Conservation Genetics

, Volume 12, Issue 6, pp 1619–1631

Genetic consequences of isolation: island tammar wallaby (Macropus eugenii) populations and the conservation of threatened species


    • School of Biological, Earth and Environmental SciencesUniversity of New South Wales
    • School of Biological SciencesThe University of Sydney
  • Mark D. B. Eldridge
    • Department of Molecular BiologyAustralian Museum
  • Keith D. Morris
    • Department of Environment and ConservationWestern Australian Wildlife Research Centre
  • Kyall R. Zenger
    • School of Marine and Tropical BiologyJames Cook University
  • Catherine A. Herbert
    • School of Biological, Earth and Environmental SciencesUniversity of New South Wales
    • Faculty of Veterinary ScienceUniversity of Sydney
Research Article

DOI: 10.1007/s10592-011-0265-2

Cite this article as:
Miller, E.J., Eldridge, M.D.B., Morris, K.D. et al. Conserv Genet (2011) 12: 1619. doi:10.1007/s10592-011-0265-2


Isolation and restricted gene flow can lead to genetic deterioration in populations. Populations of many species are increasingly becoming fragmented due to human impacts and active management is required to prevent further extinctions. Islands provide an ideal location to protect species from many mainland threatening processes such as habitat loss and fragmentation, disease and competition/predation from introduced species. However their isolation and small population size renders them prone to loss of genetic diversity and to inbreeding. This study examined two endemic and one introduced population of tammar wallaby (Macropus eugenii) on three islands in the Houtman Abrolhos Archipelago, Western Australia: East Wallabi (EWI), West Wallabi (WWI) and North Islands (NI). Nine autosomal and four Y-linked microsatellite loci, and sequence data from the mitochondrial DNA (mtDNA) control region were used to examine the impact of long-term isolation (EWI and WWI) and small founder size (NI) on genetic diversity and inbreeding. This study found all three populations had low genetic diversity, high levels of effective inbreeding and increased frequency of morphological abnormalities. Isolation has also led to significant inter-population genetic differentiation. These results highlight the importance of incorporating genetic management strategies when utilising islands as refuges for declining mainland populations.


Conservation geneticsPopulation structureGenetic differentiationGenetic diversityInbreedingMicrosatellite


Increasing habitat fragmentation and population declines pose a major threat to global biodiversity (Young et al. 1996). Island populations are typically more predisposed than mainland populations to the effects of demographic and environmental stochasticity, and significantly more susceptible to the loss of genetic diversity due to random genetic drift and genetic bottlenecks, as well as inbreeding depression (Frankham 1997, 1998; Eldridge et al. 1999). Small, isolated mainland populations face similar risks (Frankham et al. 2002). Inbreeding and loss of genetic diversity are of particular conservation concern because they not only reduce fitness (Seymour et al. 2001) and a species’ ability to adapt and evolve to change, but increase the risk of extinction (Frankham et al. 2002). The effects of genetic deterioration associated with inbreeding are known in both captive and wild populations (DeRose and Roff 1999; Hedrick and Kalinowski 2000), but have been poorly documented in island populations (Eldridge et al.1999). Detailed studies can reveal the full cost of inbreeding depressions, but these are not always feasible. However there is evidence that inbred populations can sometimes manifest physical abnormalities such as cowlicks, kinked tails, cryptorchidism and poor sperm viability (Roelke et al. 1993; Hedrick 1995; Madsen et al. 1999; Seymour et al. 2001; Sunquist and Sunquist 2001).

Studying terrestrial island populations makes an important contribution to managing fragmented mainland populations because of the many parallels between the two, such as small population size and restricted gene flow. Sea level rises during the past 6,000–15,000 years have lead to the formation of many continental islands around Australia (Main 1961), which have protected many species from threatening processes occurring on the mainland (Eldridge et al. 2004). However in the long term, island populations are highly vulnerable to extinction (Frankham et al. 2002), and reasons for this are still debated (Jamieson 2007). Many threatened species are either now confined to islands, or have island populations of high conservation significance (e.g. Abbott 2000). In addition, islands are increasingly being utilised, both in Australia and internationally, as wildlife refuges for threatened mainland species. For example, Gilbert’s potoroo (Potorous gilbertii), one of the world’s most endangered mammals has been introduced to Bald Island, Western Australia (WA) (Courtenay and Friend 2003); northern quolls (Dasyurus hallucatus) have been introduced to Pobassoo and Astell Islands in the Northern Territory, Australia, to protect them from the spread of the introduced cane toad (Chaunus (Bufo) marinus: Rankmore et al. 2008); South Island robins (Petroica australis australis) have been translocated to several offshore islands in New Zealand to establish populations protected from mammalian predators (Boessenkool et al. 2007); while several bird and mammal species are currently being translocated from Barrow Island, WA to other islands and multiple mainland sites (Stanley et al. 2010). However, the vulnerability of island populations to extinction may ultimately limit the suitability of islands as wildlife refuges.

The tammar wallaby (Macropus eugenii) is a medium-sized (5–10 kg), herbivorous macropodid marsupial with a fragmented distribution (Fig. 1) comprising several disjunct populations surviving in mainland southwest WA and on several offshore islands including the Abrolhos and Recherché Archipelagos, Garden Island (WA) and Kangaroo Island, South Australia (SA). The tammar wallaby is listed as ‘Low Risk (near threatened)’ (IUCN 2006), however there is a marked difference between the conservation status of different populations. In SA, they are extinct on four of five islands (Robinson et al. 1996), as well as the mainland and are being captive bred for reintroduction (DEH 2004). On mainland WA, the tammar wallaby was de-listed from the State Threatened Fauna List as a result of their recovery under the Western Shield conservation program which controls introduced predators (Kinnear et al. 2002; Possingham et al. 2003). In contrast to mainland populations sustained through active management, on several offshore islands they have become overabundant, for example on Kangaroo Island (SA) tammar wallabies are culled each year (Wright and Stott 1999).
Fig. 1

Present (shaded black) and former (shaded grey) distribution of the tammar wallaby (Macropus eugenii) (after McKenzie et al. 2006) A Houtman Abrolhos Archipelago; B Tutanning Nature Reserve, Western Australia (WA); C Garden Island, WA; D North Twin Peak Island, WA; E Middle Island, WA; F Kangaroo Island, South Australia; and inset the geographic relationship of East Wallabi (EWI), West Wallabi (WWI), and North Islands (NI), within the Wallabi Group of the Houtman Abrolhos Archipelago

The Houtman Abrolhos Archipelago (hereafter referred to as the Abrolhos) lies 60 km offshore from Geraldton (WA). Within the Abrolhos, the three largest islands of the Wallabi Group (Fig. 1) are the focus of this study. West Wallabi Island (WWI) is the largest of the three (587 ha) and became separated from the adjacent East Wallabi Island (EWI: 307 ha) 6,000 years ago (Peltier 2004). EWI and WWI are continental islands and tammar wallabies became isolated from the mainland following rising sea levels 11,500 years ago at the end of the last glacial maximum (Alexander 1922; Storr 1965). They are geographically close (1.9 km apart) (GeoscienceAustralia 2005) and since much of this distance is exposed at low tide (BOM 2008) it is believed that wallabies can travel between them (Cooper and McKenzie 1997). The third study population on North Island (NI: 176 ha) is more isolated, lying 14 km northwest of EWI and WWI (Abbott and Burbridge 1995; GeoscienceAustralia 2005). NI is a more recently formed sand island and the resident tammar wallaby population must have either naturally dispersed from EWI and/or WWI, or been introduced. The first recorded observation of a tammar wallaby on NI was in 1928 following their introduction (source population unknown) as a potential food source for stranded fisherman (Alexander 1922; Storr 1960; Poole et al. 1991). This introduction was thought to be unsuccessful and the population was subsequently reported extinct (Storr 1960). More recently, there have been at least two further attempts to establish tammars on NI (Poole et al. 1991) using animals sourced from EWI (C. Herbert, pers. comm.). The first was in 1983 (n = 3), but no animals were subsequently observed and the population was presumed extinct, so in 1985, five more animals were introduced (C. Herbert, pers. comm.). The sex ratio and reproductive status of the individuals introduced to NI is unknown. The latest introduction appears to have been successful with a flourishing tammar population present on NI by 2002. The population has continued to increase aided by human-mediated modifications to the landscape, with increasing concerns about overabundance and damage to the vegetation resulting in a cull of ~800 animals in 2008 (A. Desmond, pers. comm.). The Abrolhos Island populations are a good model to examine island refuges as they have been naturally subdivided into discrete populations (EWI and WWI), and NI represents a case study of the effects of establishing a population from an small, isolated population to an island environment.

A previous study by Eldridge et al. (2004) found the endemic Garden Island population of tammar wallabies had significantly lower genetic diversity than mainland WA tammars. The current study is the first to examine the three Abrolhos Island tammar populations (two endemic and one introduced), and will test whether the previously reported patterns are also found in other tammar populations. The specific aims of this study were to (i) assess the impact of long-term isolation and small population size on genetic diversity in three island populations, (ii) determine whether there has been genetic deterioration (such as morphological abnormalities) within each population, (iii) confirm the origin of the NI population, (iv) determine whether the populations within the Wallabi Group are of particular genetic or conservation significance, and (v) examine whether endemic and/or introduced island populations are suitable for the conservation of species from threatening processes on the mainland.

Materials and methods

Sample collection and DNA extraction

Individuals from each island were trapped and sampled (EWI n = 35, males = 20, females = 15; WWI n = 30, males = 16, females = 14; NI n = 36, males = 19, females = 17; Tutanning Nature Reserve (mainland) n = 10, males = 3, females = 7). Trapping took place between 2006 and 2008 using a combination of wire cage (380 × 380 × 760 mm) and mesh Thomas traps (N. Thomas, pers. comm.). All animals were identified by a unique alpha-numeric code (subcutaneous microchip; Allflex®, NSW). A range of biological data was collected from each animal including morphological measurements, reproductive status, and the presence of morphological abnormalities including tail kinks and testicular abnormalities. A small ear biopsy (2 mm) was collected and stored in 80% ethanol prior to DNA extraction using the standard salting out method (Sunnucks and Hales 1996).

DNA amplification and screening

All sampled individuals were genotyped using ten polymorphic autosomal microsatellite loci characterised from the tammar wallaby (T19-1, T46-5, T31-1, T3-1T, Me2 and Me14) and eastern grey kangaroo (Macropus giganteus) (G16-1, G26-4, G31-1, G20-2) (Taylor and Cooper 1998; Zenger and Cooper 2001; Zenger et al. 2003a). Genotyping was carried out using two multiplex combinations (i) G31-1, T46-5, T31-1, G16-1, T3-1T and (ii) G26-4, G20-2, Me2, Me14, T19-1 as previously described (Miller et al. 2010). Males were also genotyped at five Y-linked microsatellite loci, MeY01, MeY27, MeY28, MeY37A and MeY37B (MacDonald et al.2006), using 10 μl reactions (as described above) and modified PCR conditions, whereby the temperature range used was a ‘touchdown’ cycle that decreased in 1°C decrements between 64 and 59°C, per cycle. These paternally inherited markers were used to examine male-specific population histories. All PCR products were analysed in a 48 capillary AB 3730 DNA Analyser (Applied Biosystems, USA). The resultant DNA fragments were sized and quantified using GeneMapper 3.7 (Applied Biosystems, USA).

The mitochondrial DNA (mtDNA) control region [~700 base pairs (bp)] was amplified using marsupial-specific primers, L15899M and H16498M and PCR reactions were performed using previously described conditions (Fumagalli et al. 1997). The PCR products were then cleaned using the ExoSAP-IT (USB, USA) protocol, followed by a Cycle Sequencing Big Dye Reaction (ABI PRISM®, USA) to enable fluorescence-based cycle sequencing reactions. The PCR products were analysed in a 48 capillary AB 3730 DNA Analyser (Applied Biosystems, USA). The mtDNA fragments were assembled using Sequencher 4.8 (Gene Codes Corporation, USA).

Estimates of diversity, inbreeding and population bottlenecks

For the autosomal microsatellites, locus independence and Hardy–Weinberg equilibrium tests were conducted using GenePop 3.4 (Raymond and Rousset 2003) via a Markov chain method (5,000 iterations). The statistical significance levels were corrected for multiple comparisons using sequential Bonferroni adjustments (Rice 1989). Genetic diversity was estimated for each island population by calculating the allelic diversity (AD), observed (Ho) and expected (He) heterozygosities using GENALEX 6.0 (Peakall and Smouse 2006).

For the control region sequence data, ARLEQUIN 3.11 (Excoffier et al. 2005) was used to calculate haplotypic diversity (h), nucleotide diversity (π) and to assess population differentiation with the fixation index ΦST, an estimator that includes information on haplotypic frequency and molecular distance using a pairwise distance method with a gamma distribution of 0.5. An analysis of molecular variance (AMOVA) was conducted in ARLEQUIN 3.11 to assess the population differentiation using haplotypes between two populations: EWI, and a combined WWI and NI since NI shared 100% of alleles with WWI.

The differences in AD, Ho and He among populations were assessed using a Wilcoxon signed rank (WRS) test. All statistical analyses were conducted in SPSS 15.0. The Y-linked microsatellites were characterised by the presence/absence of alleles and haplotypes in each population since there was insufficient haplotypic diversity to analyse statistically. The mainland WA population was not typed for the Y-linked microsatellites as the majority of the individuals sampled were female (n = 7).

The effective inbreeding coefficient (Fe, Wright’s fixation index) was calculated from the equation:
$$ F_{\text{e}} = \, 1-H_{\text{IS}} /H_{\text{M}} $$
where HIS represents heterozygosity for island populations and HM represents heterozygosity for mainland populations (Frankham 1998), and a one-sample t-test was used to test whether Fe for each population differed significantly from zero.

BOTTLENECK 1.2.02 (Cornuet and Luikart 1996) was used to test whether a genetic bottleneck has occurred in any of the populations. A Wilcoxon test was used as recommended for relatively low number of loci, using the stepwise mutation model (SMM) and under the two-phase model (TPM) and, with a 95% single-step mutations and 5% multiple step mutations, and a variance among multiple steps of 12 as recommended by (Piry et al. 1999).

Population structure and gene flow

Estimates of microsatellite genetic differentiation (FST) were calculated using FSTAT (Goudet 2001) and significance was tested after 10,000 permutations using the Weir and Cockerham (1984) method. Population subdivision based on mtDNA was also tested in ARLEQUIN 3.11 using three estimates of variance: among groups, among populations within groups, and within populations (Excoffier et al. 2005).

To examine the presence of population structure, gene flow and assign a source population for NI, a Bayesian model-based clustering method was used within the program STRUCTURE 2.2 using autosomal microsatellite data (Pritchard et al. 2000). An admixture model was used with the alpha to be inferred from the data. Lambda, the parameter of the distribution of allelic frequency was set to one. The burn-in length was 100,000 iterations and the MCMC (Markov chain Monte Carlo) was set to 1,000,000. The range of possible populations (K) tested was from one to nine. The true number of populations (K) was identified using two methods: (i) K was identified using the maximal value of the estimated log probability of the data, Ln P(D), at each step of the MCMC, generated by STRUCTURE (Pritchard et al. 2000); (ii) K was estimated using an ad hoc statistic ΔK based on the rate of change in the log probability of data between successive K values (Evanno et al. 2005). The latter method was implemented in addition to the traditional STRUCTURE method as a comparison since Evanno et al. (2005) found it provided a more accurate estimation of the number of clusters, K.

Phylogenetic analysis

The mtDNA control region sequences were aligned using ClustalX (Thompson et al. 1997). To place the Abrolhos tammar populations in a broader intra-specific context, additional tammar control region haplotypes (Zenger, unpublished data) from mainland WA (Tutanning n = 10; Perup n = 2) and Kangaroo Island, SA (n = 2) were also included. These, along with the Abrolhos sequences, were assembled and trimmed to 595 bp using Sequencher 4.8 (Gene Codes Corporation). Phylogenetic trees of the mtDNA fragments were reconstructed using Neighbour-Joining (NJ), Maximum Parsimony (MP) and Maximum Likelihood (ML) methods in PAUP* 4.0 (Swofford 2002). Homologous sequences from an eastern grey kangaroo (M. giganteus) and western grey kangaroo (Macropus fuliginosus) (GenBank accession numbers: AF443160 and AF443174, respectively), were used as an outgroup. To determine the most appropriate model for these data, 56 tests of evolutionary models were carried out in PAUP* 4.0 and analysed using MODELTEST 3.7 (Posada and Crandall 1998). The recommended model was the HKY model (Hasegawa et al. 1985) with relative frequencies A (0.3943), C (0.2569), G (0.0662) and T (0.2826) and –ln L = 799.2745. Rate variation across sites was modelled using a gamma distribution and the proportion of sites was modelled as invariant. The NJ analysis used an uncorrected P, and gamma distribution = 0.5 for 10,000 replicates. The MP analysis used a Branch-and-Bound search. The HKY model was used for the ML analysis in PAUP* 4.0 using 1,000 replicates. For the Bayesian analysis, the HKY model was also used in MRBAYES 3.1.2 (Huelsenbeck and Ronquist 2001), and run for 10,000 generations with four chains, sampling trees at every 100 generations, with a burn-in of 1,000. A 50% majority consensus tree was obtained from the last 9,000 trees. Heuristic searches were conducted for bootstrapping and the robustness of each branch was evaluated with 10,000 replicates in all the Bayesian, MP and ML analyses.


Amongst the 101 Abrolhos tammars there was a high prevalence of morphological abnormalities observed in individuals from each island population. Kinked tails (Fig. 2a, b) were observed in all populations: 60% EWI (21 of 35 animals); 46.6% WWI (14 of 30 animals) and NI 57.1% (16 of 28 animals). A small proportion of males on NI 3.4% (3 of 89) exhibited unilateral testicular aplasia whereby only one testis was present in the scrotum (Fig. 2c, d). In contrast, no evidence of tail kinks or testicular abnormalities were observed in tammars from Tutanning (n = 10, males = 3, females = 7), nor from Kangaroo Island, South Australia (n = 72, male = 32, females = 40) (unpublished data).
Fig. 2

Evidence of morphological abnormalities in the Abrolhos Island populations of tammar wallabies (Macropus eugenii) showing: evidence of tail kinks (a, b), on the right two individuals showing tail kinks, on the left a normal straight tail (a), and a close up of kinked tail (b); and, variations in testicular morphology in sexually mature males, evidence of unilateral testicular aplasia (c) on North Island, Western Australia, and a male with normal testicles from Kangaroo Island, South Australia (d)

Estimates of diversity, inbreeding and population bottlenecks

Nine of the ten autosomal microsatellite loci amplified successfully. One locus (T19-1) could not be amplified consistently for all individuals and was excluded from further analysis. EWI, WWI and Tutanning populations were polymorphic for all autosomal microsatellite loci however NI was monomorphic for one locus (G20-2). All populations were in Hardy–Weinberg equilibrium at all loci (P > 0.05). The mainland Tutanning population had significantly higher AD, Ho and He than all three Abrolhos populations (Table 1; AD: WRS z = −2.023 to −0.041, P = 0.041–0.043; Ho: WRS z = −1.761 to −0.826, P = 0.040–0.048; He: WRS z = −2.023, P = 0.043). Within the Abrolhos, genetic diversity (AD, Ho, and He) was highest on WWI, and lowest on NI (Table 1). There were no significant differences in Ho and He among the Abrolhos populations (P > 0.05). NI had significantly lower AD than both EWI (WRS z = −2.271, P = 0.023) and WWI (WRS z = −2.220, P = 0.026). Across the nine autosomal microsatellite loci, WWI had eight private alleles and EWI had 17 (Table 1). NI had no private alleles; all were shared with WWI and tended to be the higher frequency alleles (Table S1). All three Abrolhos populations were significantly inbred. Fe was significantly different from zero for all populations. Fe was highest in NI (Fe = 0.55; P = 0.02) and EWI (Fe = 0.55; P = 0.04), with WWI having the lowest (Fe = 0.38; P = 0.01). The Wilcoxon test in BOTTLENECK detected no evidence of a genetic bottleneck under the TPM or SMM models, in any population. There was also no evidence of a mode shift (Table 2).
Table 1

Summary of autosomal microsatellite and haplotypic mtDNA control region haplotypic diversity indices for the East Wallabi Island (EWI), West Wallabi Island (WWI), North Island (NI) and mainland Western Australian tammar wallaby (Macropus eugenii) populations





Microsatellite diversity

mtDNA haplotypic diversity





No. haplotypes

h (±SD)

π (±SD)

Mean ± SE


Mean ± SE


Mean ± SE






3.7 (±0.6)


0.38 (±0.10)


0.37 (±0.10)




0.23 (±0.09)

0.35 (±0.39)





4.1 (±0.7)


0.49 (±0.10)


0.51 (±0.09)




0.74 (±0.04)

2.76 (±1.67)





2.8 (±0.5)


0.39 (±0.09)


0.37 (±0.09)




0.50 (±0.04)

0.99 (±0.77)

Mainland (Tutanning)




10 (±1.6)


0.71 (±0.10)


0.82 (±0.04)




4.42 (±3.79)

15.98 (±8.82)

n sample size, Na number of alleles, AD allelic diversity, Ho observed heterozygosity, He expected heterozygosity, uA proportion of unique alleles, h haplotypic diversity, π nucleotide diversity

*One shared haplotype

Table 2

Pairwise genetic differentiation amongst the Abrolhos tammar wallaby (Macropus eugenii) populations

















Pairwise comparison: FST above and ΦST below the diagonal

EWI East Wallabi Island, WWI West Wallabi Island, NI North Island

P = 0.008 for each comparison

** P < 0.0000 for each comparison

A total of 55 males were genotyped at four Y-linked microsatellite loci (MeY27, failed to amplify so was excluded). Two loci were polymorphic (MeY01 and MeY28) and two were monomorphic (MeY37A and MeY37B). A total of seven alleles and three haplotypes were identified across all three populations (Tables 3, S2). EWI had one unique haplotype; WWI had two haplotypes, one of which was shared with NI (Table 3).
Table 3

Distribution of mtDNA control region and Y-linked microsatellite haplotypes across the three Abrolhos Island tammar wallaby (Macropus eugenii) populations


mtDNA haplotypes

Y-linked haplotypes




























EWI East Wallabi Island, WWI West Wallabi Island, NI North Island

Within the 92 Abrolhos individuals examined for mtDNA control region diversity, eight haplotypes were identified (Table 3), (nine individuals failed to amplify). Tammar wallabies from EWI had three haplotypes none of which were shared with WWI. NI had the lowest number of haplotypes (F and H), one of which was shared with WWI (haplotype F), the other, haplotype H, was unique to NI (Table 3). These Abrolhos control region haplotypes were distinguished by nine variable sites. Higher haplotypic diversity and nucleotide diversity was detected in WWI and NI, than in EWI (Table 1).

Population structure and gene flow

All three island populations were significantly differentiated from one another (FST range = 0.118–0.520; P = 0.008; Table 2) and from the mainland population (P = 0.008). The STRUCTURE analysis indicated there were two populations (ΔK = 2; K = 2). The assignment probabilities showed evidence of population structuring within the three island populations. There was no evidence of recent gene flow/admixture between EWI and WWI (Fig. 3), while WWI and NI tammars were classed as a single population. The STRUCTURE AMOVA analysis repeated for WWI and NI alone revealed no further population subdivision. There was significant divergence between populations compared with the molecular variation within populations (ΦST = 0.660, P = 0.00000). Fifty percent of the variation occurred between the two groups (EWI, and combined WWI and NI), 16% within the groups and 34% occurred within populations.
Fig. 3

The level of admixture among tammar wallabies (Macropus eugenii) in the three Wallabi Group Islands. Each vertical bar represents a single individual. Individuals were assigned to a population using the program STRUCTURE (Pritchard et al. 2000). The population number, K, was estimated using no prior population information

Phylogenetic analysis

The reconstruction of the relationships amongst EWI, WWI and NI mtDNA haplotypes using NJ, MP and ML methods produced trees with similar topologies and bootstrap support for the resolved groups (Fig. 4). The ML analysis resulted in a phylogenetic tree that distinguished four main groups [Abrolhos, mainland WA, SA (KI) and the grey kangaroos] all with strong bootstrap support (≥96% in ML) and Bayesian credibility values (≥0.70). Within tammars, there was strong support (100%) for two major clades consisting of tammars from WA and SA respectively. There was also strong (96%) support for the mtDNA from Abrolhos Island tammars being monophyletic with respect to mainland WA tammars, but not reciprocally monophyletic. The haplotypes from mainland WA tammars generally exhibited an unresolved polytomy, although this was resolved in the MP tree. Sequence divergence (mean ± se) within the Abrolhos was minimal 0.54 ± 0.07% but more extensive between the Abrolhos and mainland WA (4.3 ± 0.06%). Sequence divergence amongst WA mainland haplotypes averaged 3.6 ± 0.26%. Within tammars the largest divergence was between the Abrolhos and SA (11.3 ± 0.05%). As expected, sequence divergence between the tammar wallaby and grey kangaroo haplotypes was also high (17.9 ± 0.1%).
Fig. 4

Maximum likelihood (ML) tree of relationship amongst Abrolhos, mainland WA and SA tammar wallaby (Macropus eugenii) mtDNA control region sequences (595 bp). Robustness is indicated by bootstrap values (≥50%, above branches) and Bayesian posterior probabilities (≥0.50, below branches)


Island populations are increasingly being utilised for threatened species conservation and make an important contribution to preventing extinctions (Abbott 2000). However, if island populations are not managed to minimise the impact of low genetic diversity, high levels of inbreeding and genetic deterioration, these efforts may be counterproductive. This study has demonstrated that some island populations are genetically compromised and that the regular exchange of individuals amongst isolated populations will be an important strategy for retaining genetic diversity and minimising the effects of inbreeding that can lead to genetic deterioration. The mtDNA and microsatellite data revealed that all three island populations were inbred, had significantly lower levels of genetic variation than their WA mainland counterpart, and were significantly differentiated from one another, as well as the mainland population. There was no evidence of recent gene flow between EWI and WWI, despite their close geographic proximity. The microsatellite and mtDNA data indicate that the NI population was most likely founded by WWI, contrary to local belief that they were founded by individuals sourced from EWI. Two main lineages were detected within tammar wallabies (SA and WA) following phylogenetic analysis of the mtDNA control region sequence data. Within the WA clade, the Abrolhos tammar wallabies were monophyletic suggesting that they should be considered as separate Management Units (MU).

Genetic diversity

Genetic diversity in the Abrolhos tammar populations was significantly lower than the adjacent WA mainland population, which is consistent with their long term isolation and limited population size. Within the endemic Abrolhos populations, the largest island (WWI) had the highest microsatellite and haplotypic diversity, and the smaller island (EWI), had the lowest. These data match theoretical predictions that (i) genetic variation within a species is related to population size; (ii) genetic variation is related to island size; and (iii) island populations have less genetic variation than mainland populations (Soulé 1976; Frankham 1997). Since no census data are available, it is assumed that the larger island supports a larger tammar population since more habitat is available. Although the sample size for the mainland WA population is relatively low compared to the Abrolhos populations, the levels of diversity detected were comparable to those previously reported (Eldridge et al. 2004) for a larger sample of this population using a different panel of microsatellite markers.

A similar pattern of significantly reduced diversity in island compared to mainland populations has also been found for another island tammar population (Eldridge et al.2004), other marsupials (Robinson et al. 1993; Sinclair 2001; Eldridge et al. 2004; Mills et al. 2004), as well as a variety of other taxa (Frankham 1997). This demonstrates that while the island populations may have been shielded from threatening processes occurring on the mainland, such as predation and habitat loss, they have nonetheless been considerably affected by the genetic and demographic processes intrinsic to island populations, such as loss of genetic diversity through isolation and genetic drift, making them inherently vulnerable to extinction.

Significant levels of effective inbreeding were detected within each island population, but there was no evidence of a recent genetic bottleneck. Inbreeding is a consequence of small population size and long-term isolation, and in this study, is associated with the high prevalence of kinked tails in animals on all three islands, and an increase in the proportion of males on NI with only one testis in the scrotum. These traits, are known to be associated with inbreeding (Frankham et al. 2002), which can impact many life history traits resulting in reduced fitness. For example, kinked tails, cowlicks, cryptorchidism, poor sperm viability and sterility have been documented in the Florida panther (Felis concolor coryi), which is suffering from reduced fitness due to fixation of detrimental alleles, and low genetic diversity (Roelke et al. 1993; Hedrick 1995; Sunquist and Sunquist 2001). Similarly, morphological abnormalities, unilateral testicular aplasia and associated sperm abnormalities, have also been observed in some introduced koala (Phascolarctos cinereus) populations in south-eastern Australia (Seymour et al. 2001). Despite the evidence of morphological abnormalities and high levels of inbreeding, the NI tammar population has reached high densities (>4.5 animals per hectare; A. Desmond, pers. comm.) in the last two decades. The success of tammars on NI can be attributed to their ability to occupy a vacant ecological niche, and exploit abundant resources, on an island that is devoid of natural terrestrial predators and with increased availability of water as a result of human activity. In such circumstances, even species with reduced fitness are likely to be successful, for example introduced koala populations in South Australia (Seymour et al. 2001).

Population structure and gene flow

There is no evidence for recent gene flow amongst populations within the Abrolhos, and all populations showed significant genetic differentiation at both microsatellite loci and mtDNA. Within the Abrolhos, EWI and WWI had the highest and WWI and NI showed the lowest level of differentiation. The lack of shared diversity between EWI and WWI indicates there has been no recent gene flow and each island has been evolving independently since their separation 6,000 years ago. This result was unexpected as EWI and WWI are separated by only 1.9 km of shallow ocean, much of which is exposed at low tide. This indicates that tammars are poor dispersers over water. Given the isolation and the absence of gene flow, these Abrolhos populations are expected to continue to lose variation due to genetic drift and diverge from one another, and the mainland.

NI founder and population genetics

A surprising result from this study was the relationship between tammars from WWI and NI. Despite the local belief that NI tammars originated from EWI, the genetic data indicate that the NI population is unequivocally derived from the WWI population. NI tammars shared 100% of their autosomal and Y-linked microsatellite alleles, as well as one mtDNA haplotype with WWI and 0% of their diversity with EWI. The second mtDNA haplotype found on NI (haplotype H) is most closely related to WWI haplotype G (one base pair difference) and so most likely represents a recent mutation on NI or a haplotype that occurs at low frequency on WWI and was not detected. This finding was important as management decisions based on incorrect information can lead to the inaccurate allocation of Evolutionarily Significant Units (ESU) and MUs.

If tammars do not disperse readily across the narrow sheltered gap (1.9 km) between EWI and WWI, it is inconceivable that they were able to disperse 14 km north-west through the open ocean to reach NI from WWI. Therefore the NI population has most likely been deliberately introduced. Although local oral history records two deliberate introductions of tammars to NI from EWI in the 1980s, the current population is clearly of WWI origin. So either the oral history is inaccurate or the animals introduced from EWI failed to establish and there has been a subsequent undocumented but highly successful introduction of animals from WWI to NI. It is possible that some animals of EWI origin or descent do exist on NI but at such low frequencies that they were not sampled. However, this is unlikely since over 100 NI individuals have now been genotyped (unpublished data) and no microsatellite alleles, or mtDNA haplotypes characteristic of EWI have been detected.

The NI tammar population shows evidence of a limited number of founders since it has significantly lower AD compared to its source population (WWI) and is monomorphic at one locus. As predicted by theory and demonstrated by these data, allelic diversity is more sensitive to founder effects than heterozygosity (Nei et al. 1975; Allendorf 1986). Establishing populations with a small number of individuals can result in a genetic bottleneck, changes in allele frequencies and a loss of diversity because often the source individuals represent only a fraction of the genetic variants from the source population (Nei et al. 1975). The negative impacts of small founder numbers and genetic bottlenecks have been documented in vertebrates (Tarr et al. 1998; Hedrick et al. 2001; Jones et al. 2004; Ramstad et al. 2004), invertebrates (Johnson 1988) and invasive species that have undergone sequential founding events (Cabe 1998; Dlugosch and Parker 2008; Ficetola et al. 2008). However no genetic signature of a bottleneck was detected in the NI population perhaps due to a Type II error (failing to detect a bottleneck when there was one), or possibly the population had made a rapid demographic recovery (Zenger et al.2003b). Detecting bottlenecks can also be difficult when using single samples opposed to temporally spaced samples (Cornuet and Luikart 1996; Williamson-Natesan 2005).

Phylogenetics and conservation significance

The phylogenetic analysis of Abrolhos, mainland WA and SA tammar wallaby mtDNA showed reciprocal monophyly between the WA and SA tammar lineages, which was well supported by bootstrap values. Within the WA clade, the Abrolhos tammar wallaby haplotypes were monophyletic. Although the mainland WA haplotypes are clustered together they do not exhibit complete reciprocal monophyly with the Abrolhos haplotypes. Since the mtDNA control region shows reciprocal monophyly between the WA and SA tammar wallaby populations, they should be classified as separate ESUs (Waples 1991; Moritz 1994, 1999). MUs are identified by significant differences in allele frequency distributions and significant divergence in their mtDNA, therefore the mainland WA, EWI and WWI/NI warrant being treated as separate MUs.

Existing data indicate that the tammar wallaby populations in SA and WA represent two major reservoirs of diversity within M. eugenii, and both should be conserved as they reflect the long term historical isolation of populations within the species. The separate MUs including EWI, and WWI/NI, and mainland WA should be managed in such a way as to maintain diversity within the WA tammar wallaby ESU. There are several management options for the Abrolhos tammar populations. NI is an introduced population that has become overabundant on the island and was subject to a culling operation in 2008. The loss of the NI population through extinction or eradication would not be catastrophic as the diversity present on NI is present in the WWI population. However, if either EWI or WWI were to go extinct it would represent the irreplaceable loss of unique populations that are significantly different from one another, and all other tammar populations. Since island populations are known to be prone to extinction (Frankham 1998), NI could provide a potential ‘backup’ for WWI. However, the current NI population is not particularly representative of the WWI population as it was founded by a small number of individuals. The NI population also has lower genetic diversity and higher levels of inbreeding compared to the other Abrolhos populations.

Since all the Abrolhos tammar populations are inbred, subdivided and have no gene flow, they may benefit from augmentation, that is, translocation of individuals from the larger mainland population into the smaller island populations (Moritz 1999). From a demographic and genetic perspective, the Abrolhos populations would benefit by increasing genetic diversity, reducing the impact of inbreeding and preventing further genetic deterioration (e.g. kinked tails). Recent studies have shown that these effects can be ameliorated through ‘genetic rescue’, that is, the introduction of genetically unrelated individuals into small, isolated populations (Hedrick 2001; Pimm et al. 2006; Hedrick and Fredrickson 2008). To restore genetic variation and remove some of the detrimental alleles in the Florida panther population, Texas cougars (Felis concolor stanleyana), a related subspecies, were translocated into the population (Hedrick 1995), and the resultant offspring have shown a reduced occurrence of kinked tails, cowlicks and cryptorchidism (Land et al. 2004). Augmenting the Abrolhos populations with mainland WA wallabies may similarly help overcome any deleterious traits related to inbreeding. But is this appropriate? The Abrolhos tammar wallabies are morphologically and reproductively different from the mainland (both WA and SA) populations (Herbert et al., unpublished data). The environmental conditions on the Abrolhos are so harsh (e.g. no fresh water) that a reciprocal augmentation may be unsuccessful as mainland tammar wallabies may be unable to survive, particularly with the potential future effects of climate change likely to increase aridity. Furthermore, it is not recommended that inbred or genetically depauperate populations be used to augment other populations in an effort to maximise genetic diversity as it can compromise their long-term evolutionary plasticity (Eldridge et al. 1999; Moritz 1999). Augmenting the Abrolhos tammar populations with mainland WA tammars also raises the issue of whether these island populations should be allowed to retain their genetic integrity or be introgressed to reduce their risk of extinction. Alternatively, NI could be augmented with individuals from WWI and EWI, transforming NI into a reservoir for a generic Abrolhos tammar population and thus maintaining the genetic integrity of both EWI and WWI.

Implication for using islands as wildlife refuges

This study has shown that endemic and introduced island populations can have reduced genetic diversity, increased inbreeding and associated deleterious effects. Such genetic deterioration is a likely contributor to the extinction proneness of island populations. While inbred populations may persist, and even do well, in their current environment, they are likely to have a reduced ability to adapt to change (Spielman et al. 2004). In addition, their isolation exacerbates the effects of small population size, restricted gene flow, random genetic drift and inbreeding (Frankham 1997, 1998; Eldridge et al. 1999). This is of particular concern as island populations are increasingly being used as source populations for conservation initiatives or as wildlife refuges and without supplementation the ongoing deterioration of these populations is inevitable. Currently, the genetic management of fragmented populations is one of the most important unresolved issues in conservation biology (Frankham 2009). If islands are to be used as a long-term refuge for threatened species, it is fundamental to incorporate active genetic management such as reestablishing gene flow. Establishing colonies on islands may protect them from deterministic threats on the mainland, but will increase the effect of stochastic processes such as random genetic drift, particularly if they are founded with a small number of individuals that may already be inbred due to population declines on the mainland. In such cases, regular exchange of individuals between populations to simulate natural gene flow will be essential to assist in the maintenance of genetic diversity, the reduction of inbreeding and increase the likelihood of long-term persistence.


This research was funded by an ARC Linkage Grant (LP0560344) and forms part of the Koala and Kangaroo Contraception Program. Special thanks to Neil Thomas, Brent Johnson, Peter Orell from DEC (WA) for field assistance; John Fitzharding; The Rat Patrol; Lee Ann Rollins (UNSW) and Greg Eldridge. All experimental work carried out was approved by the Department of Environment and Conservation, Western Australia under the approval numbers 10/2005 and 40/2007.

Supplementary material

10592_2011_265_MOESM1_ESM.doc (136 kb)
Supplementary material 1 (DOC 136 kb)

Copyright information

© Springer Science+Business Media B.V. 2011