hog deer, which identified that the Victorian population has hybrid origins with the closely related chital (Axis axis), a species that is no longer present in the wild in Victoria. In addition, the mitochondrial D‐loop region within the Victorian hog deer is monomorphic, demonstrating that mitochondrial genetic diversity is very low within this population. This study is the first to report of long‐term persistence of hog deer and chital hybrids in a wild setting, and the continual survival of this population suggests that hybrids of these two species are fertile. Despite the newly discovered hybrid status in Victorian hog deer, this population may still be beneficial for future translocations within the native range. However, more in‐depth analysis of genetic diversity within the Victorian hog deer population and investigation of hybridization rates within the native range are necessary before translocations are attempted.

itat of tall floodplain grasslands for agriculture and commercial development (Timmins et al., 2015). The largest non-native population of hog deer exists in Australia, where a stable, continuous population occurs throughout the Gippsland region in Victoria (Scroggie, Forsyth, & Brumley, 2012). Hunting of hog deer is restricted to April every year, and only one male and one female may be harvested per season per hunter, in part because the population is thought to be important for the long-term conservation of the species. Assessing the potential for the hog deer in Victoria to be used for conservation efforts is therefore not only important for the declining populations within the native range, but also for the management of the deer in Victoria. Negative impacts to native flora and fauna due to deer damage are well known, and hog deer can therefore be managed more effectively to mitigate these impacts if their conservation worth is effectively evaluated (Davis et al., 2016;Davis, Coulson, & Forsyth, 2008;Davis, Forsyth, & Coulson, 2010).
There are a number of considerations to be made when assessing the Victorian hog deer's suitability for translocations as part of conservation programs. Firstly, it is important to identify which subspecies has established in Victoria. The Acclimatisation Society of Victoria brought hog deer from South-East Asia to Victoria in the 1860s, primarily sent from ports in Sri Lanka and India (Mayze & Moore, 1990). The Sri Lanka population itself is thought to be comprised of introduced hog deer of unknown origin (Timmins et al., 2015). Furthermore, following these introductions to Australia, a new subspecies of hog deer was described in South-East Asia, the Indochinese hog deer (Axis porcinus annamiticus: Heude 1888). Today, this subspecies is considered critically endangered and occurs in isolated populations in Cambodia and Thailand; however, in the past its distribution was more widespread (Brook, Nask, & Channa, 2015;Maxwell, Nareth, Kong, Timmins, & Duckworth, 2006). Hunters in Victoria have also previously reported two distinct "forms" of hog deer found throughout Gippsland, with one form described as being smaller with a stockier build (Bentley, 1978).
The second consideration when evaluating the Victorian hog deer population for translocations is the possibility that either past or contemporary hybridization has occurred. Hybridization is prolific within the family Cervidae, and hybrid zones have been recorded in the genera Cervus (Lowe & Gardiner, 1975;McDevitt et al., 2009;Moore & Littlejohn, 1989;Senn & Pemberton, 2009), Odocoileus (Ballinger, Blankenship, Bickham, & Carr, 1992;Carr, Ballinger, Derr, Blankenship, & Bickham, 1986;Cathey, Bickham, & Patton, 1998), and most recently in Rusa (Martins, Schmidt, Lenz, Wilting, & Fickel, 2018). While hybridization between species within the Axis genus has not been reported in the wild, there have been cases of chital (Axis axis) and hog deer hybrid offspring being born in captivity, with animals from the two species needing to be separated due to their proclivity to interbreed (Gray, 1972;Mayze & Moore, 1990;McMaster, 1871). Chital have been released in Victoria in the past, including in areas where hog deer are now known to occur; however, these chital populations have since become locally extirpated (Forsyth, Stamation, & Woodford, 2016). Many species of deer were housed together in Royal Park (now the Royal Melbourne Zoological Gardens) prior to release ("A Few Hours in the Zoological and Acclimatisation Society's Grounds" 1873; "The Naturalist", 1873), so it is possible that hog deer and chital were housed in captivity together prior to their liberation. Additionally, previous research has shown that while native hog deer and chital mitochondrial genomes share a 94.65% identity, the mitochondrial genomes of Victorian hog deer and chital show a greater degree of similarity which may indicate hybridization; however, this was detected in only four samples from a managed island population in Victoria, and so may not be representative of the entire population (Hassanin et al., 2012;Hill, Linacre, Toop, Murphy, & Strugnell, 2017). There are also unconfirmed reports that another species from the Axis genus, the Bawean hog deer (Axis kuhlii), was introduced to Victoria and possibly released; however, there is some debate that the species introduced was actually the Javan rusa (Bentley, 1978;Mayze & Moore, 1990). Today, the hog deer range in Victoria overlaps with both sambar and fallow deer (Dama dama ;Forsyth, Stamation, & Woodford, 2015;Forsyth et al., 2016). While hybridization across two different genera of deer is likely to be rare, hybridization between fallow and hog deer has been recorded in captivity in the past, although it is unknown how long the offspring survived, or if they were fertile (Gray, 1972).
The final consideration necessary to evaluate the suitability of hog deer for conservation efforts is to estimate the genetic diversity present within the Victorian population. When choosing individuals to use for translocations, it has been suggested that capturing >95% of the genetic variation present in the source population is necessary in order to offset the negative genetic impacts present in the receiving population (Weeks et al., 2011). The founding population F I G U R E 1 Maps showing the collection sites for samples used for sequencing analysis in this study: (a) chital (Axis axis) sampling locations in Queensland, Australia; (b) hog deer (Axis porcinus) sampling locations in Victoria, Australia; and (c) Indochinese hog deer (Axis porcinus annamiticus) sampling location in Koh Kong Province, Cambodia. Green circles indicate sites where sequences for all five genes were obtained, and purple circles indicate sites where only the D-loop was sequenced of hog deer in Victoria was only comprised of 15 individuals (nine females, two males, and four of unknown sex); however, the population has thrived since being released, and a continuous breeding population is present across a 2,336 km 2 range Mayze & Moore, 1990). In order to capture as much genetic diversity as possible, it is crucial to sample individuals throughout the entire range in Victoria to determine how much variation is present in the population, and if particular sites contain greater diversity and are therefore more suitable for translocation.
Genetic analyses of mitochondrial and nuclear DNA are able to address the suitability of the Victorian population of hog deer as a potential source population for conservation efforts of the species.
Mitochondrial markers, and the nuclear markers alpha-lactalbumin (αLalb) and protein kinase C iota I (PRKCI), have been used in the past to investigate the phylogeny and genetic diversity of various deer species (Ludt, Schroeder, Rottmann, & Kuehn, 2004;Vernesi et al., 2002) and have been shown to be effective at delineating species (Hassanin & Douzery, 2003;Ropiquet & Hassanin, 2005). These markers will be utilized in this study to first establish which species or subspecies of hog deer was introduced to Australia, and then to assess the genetic diversity and any occurrence of hybridization within this population in order to ascertain the value of the Victorian population as a source for genetic rescue within the native range.

| Sampling and DNA extraction
A total of 78 liver and tongue samples were collected from wild, free-ranging hog deer (presumed Axis porcinus) during the hunt-

| PCR amplification
Three mitochondrial markers and two nuclear markers were used in this study. Universal markers for the mitochondrial genes, cytochrome b (Cyt b) and cytochrome oxidase subunit I (COI), were utilized, as well as custom-made D-loop primers (Branicki, Kupiec, & Pawlowski, 2003;Folmer, Black, Hoeh, Lutz, & Vrijenhoek, 1994;Kocher et al., 1989; Table 1). Nuclear regions of intron 2 of the α-lactalbumin gene (αLalb) and the intron of protein kinase C iota (PRKCI) were also sequenced (Hassanin & Douzery, 2003;Ropiquet & Hassanin, 2005; Table 1). Twenty-two samples of VIC hog deer and five samples of QLD chital were sequenced for the above-described genes. All regions were also sequenced for one Axis porcinus annamiticus sample, except the PRKCI gene which failed to amplify. The samples of hog deer selected for sequencing of all five gene regions were chosen to represent the spatial distribution of the hog deer population in Victoria. A further 58 samples of hog deer, 30 samples of chital, and one additional sample of A. p. annamiticus were sequenced for the D-loop region to assess the genetic diversity of the populations. This region was chosen to further investigate genetic diversity as the mitochondrial control region, where the D-loop is located, is considered to be the most polymorphic section of the mitochondrial genome, and several studies have utilized the D-loop to assess genetic diversity in several deer species in the past (Hu, Fang, & Wan, 2006;Moritz, Dowling, & Brown, 1987;Pérez-Espona et al., 2009;Skog et al., 2009).

| Phylogenetics
Sequences were aligned using the Geneious alignment method in Geneious 9.0.5 (Kearse et al., 2012). Additional sequences of hog deer and chital were downloaded from GenBank to include in the dataset (Table 2). Further sequences of all deer species that have been introduced to Australia, including those that did not become established, were also included in the dataset where available ( Table 2). Sequences of moose (Alces alces) were used as an outgroup for phylogenetic analyses. These sequences were trimmed and aligned with the hog deer, chital, and Indochinese hog deer sequences. Mitochondrial genes were concatenated for analyses, while the nuclear genes αLalb and PRKCI were analyzed separately. Alignments were run in JModelTest 2.1.7 to determine the best-fit model for each alignment, using the Akaike information criterion (Akaike, 1974;Darriba, Taboada, Doallo, & Posada, 2012). Best-fit models were GTR + I + G for the concatenated mitochondrial alignment, TrN + G for the αLalb gene, and F81 + G for the PRKCI gene. Bayesian phylogenetic analyses were run in MrBayes 3.2.6 to calculate posterior probabilities, which were run for 1,000,000 generations, after a burn-in period of 100,000, and sampling trees every 200 generations (Huelsenbeck & Ronquist, 2001). This analysis was performed twice for every gene, to ensure convergence toward the same likelihood score. Maximum likelihood phylogenetic trees were created using the program PHYML, using the same models described previously, with 1,000 bootstrap replicates (Guindon et al., 2010). The program PAUP 4.0 was used to calculate genetic distances for the concatenated mitochondrial tree, using the model identified in JModelTest 2.1.7 (Swofford, 1999). Note: These species were all historically introduced to Australia (present-day species in Australia in bold), as reported by Moriarty (2004). Sequences could not be found for C. canadensis and Moschus sibiricus; so, closely related species were used as a substitute (asterisk). a Hassanin et al. (2012). b Li, Ba, and Yang (2016). c Gilbert et al. (2006). d Hassanin and Douzery (2003). e Matthee, Burzlaff, Taylor, and Davis (2001).
TA B L E 2 Sequences obtained from GenBank that were combined with the main dataset

| Population genetics
Sequences of the D-loop region were trimmed to 576 bp and aligned using the Geneious alignment method in Geneious 9.0.5 (Kearse et al., 2012). Measures of haplotype diversity and nucleotide diversity were calculated for the VIC hog deer and QLD chital using DNAsp 5.10.1 (Librado & Rozas, 2009). The neutrality tests Tajima's D and Fu's Fs were calculated for each population in the program Arlequin 3.5 (Excoffier & Lischer, 2010;Fu, 1997;Tajima, 1989 for the genetic diversity and neutrality indices described above.

| Phylogenetics
All unique sequences generated in this study for each species have been deposited in GenBank (accession no. MN226858-MN226880).    Table 5).

| Population genetics
In comparison, much greater diversity was detected in the native The median-joining haplotype network created with the combined chital and hog deer data generated in this study and taken from GenBank shows four distinct groups within the data ( Figure 6).
Only two haplotypes are present in the Australian populations, both of which are found in the broader chital group. One of these hap-

| D ISCUSS I ON
The mitochondrial and nuclear data presented here portray an interesting insight into the recent genetic history of hog deer in Victoria, Australia. Due to the discovery of chital haplotypes in the mitochondrial DNA of hog deer in Victoria, it is not possible to identify which species/subspecies of hog deer was initially introduced using traditional barcoding methods. Results from the nuclear gene region αLalb indicate that these haplotypes are more closely related to Axis porcinus rather than Axis porcinus annamiticus, which may suggest that the species introduced to Australia was the Indian hog deer A. porcinus. However, given the low levels of resolution at this nuclear marker, additional analysis using more variable nuclear STRs or SNPs is needed to firmly conclude which hog deer species was released in Victoria.
The mitochondrial data showed that hog deer in Victoria possess haplotypes that are most closely allied with the chital, a species that historically was released and established a population in Victoria but has since become locally extirpated . This may be a result of incomplete lineage sorting; however, it is more likely that hybridization has occurred between these two species. The time of divergence between chital and Axis porcinus occurred during the Pliocene, approximately 2.6 Mya, and a number of mitochondrial haplotypes exclusive to each species have been detected, as seen in the haplotype network presented in this study (Gilbert et al., 2006;Gupta et al., 2018;Hassanin et al., 2012). The presence of these species-specific haplotypes suggests that incomplete lineage sorting is unlikely to be a factor in-  (Smith, Carden, Coad, Birkitt, & Pemberton, 2014;Smith et al., 2018). In both cases, hybridization occurs between the female of the larger deer species (chital and red deer) and the male of the smaller deer species (hog deer and sika deer), and with males generally being the larger sex this hybridization pattern may be reflective of the phenotypic limitations that the reciprocal cross would present. However, this may be less of a consideration between hog deer and chital as overall sizes are relatively similar, and crosses have been reported to occur both ways (Gray, 1972). Hybrids between hog deer and chital also appear to favor hog deer-like phenotypes. McMaster (1871) describes hybrids as having similar behavioral characteristics as hog deer, and with darker fur and fainter spots than chital, while Gray (1972) reported a hybrid that resembled hog deer in "head, face, and horns," with a white-spotted coat resembling chital. It is also possible that backcrossing to a parental species has occurred following the hybridization event, which may further explain the hog deer appearance seen in the current population. Furthermore, F I G U R E 6 Median-joining haplotype network based on the mitochondrial D-loop, comprising Victorian hog deer, Queensland chital, and native Axis axis, Axis porcinus, and Axis porcinus annamiticus samples. Circle size is indicative of the frequency of each haplotype. Numbers indicate sample sizes for common haplotypes. Hatch marks represent the number of bp changes between haplotypes. Black circles represent median vectors the presence of a shared nuclear haplotype between Victorian hog deer and Axis porcinus at the αLalb gene provides further support for hybridization between the two species of deer.
This study is the first to report on long-term persistence of chital and hog deer hybrids in a wild setting. It is important to note, however, that it is unknown when the initial hybridization occurred; hybridization may have arisen in the wild in either the native range or after release of both chital and hog deer in Wilsons Prom in the 1860s or may have occurred in captivity prior to release. The scenario presented here is also somewhat unusual; traditionally, hybridization occurs when the distribution of two genetically distinct populations shares overlapping ranges, mates, and produces viable offspring, with the offspring forming a hybrid zone where overlap occurs (Shurtliff, 2013). However, no pure stock of chital is present within Victoria as the population became locally extinct in the 1920s, and no pure hog deer were identified during the course of this study. This would suggest that either both parental species have been essentially bred out of existence over generations, or only hybrids were ever released into the wild. As the original parental species are no longer present, further assessment of hybridization within Victoria, particularly detecting backcrossing to either species, is difficult as samples of parental species that contributed to the hybridization are needed for more in-depth analysis.
The continued survival of the hybrid population in Victoria over many generations without the presence of either parental species demonstrates that hybrids between chital and hog deer are fertile.
The chromosome numbers differ in these two species (chital 2n = 66, hog deer 2n = 68; Khongcharoensuk et al., 2017;Pinthong et al., 2017); however, this is not unique to hog deer and chital, with other species known to hybridize in the family Cervidae also comprising different chromosome numbers (Bonnet-Garnier, Claro, Thevenon, Gautier, & Hayes, 2003;Gustavsson & Sundt, 1968). Robertsonian translocations of chromosomes, whereby the fusion of whole arms of two acrocentric chromosomes occurs, are common in cervids (Bonnet-Garnier et al., 2003;Huang, Chi, Nie, Wang, & Yang, 2006), and it is likely that the prevalence of these chromosome translocations has assisted in the production of fertile hybrids where chromosome numbers are different. Robertsonian translocations have been detected in red deer and sika deer hybrids (Herzog & Harrington, 1991), and it is feasible that further investigation of hybrid hog deer and chital through karyotyping will reveal similar mutations that have made fertile hybrids possible.
The mitochondrial D-loop region was discovered to be monomorphic within the Victorian hog deer population, suggesting that the diversity at this region of the mitochondrial genome is very low. Similar findings have been reported for other species introduced to Australia belonging to the Order Artiodactyla; mitochondrial analysis of Banteng revealed that this species was monomorphic at one mitochondrial gene and two nuclear genes (Bradshaw et al., 2006), and analysis of the D-loop of Australian populations of dromedary camels (Camelus dromedarius) revealed only 13 haplotypes, which was considered low by the authors as the founder size of dromedary camels was <5,000 individuals, and their population size is now considered to be greater than 1 million animals (Spencer et al., 2012 (Brook et al., 2015;Humphrey & Bain, 1990;Maxwell et al., 2006). These distinct annamiticus haplotype differences between Cambodia and Thailand warrant further research to ascertain whether the Cambodian A. p. annamiticus haplotype is distributed elsewhere and is in need of conservation intervention.
Despite low reported genetic diversity which is suggested to negatively impact populations, the initial population size of hog deer has expanded considerably since their release in Victoria. This may be explained by the enemy release hypothesis, whereby an introduced species becomes abundant and a successful invader as their population sizes are no longer affected by their native predators or pathogens (Keane & Crawley, 2002). In their native range, hog deer are an important prey item for many species, including leopard (Panthera pardus fusca), clouded leopard (Neofelis nebulosa), Burmese python (Python bivittatus), Bengal tiger (Panthera tigris tigris), and dhole (Cuon alpinus), and so, liberation from these predators is likely to positively impact abundance of the hog deer in Victoria (Dhungel & O'Gara, 1991;Grassman, Tewes, Silvy, & Kreetiyutanont, 2005;Prasanai, Sukmasuang, Bhumpakphan, Wajjwalku, & Nittaya, 2012;Wegge, Odden, Pokharel, & Storaas, 2009). Alternatively, there may have been some genetic fitness associated with the population, particularly immediately after hybridization. Heterosis, or "hybrid vigor," may have made initial establishment of the hog deer in Victoria much easier than if pure stock alone had been introduced. Hybridization introduces many novel alleles into the population with which natural selection can act upon, thereby increasing overall fitness (Weeks et al., 2011;Whiteley, Fitzpatrick, Funk, & Tallmon, 2015). Hybrid vigor has been implicated in the successful introductions of many plant species (Durand et al., 2002;Moody & Les, 2002) and is now being recognized as advantageous in several animal species as well (Drake, 2006;Facon, Jarne, Pointier, & David, 2005). However, as a small founding population was established and no additional hog deer were introduced to the main population following initial release, these potential genetic benefits are likely no longer affecting the persistence of the Victorian population.
Although genetic variation is reported to be low at the mitochondrial regions in the Victorian hog deer population, it may still be worthwhile to use this population as a source for genetic rescue. Due to the discovery of hybridization with chital, translocation should be restricted to areas where both species are present in the northern regions of India. It is currently unknown how prolific hybridization is between hog deer and chital in their native range, but considering that the two species share overlapping ranges, the existence of a hybrid zone is probable. Alternatively, if natural hybrid zones are not detected in the native range after extensive study, this may give further credence to the idea that hybrids between the two species only occur in captivity and, as such, would narrow down the possibilities of where hybridization most likely occurred in the Victorian population. Often, the most significant concern cited when attempting genetic rescue is the possibility of introducing outbreeding depression in the population, whereby the offspring of parents with distinct genetic differences show a decrease in overall fitness, as they are no longer well adapted to their current environment. Currently, there is a shift away from the belief that translocation of distinct populations will lead to outbreeding depression and "genetic swamping" of locally adapted alleles (Frankham, 2015;Weeks et al., 2011 Bhutan, and Cambodia with little to no genetic assessment of these populations (Timmins et al., 2015). Future research within the native range should focus on the connectivity of hog deer across the native landscape and identifying the subspecies boundary between A. porcinus and A. p. annamiticus, with an overarching goal of promoting genetic diversity and effective management of hog deer across South-East Asia. Hog deer have also previously been reported in China, Myanmar, Viet Nam, and Laos; however, it may be locally extinct in these areas, so monitoring in habitats where hog deer have previously occurred in these countries is needed to firmly establish whether local extirpation has occurred. Additionally, an investigation into the possibility of a natural hybrid zone between hog deer and chital in India is necessary, as this will likely dictate whether translocation of Victorian hog deer is suitable for genetic rescue of the species in India. Further analysis of the Victorian population of hog deer is also warranted; this study was unable to examine the genetic diversity of Victorian hog deer in-depth as the D-loop region of the mitochondria within this population was monomorphic. Moreover, the nuclear markers chosen for this study were shown to provide low discrimination power when comparing hog deer and chital sequences. The inclusion of polymorphic nuclear STR or SNP analysis would likely address these lingering questions and could also be used for monitoring hog deer populations pre-and post-translocation, to understand the long-term effects of using Victorian hog deer hybrids as a source for genetic rescue within the hog deer native range.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflict of interest.