Genetic origins and diversity of bushpigs from Madagascar (Potamochoerus larvatus, family Suidae)

The island of Madagascar, situated off the southeast coast of Africa, shows the first evidence of human presence ~ 10,000 years ago; however, other archaeological data indicates a settlement of the modern peoples of the island distinctly more recent, perhaps > 1500 years ago. Bushpigs of the genus Potamochoerus (family Suidae), are today widely distributed in Madagascar and presumed to have been introduced from Africa at some stage by human immigrants to the island. However, disparities about their origins in Madagascar have been presented in the literature, including the possibility of endemic subspecies, and few empirical data are available. Furthermore, the separation of bushpigs in Madagascar from their mainland relatives may have favoured the evolution of a different repertoire of immune genes first due to a founder effect and then as a response to distinct pathogens compared to their ancestors. Molecular analysis confirmed the species status of the bushpig in Madagascar as P. larvatus, likely introduced from the central region of southern Africa, with no genetic evidence for the recognition of eastern and western subspecies as suggested from previous cranial morphology examination. Investigation of the immunologically important SLA-DQB1 peptide-binding region showed a different immune repertoire of bushpigs in Madagascar compared to those on the African mainland, with seventeen exon-2 haplotypes unique to bushpigs in Madagascar (2/28 haplotypes shared). This suggests that the MHC diversity of the Madagascar populations may have enabled Malagasy bushpigs to adapt to new environments.

The first evidence of human presence in Madagascar, situated off the south-eastern coast of Africa, dates from close to 10,000 years ago 1 . Subsequently, the island was settled by Austronesians approximately 1500-3000 years ago and then soon thereafter by Bantu groups from East Africa, but evidence of this remains unclear [2][3][4] . The bushpig (Potamochoerus larvatus; Suidae, Artiodactyla) is suggested to have been introduced to Madagascar from eastern Africa by early sea navigators who settled on the island 5,6 . The earliest archaeological evidence for the bushpig on Madagascar dates to the tenth to thirteenth centuries 7 . On other regional islands, archaeological records show evidence of bushpigs on the Comoro Islands from the ninth to tenth centuries 8,9 . However, there are large gaps in the paleontological record of Madagascar between the Late Cretaceous and the Late Pleistocene (~ 66 million to 120,000 years ago) 10 , as well as the Holocene (the earliest possible Potamochoerus introduction to Madagascar), hampering the determination of when and where the species was first established on the island and whether this predates human arrival. Genetic evidence is useful to assess where the Potamochoerus lineage(s) on the island originated from and determine whether their separation resulted in any significant Scientific Reports | (2020) 10:20629 | https://doi.org/10.1038/s41598-020-77279-5 www.nature.com/scientificreports/ a role in maintaining ASFV in Madagascar 11,23 . The major histocompatibility complex (MHC) is an important part of adaptive and innate immunity in vertebrates 29 and has been studied comprehensively in model species including the domestic pig (Sus scrofa) and humans [30][31][32][33] . Genetic diversity in this genomic region contributes to individual and population fitness in natural populations through pathogen recognition 12,34 . In particular, the antigen-binding sites (ABS) within the class I and class II MHC molecules are responsible for self and non-self recognition and present peptides to the CD4+ T lymphocytes for surveillance 35,36 . The high genetic diversity in the MHC can be caused by different evolutionary mechanisms such as gene duplication and pathogen-mediated co-evolution and consequently diversification of the ABS through balancing and positive selection for advantageous alleles [37][38][39] . Balancing selection has been described as one of the major factors driving this diversity as a higher number of different alleles (heterozygote advantage) could bind and present different antigens 40,41 . Thus, examining the MHC of bushpigs from Madagascar could provide an opportunity to better understand how these mechanisms have changed the MHC repertoire of Potamochoerus over space and time.
In this study, the origins and taxonomic position of bushpigs from Madagascar were assessed using the Sanger sequencing data of mitochondrial (control region; CR, and cytochrome b; cytb) and nuclear markers (glucosephosphate isomerase-processed pseudogene; GPIP, and melanocortin 1 receptor; MC1R). A range of suids and peccaries (nine species), and specimens of Potamochoerus from Madagascar were sequenced in this study. Due to the varied bioclimatic regime found in different areas of the island 42 , as compared to mainland Africa, we sequenced three MHC loci to assess whether the separation of Malagasy bushpigs generated a different repertoire of immune genes compared to mainland bushpigs. These genetic analyses provide insights into the diversity of the Malagasy bushpig adaptive immune system in comparison to mainland bushpigs since their separation. Similar comparisons could be used to assess the fitness and historical changes incurred by other species translocated to Madagascar.

Materials and methods
Sampling. Genomic DNA was extracted from blood and tissue samples from a total of 80 individuals (Table 1). Samples included 55 bushpigs (P. larvatus) consisting of 13 specimens from mainland Africa (Zimbabwe, Tanzania-museum specimen, South Africa; SA), and 42 specimens from regions of "West" Madagascar (general Mahajanga Province area, which includes a museum specimen collected near the Ambohijanahary Special Reserve and two other protected areas: the Menabe Antimena and the Ankarafantsika National Park), "East" Madagascar (Andasibe-Mantadia protected area), and "North" Madagascar (museum specimen collected from Forêt d'Antsahabe, Antsiranana Province) (Fig. 1). Genomic DNA for non-museum specimens and museum specimens were extracted using the DNeasy Blood & Tissue Kit (Qiagen, France) and phenol/chloroform method, respectively. Museum specimens were provided by the Field Museum of Natural History, Illinois, USA. Samples were collected under permission from the Direction Générale des Eaux et Forêts de Madagascar (Supplementary Table S1). The tissue samples were from recently killed bushpigs found during field trips and skulls were kept as voucher specimens. Although Madagascar no longer uses the province system in an administrative manner, we maintain these names for geographical descriptive purposes. For phylogenetic comparison and to provide an overview on the similarity of immune genes, the study also included sampling of three red river hogs (P. porcus), 16 collared peccaries (Pecari tajacu), and one specimen of the following species: European wild boar (S. scrofa), Bornean beared pig (S. barbatus), Celebes warty pig (S. celebensis), Buru babirusa (Babyrousa babyrussa), common warthog (Phacochoerus africanus), and giant forest hog (Hylochoerus meinertzhageni) ( Table 1). Outgroup species were obtained from previous studies [43][44][45] .  46 and cytb 47 , and the nuclear GPIP region 48,49 and MC1R 50 region were sequenced. Mitochondrial markers have been extensively used to infer evolutionary relationships of many organisms [51][52][53][54] . The control region and cytochrome b has been particularly valuable in Suidae and Tayassuidae; identifying differences amongst species and pig breeds 48,[55][56][57] . The mitochondrial markers were concatenated as they are essentially a single, non-recombining locus. The nuclear marker GPIP was sequenced due to minimal selection pressures and previous studies have shown the ability of this marker to differentiate between S. scrofa clades 55,58 but has not been used in the context of wild suid populations. The nuclear marker MC1R plays a key role in regulating eumelanin (black/brown) and phaemelanin (red/yellow) and is responsible for coat, hair, and skin colour variation in various pig breeds 48,59 and may be useful in distinguishing the presence of population differences based on coat colour. Partial genes from each of the MHC class I, II and III were also sequenced to assess the immunogenetic diversity of bushpigs from Madagascar. The MHC class I classical genes were not targeted due to their high level of gene duplication which can reduce the resolution of analyses. Instead, the MHC class I nonclassical SLA-6 exon-2 region was targeted (alpha-1), as well as SLA-DQB1 exon-2 (class II), and region spanning residues 450-518 (exons [11][12][13][14] of the class III BAG6 gene, which has been suggested to be associated with ASFV infection in the host 60 Inferring the relationship of bushpigs in Madagascar. Phylogenetic inferences were generated based on Datasets 1-3 (Table 3) using the RAxML-NG v0.09.0 web-based server 79 . The partitions to accommodate for different evolutionary models were based on the Bayesian Information Criterion 80 identified by

Analyses of MHC genetic diversity and selection.
To observe whether the MHC sequence of bushpigs from Madagascar has been subjected to purifying or diversifying selection since their introduction to the island, we performed selection analysis using DataMonkey 83 . We estimated selection based on (1) all haplotypes identified, (2) bushpigs from Madagascar only, and (3) mainland bushpigs-selection for SLA-6 was based on all haplotypes only as three or more input sequences in DataMonkey is required. Selection tests can be used for identifying certain positions in genes or regions (particularly in coding regions) that are conserved or have underwent radical changes in some taxa, and for testing an increase of nonsynonymous substitutions than expected under neutral evolution 84 . Diversifying selection is observed by a higher nonsynonymous substitution rate (dN) compared to the synonymous substitution rate (dS). A dN/dS ratio > 1 suggests the presence of positive or diversifying selection; dN/dS < 1 indicates negative or purifying selection; and dN/dS ~ 1 indicates no selection. We tested four different methods for detecting sites under selection: MEME (Mixed Effects Model of Evolution), FEL (Fixed Effects Likelihood), SLAC (Single-Likelihood Ancestor Counting), and REL (Random Effects Likelihood). For each method, the significant levels were the following: P ≤ 0.05 in MEME 85 , P ≤ 0.25 in FEL and SLAC, and a Bayes Factor of > 50 in REL 86 . Only sites significantly identified under positive selection by at least two methods were considered 87,88 . Genetic diversity was calculated in MEGA 7 89 using the model-of-best-fit Tamura 92 (T92) 90 , Kimura-2 parameter (K2) 91 , and Jukes-Cantor (KC) 92 for SLA-DQB1, SLA-6, and BAG6, respectively. Standard error was calculated using 1,000 bootstraps. Based on initial genetic diversity analysis, genetic distance was also calculated for SLA-DQB1 in MEGA 7 using p-distance in coding positions only and sites with < 95% coverage were discarded from analysis.
Divergence of bushpigs from mainland Africa and Madagascar. BEAST v2.4.8 93 was used to estimate the divergence time between bushpigs in mainland Africa and Madagascar. Concatenated mitochondrial CR and cytb sequence alignments were subsampled to include members of Potamochoerini, Phacochoerini, and Suini, a Babyrousa, and a Pecari as an outgroup (Dataset 1a; Table 3; Supplementary Table S1). Independent test runs using the MC1R and GPIP nuclear genes were also performed and evaluated. The sequences were aligned using ClustalW algorithm in Geneious v.11.0.4 (https ://www.genei ous.com) 94 . PartitionFinder2 on XSEDE 95 was used to determine the best substitution model and best partitioning scheme for the dataset. It provided four data blocks including the CR, and the first, second, and third codon of the cytb gene (Supplementary Table S3). The best partitioning scheme and substitution model as indicated by PartitionFinder were used in the downstream BEAST analysis. BEAST was implemented using a calibrated Yule tree prior and a single uncorrelated relaxed lognormal clock model. The molecular clock was calibrated by constraining the time to most recent common ancestor (TMRCA) of the Suoidea (Tayassuidae-Suidae split) according to normal distribution prior with a mean of 37.09 Mya and a sigma of 0.5. The Suiodea TMRCA prior around of 34.50-39.69 Mya was previously estimated and is in consensus with other estimated molecular and fossil analyses 67 . The BEAST MCMC chain was set to run for 20 million generations (replicated three times) with the first 10% set as burn-in. The models and parameters for the BEAST run were prepared using BEAUTi 96 . The BEAST analysis was independently replicated three times using a different starting seed each time. All the generated log and tree files were combined using LogCombiner 97 . The BEAST analysis was evaluated using Tracer v1.7.1 98 by ensuring an effective sample size (ESS) of > 200 and convergence. TreeAnnotator 97 was used to summarize the tree topologies, generate the best supported BEAST trees (maximum clade credibility tree) corresponding to the 95% HPD ranges, and to estimate the posterior clade probability of each node. FigTree v.1.4.3 (https ://tree.bio.ed.ac.uk/softw are/figtr ee/) 99 was used to visualize and evaluate the estimated tree.

Results
Phylogenetic relationship and divergence of bushpigs from Madagascar to other Suidae. Phylogenetic analysis of concatenated mtDNA (Dataset 1a; Fig. 2) and non-concatenated mtDNA genes (Datasets 2-3; Supplementary Fig. S1 and Fig. S2) using the maximum likelihood method showed that specimens of bushpigs from Madagascar and mainland bushpigs formed a monophyletic clade. Bushpigs from "East", "West" or "North" Madagascar, did not form reciprocal monophylogetic lineages. Furthermore, some mainland bushpigs, including those from Zimbabwe (n = 3) and Tanzania (n = 1) appear as a sister clade to Madagascar bushpigs. This is followed by those from South Africa (n = 2), and the remaining specimens from Zimbabwe (n = 6). Trees inferred from Dataset 1b ( Supplementary Fig. S3) was consistent to Dataset 1a. The topology for nuclear DNA markers was less resolved. Both GPIP and MC1R showed little resolution of bushpig samples but were sufficient to cluster other extant suids separately from bushpigs (Supplementary Fig. S3 and Fig. S4). Surprisingly, one bushpig sample obtained from Tervuren (Belgium) clustered with S. scrofa samples for each analysis; presumably due to sampling error/mislabelling. The specimens of bushpigs from Madagascar employed herein are genetically distinct, even in the highly conserved GPIP and MC1R genes, from the P. porcus specimens available for this study. We did not estimate divergence time between Madagascan and mainland African bushpigs using the nuclear gene MC1R and pseudogene GPIP as test runs produced very unresolved trees. In addition, previous work did not include the MC1R gene for divergence time estimation due to strong phenotypic selection 48 .  Fig. S6 and Fig. S7) showed no shared haplotypes between bushpigs from Madagascar and mainland Africa. Both CR and cytb contributed the most to the genetic differentiation between these two groups (13 and 14 haplotypes, respectively) each with five haplotypes unique to Madagascar. In contrast, the nuclear GPIP and MC1R genes ( Supplementary Fig. S8) were highly conserved between bushpigs showing respectively three (two found in Madagascar) and four haplotypes (three found in Madagascar), and provided the lowest resolution of bushpig relationships. The above was supported by concatenated sequences in which both mtDNA genes were successfully sequenced (Dataset 1a; Table 3), showing that specimens from Madagascar clustered closely together and separately from mainland specimens (Supplementary Fig. S9). In this case, nine haplotypes were identified for bushpigs from Madagascar, and of these, four haplotypes here were classified as mt1 using Dataset 1b (Fig. 3). Two of these haplotypes were found only in the Ankarafantsika specimens and the general Mahajanga Province region. Based on MJN analysis which showed the close relatedness of specimens from across Madagascar (one mutational step between haplotypes), the occurrence of multiple bushpig introductions to the island is unlikely.
Regarding mainland bushpigs, based on Datasets 1a and 1b, five unique haplotypes (mt7, 8, 9, 10, and 11) were identified in Zimbabwe bushpigs and were more distantly related to each other compared to bushpigs from Madagascar (higher number of substitutions; Fig. 3 and Supplementary Fig. S9). The Zimbabwe bushpgis were also genetically different from bushpigs sampled from the Soutpansberg (mt5), KwaZulu-Natal (mt12), and Tanzania (mt4). The Soutpansberg (mt5) and KwaZulu-Natal (mt12) bushpigs were consistently grouped separately from other bushpigs from Madagascar in all other mtDNA analyses ( Supplementary Fig. S6 and Fig. S7). These results suggest that bushpigs from the island are related to the specimens from the central region of southern Africa. However, the absence of genetic data from southern Tanzania and Mozambique also limits the resolution of the relatedness between bushpigs from other regions in mainland Africa and Madagascar. www.nature.com/scientificreports/ Overall, based on Dataset 1, the most frequent haplotype was mt1 found in 91.5% of bushpigs from Madagascar (Table 4). It was mostly found in specimens from "West" Madagascar (90.5%), where mt13-15 were also found at low frequencies (2.7-3.6%).
Haplotype structure and genetic diversity of three MHC loci. Median-Joining Network analysis found no specific relationship or clustering for SLA-DQB1 exon-2 sequences (Figs. 5 and 6). However, phylogenetic analysis suggests that certain haplotypes clustering together in the same clade may perform similar functions (as supertypes) due to shared antigen-binding sites. Most of these were shared within bushpigs ( Fig. 7; groups 1, 2, 3, and 4), while group 5 was shared between bushpigs and P. africanus, and group 6 was found in B. babyrussa only. Antigen-binding site Groups 2 and 5 were found in relatively low frequencies (average 3.4% and 2.3%, respectively).

with 28 haplotypes-19 in bushpigs from Madagascar and 11 in mainland individuals (Supplementary
Genetic distance analyses indicated that the haplotypes are also very distinct from those found on the mainland; genetic distance between Madagascar and mainland bushpig sequences ranged from 0.031 to 0.058, while sequences from Madagascar (Ankarafantsika, Menabe Antimena, and Andasibe-Mantadia) had a genetic distance ranging from 0.026 to 0.035 (Supplementary Table S6).
In contrast to SLA-DQB1, the class II SLA-6 and class III BAG6 genes were more conserved with only two and four haplotypes found, respectively. For both genes, only one haplotype was found in Madagascar (SLA-6-1   www.nature.com/scientificreports/ Table S7), BAG6-3 to 5 were found in lower frequencies but the aforementioned and BAG6-6 produced the same amino acid sequence as the most common haplotype (BAG6-1) and thus appears fairly conserved across mainland and Madagascar bushpigs and not found in other species. Based on amino acid translation, haplotypes BAG6-1 and BAG6-6 were found in other sub-Saharan African species and B. babyrousa, respectively. BAG6-7 and BAG6-8 were only found in P. tajacu. The overall genetic diversity of SLA-6 was lower (nt: 0.03; AA: 0.04; Table 6) than BAG6 (nt: 0.017; AA: 0.01; Table 6), although the diversity was lower in BAG6 when only considering bushpigs (nt: 0.004; AA: 0.004). However, the S.E. for both these genes was high, possibly due to the low number of haplotypes found. In addition, the number of variable sites was extremely low. BLAST searches also found three identical matches in public databases, with SLA-6-2 matching several SLA-6 haplogroups of S. scrofa, SLA-6-4 being identical to SLA-6*10:01 (Supplementary Table S9), and BAG6-2 being identical to an existing S. scrofa sequence (Supplementary Table S10). These sequences were also shared with other Sus species. The 86 SLA-DQB1, 74 SLA-6, and 65 BAG6 sequences were submitted under GenBank accession numbers: MT853335-MT853419, MT853262-MT853334, and MT853420-MT853483.

Selection analysis of MHC in bushpigs from Madagascar.
Several SLA-DQB1 sites were found to be under (diversifying) selection when considering all haplotypes or subcategories (mainland Africa or Madagascar; Fig. 8). Two and one sites were detected under diversifying selection respectively in mainland bushpigs (8G: MEME P-value: 0.03, REL BF: 208.31; 21G: MEME P-value: 0.03, SLAC P-value: 0.13, REL BF: > 50), and bushpigs from Madagascar (52F: FEL P-value: 0.16, SLAC P-value: 0.16). Site 21 was the most diverse in mainland bushpigs, coding for up to four different residues, the same site only coding for one residue in bushpigs from Madagascar. In contrast, site 52 coded for up to five different residues in bushpigs from Madagascar, compared to only three in mainland bushpigs (Fig. 8). In addition, one site was significant for purifying selection in bushpigs from Madagascar (15G: FEL P-value: 0.06, SLAC P-value: 0.13, REL BF: > 50), and was conserved across all haplotypes.
Results show that the SLA-6 exon-2 region was highly conserved (Supplementary Fig. S12), with two sites subjected to purifying selection in more than two methods (42 N: FEL P-value: 0.07, SLAC P-value: 0.25; and 65D: FEL P-value: 0.09, SLAC P-value: 0.25) and one significant for diversifying selection (57R: FEL P-value: 0.25, MEME: 0.23, REL Bayes factor: 346.4). Site 57 coded for two different residues (arginine or serine) but was conserved in bushpig haplotypes (Supplementary Fig. S12). BAG6 was similarly conserved ( Supplementary  Fig. S13); when all haplotypes were considered, we detected one site evolving under purifying selection (26I: FEL As all Malagasy samples formed a well-supported monophyletic clade ( Fig. 3 and Supplementary Fig. S9), our results do not support the idea that bushpigs in Madagascar originated from multiple African source populations. It remains possible, however, that additional haplotypes could have been identified with a greater geographical www.nature.com/scientificreports/ sampling, particularly in the far north (Antsiranana Province) and also the Comoro Islands (i.e., Grande Comore and Mayotte 9,101 )-but evidence of bushpigs on this island are unverified reports and may have been mistaken for feral pigs 101 . BEAST analysis suggests that Malagasy bushpigs diverged from an African source population, that was closest to bushpigs from Zimbabwe, approximately 480,000 years ago. These populations are distinct from those other Zimbabwe specimens and those in KwaZulu-Natal, suggesting the existence of separate phylogeographic groups before their translocation into Madagascar. The estimates does not exceed the time to most recent common ancestor of Potamochoerus (approximately 2.71 Ma) 67 . The time of divergence between Malagasy and Zimbabwe bushpigs dramatically precedes that of any conclusive evidence of human occupation on Madagascar. Given that there is no evidence of bushpigs in Madagascar during the Holocene palaeontological or in pre-tenth-century archaeological records, the presence of bushpigs in Madagascar would have likely involved human intervention unless future fossil discoveries suggest otherwise. This early divergence indicates  www.nature.com/scientificreports/ www.nature.com/scientificreports/ that the Zimbabwe population does not represent the source population of Malagasy per-se. Altogether, our results indicate that to identify the source population of Malagasy bushpigs, future studies should collect additional samples from locations such as Mozambique, southern Tanzania, and offshore islands-specifically the Comoros, Mafia, and Zanzibar. The study showed that mitochondrial DNA was useful in inferring the relationships of Malagasy bushpigs but this was not the case for the nuclear DNA GPIP and MC1R. Both the phylogenetic topologies produced for GPIP and MC1R were less resolved. The MC1R gene was useful in discerning coat colour for breeds of domestic pigs but not in bushpigs, however further studies identifying other melanin producing genes may be suitable for similar studies 50,59,102,103 . The usage of GPIP was also not informative for distinguishing between bushpig populations as for domestic pig breeds which was valuable in supporting information about domestication 48,49 . Sequencing of additional mitochondrial genes could also provide a more comprehensive insight into the divergence of bushpigs from Madagascar and the mainland.
Immunogenetic diversity of bushpigs from Madagascar. Our results show that Malagasy bushpigs have maintained similar levels of and unique genetic diversity within the important MHC class II SLA-DQB1 ABS through diversifying selection when compared to mainland animals (Fig. 8). This suggests that founder effects, such as genetic drift and founder population size 12,28,104 , had minimal effect on the SLA-DQB1 ABS and the adaptation to environmental challenges after the introduction to Madagascar may have driven the rise of new alleles and maintenance of genetic diversity. Most samples here are also from the tropical zones of Madagascar in contrast to the arid and temperate zones of Zimbabwe and southern Africa 105 . It is possible that the prevalence, abundance, and infection of novel parasites, driven by specific precipitation and temperature in Madagascar influenced MHC allele frequencies as has been suggested for other species 87,106,107 . In comparison, the retention of ancestral SLA-DQB1 ABS amongst Malagasy and mainland bushpigs may indicate the necessity to combat common pathogens such as as Taenia solium 108,109 , Trichinella 110 , and Burkloderia 111 . However, detailed studies of bushpigs are still limited, particularly due to their elusive behaviour. Investigations into parasitic infections in both the mainland and Madagascar, along multiple habitat ranges, and pathogen-associated studies could provide in-depth insights into the roles of specific haplotypes of MHC genes 106,112,113 .
In contrast to the above, the conserved SLA-6 haplotypes found in some of these species when observing amino acid translations (SLA-6-1, -4, and -7 were found in five genera: P. larvatus, P. africanus, H. meinertzahgeni, S. scrofa [wild boar], and B. babyrussa; and the bushpig specimen from Duisburg Zoo was identical to S. celebensis) are expected as this gene is often described as monomorphic 114 . SLA-6 has a wide tissue-expression but is the least overall transcribed non-classical genes 114 with potential roles as a membrane-anchored glycoprotein and may have roles specific to each species 115 . The lack of genetic diversity of the SLA-6 exon-2 region, coding for the alpha-1 domain, provides little support for species-specific functions and additional sequencing of the alpha-2 domain may provide more information. There is also potential for the SLA non-classical genes to have similar functions to those described in humans (HLA-E, HLA-F, and HLA-G) such as cell population regulation (e.g. impair T cell proliferation and cytotoxicity 116 , apoptosis induction 117 ), and possibly immunomodulation during pregnancy 118 .
The class III BAG6 gene was highly conserved and indicates a retention of these haplotypes across Suidae species due to similar functions, and the low diversity found should not be due to founder effects. The Class III BAG6 gene has been shown to play a role in the presentation of MHC class I on the cell surface and protects against cell death by metabolising defective ribosomal products 119 . The importance of this gene in African swine fever infection has been implicated in the ability of viral genes to exploit host genes to evade host immunity 60 . African swine fever is known to asymptomatically infect bushpigs but can cause high mortality rates in species of the genus Sus 120 . Due to the high conservation between the sequences, it is highly likely that the role of BAG6 in ASFV resistant is small. Identical amino acid sequences in some haplotypes (BAG6-1, -3, -5, -6) only present in sub-Saharan African species, are distinct from those found in Sus. This may suggest functions that are different between genera. African swine fever infection in other species such as B. babyrussa and peccaries have been poorly studied and are suggested to not be affected by ASFV 121,122 . It is possible that mutations in genes can reduce activity or interfere with ASFV production 60 and the role of BAG6 in ASFV resistance should be studied in the context of other Suidae and Tayassuidae species as well.

Conclusion
The current study provided genetic evidence that the bushpigs samples obtained in Madagascar are Potamochoerus larvatus which originated from mainland Africa and there was no genetic differentiation between eastern and western Malagasy bushpigs as suggested by previous morphological work. The introduction of bushpigs in Madagascar was associated with few maternal lineages and based on the current sampling seem closest to the sampled natural populations of Zimbabwe, but this conclusion is preliminary until more detailed phylogeographic data is available from other areas of eastern and south-eastern Africa, including southern Tanzania and Mozambique. The MHC sequences examined also showed differences between bushpigs from mainland Africa and Madagascar, with few haplotypes of the polymorphic SLA-DQB1 shared between the two groups, suggesting they are undergoing diversification from their mainland ancestors. This also indicates that the introduction and isolation of bushpigs in Madagascar did not severely reduce their immunogenetic repertoire and appear to facilitate adaptation to new environments. Additional sequencing of specimens from other locations around Madagascar, as well as eastern Africa, is needed to provide a more in-depth indication into the source population of bushpigs introduced to Madagascar.