Conservation Genetics

, Volume 12, Issue 1, pp 91–103

Reduced MHC class II diversity in island compared to mainland populations of the black-footed rock-wallaby (Petrogale lateralis lateralis)

Authors

    • Department of Biological SciencesMacquarie University
    • DNA Laboratory/Evolutionary Biology UnitAustralian Museum
  • Teena L. Browning
    • Department of Biological SciencesMacquarie University
    • DNA Laboratory/Evolutionary Biology UnitAustralian Museum
  • Mark D. B. Eldridge
    • Department of Biological SciencesMacquarie University
    • DNA Laboratory/Evolutionary Biology UnitAustralian Museum
Research Article

DOI: 10.1007/s10592-009-9993-y

Cite this article as:
Mason, R.A.B., Browning, T.L. & Eldridge, M.D.B. Conserv Genet (2011) 12: 91. doi:10.1007/s10592-009-9993-y

Abstract

Many animal populations that are endangered in mainland areas exist in stable island populations, which have the potential to act as an “ark” in case of mainland population declines. Previous studies have found neutral genetic variation in such species to be up to an order of magnitude lower in island compared to mainland populations. If low genetic variation is prevalent across fitness-related loci, this would reduce the effectiveness of island populations as a source of individuals to supplement declining mainland populations or re-establish extinct mainland populations. One such species, the black-footed rock-wallaby (Petrogale lateralis lateralis), exists within fragmented mainland populations and small island populations off Western Australia. We examined sequence variation in this species within a fitness-related locus under positive selection, the MHC class II DAB β1 locus. The mainland populations displayed greater levels of allelic diversity (4–7 alleles) than the island population, despite being small and isolated, and contained at least two DAB gene copies. The island population displayed low allelic diversity (2 alleles) and fewer alleles per individual in comparison to mainland populations, and probably possesses only one DAB gene copy. The patterns of DAB diversity suggested that the island population has a markedly lower level of genetic variation than the mainland populations, in concordance with results from microsatellites (genotyped in a previous study), but preserved unique alleles which were not found in mainland populations. Where possible, conservation actions should pool individuals from multiple populations, not only island populations, for translocation programs, and focus on preventing further declines in mainland populations.

Keywords

Rock-wallabyDAB exon 2Gene copy numberFitness-related genetic diversityPeptide binding regionWildlife translocation

Introduction

Since European settlement of the Australian mainland, 14% of native mammals have declined and 10% have become extinct due to habitat clearance, changes in fire regimes, and predation and competition from introduced mammals (Burbidge and McKenzie 1989). Many islands off the Australian coast harbor populations of mammals that have declined massively on the mainland (Burbidge 1999; Burbidge et al. 1997). Many of these islands are free from introduced mammals and other wide-scale disturbances, providing a secure habitat for native mammals. These islands are viewed as having a high conservation potential for two reasons. Firstly, the preservation of mammal species on islands can prevent total extinction of those species if they disappear from the mainland (Burbidge 1999). Secondly, groups of individuals from island populations can be used as founders to re-establish populations in mainland locations where they have disappeared (Christensen and Burrows 1994; Short et al. 1994).

However, due to their small size, island populations are in theory more prone to decimation due to random stochastic events (e.g. disease, cyclones), or to loss of genetic diversity (Frankham 1996; Frankham 1997). Island populations with reduced genetic diversity may be more likely to suffer extinction, and so are less reliable as reservoirs for the preservation of their species. This phenomenon may be exacerbated by the adaptation of island populations to their immediate environment. Low genetic diversity and local adaptation decrease the genetic evolutionary potential of a species, and thereby the value of a population to supplement another population, as it is less likely to survive in a new environment.

Many studies have evaluated neutral genetic diversity in island populations with potential conservation value (e.g. Eldridge et al. 1999, 2004; Mills et al. 2004; Wilson et al. 2008). Studies of four marsupial species documented levels of heterozygosity at microsatellite loci that are between a third and an order of magnitude lower in island populations compared to mainland populations (Eldridge et al. 2004; Mills et al. 2004). Meanwhile, small and fragmented mainland populations maintain levels of heterozyosity similar to that found in undisturbed wild populations (Eldridge et al. 1999). This pattern is mirrored at neutral loci in a wide variety of other organisms, where island populations have (on average) 29% lower allozyme genetic variation than corresponding mainland populations (Frankham 1997). These studies suggest that island populations are unsuitable to act as reservoirs to preserve their species or as sources of individuals for re-introduction (Eldridge et al. 2004; Mills et al. 2004).

However, genetic diversity at fitness-associated loci may not be as low in island populations as these studies imply. Estimates of genetic diversity based on neutral loci such as microsatellites (Jarne and Lagoda 1996) may not reflect diversity at fitness-related loci, which are under the influence of selection (Ford 2002), and at which genetic diversity has a functional significance. If it is strong enough, selection can prevent the erosion of variability at a locus by genetic drift (Frankham et al. 2002).

Major histocompatibility (MHC) loci have recently gained prominence as useful loci at which to examine fitness-related genetic diversity in wild populations due to their high polymorphism and their direct association with immune fitness (Bernatchez and Landry 2003). The MHC encodes antigen presentation molecules, as well as a number of other immune-related loci (Bodmer et al. 1978). Class II MHC molecules present antigens on the surface of professional Antigen Presenting Cells (APCs). These antigens are derived from pathogenic cells that have been phagocytised by APCs, and the presentation of antigenic peptides to T-cells by class II molecules is a vital step in inducing an immune response (Klein 1986). The peptide binding region (PBR) of each MHC molecule binds antigenic peptides that contain amino acid side chains at the appropriate positions to fit the topology of the PBR (Hammer et al. 1993; Marshall et al. 1994). Balancing selection has driven the diversification of the peptide binding region, so that, within many species, a vast diversity of MHC molecules exists, capable of binding antigens from many different pathogens (reviewed by Hughes and Yeager 1998). An examination of the MHC will enable us to gauge if the findings of low neutral genetic diversity on island populations (Frankham 1997; Eldridge et al. 2004; Mills et al. 2004) may also be valid at fitness related loci.

The black-footed rock-wallaby (Petrogale lateralis lateralis) inhabits outcrops of weathered and fissured rock in semi-arid and arid regions of Western Australia (Eldridge and Close 1995), which are crucial for shelter during high temperatures to reduce loss of body water (King and Bradshaw 2008), and the species can survive in very small areas of habitat, provided that rock piles exist (Main and Yadav 1971). The black-footed rock-wallaby has declined in the extent of its distribution on the mainland by over 90% since European settlement due to fox predation and land clearing. It exists in stable populations on pristine islands that are free from foxes (Fig. 1a) (Kinnear et al. 1988, 1998; Maxwell et al. 1996). Translocation of individuals to new areas is intended as part of the conservation management of this species (Davies et al. 2007); black-footed rock-wallabies have already been translocated from Wheat-belt populations to conservation lands near Perth (Mawson 2003) and from one inland location (Nangeen Hill) to another (Querekin) where the original population had become extinct (J. E. Kinnear, unpublished data, cited in Eldridge et al. 2001).
https://static-content.springer.com/image/art%3A10.1007%2Fs10592-009-9993-y/MediaObjects/10592_2009_9993_Fig1_HTML.gif
Fig. 1

a Distribution of the black-footed rock-wallaby in Western Australia. Grey shading: distribution at commencement of European settlement; black-dots: present day populations; large font: populations examined in this study; small font: populations not examined in this study; Italicized font: populations established from translocated black-footed rock-wallabies—Paruna Sanctuary, Avon Valley National Park and Walyunga National Park populations were established from both Mount Caroline and Querekin populations, the Querekin population was established from the Nangeen Hill population. N.P. National Park. Diagram based on Eldridge et al. (1994, 2001), Hall and Kinnear (1991), Mawson (2003). Outline of Australia and the Australian States is © copyright Commonwealth of Australia (Geoscience Australia) [2005]. b Black-footed rock-wallabies populations in the Wheat-belt of Western Australia. Large font: populations examined in this study; small font: populations not examined in this study. N.R. Nature Reserve. Diagram based on Kinnear et al. (1988) and Freegard and Orell (2005)

Complicating the management of the black-footed rock wallaby is the finding that neutral genetic diversity (assessed through microsatellites) in the pristine Barrow Island population represents one of the lowest levels of microsatellite diversity ever found in a mammal (Eldridge et al. 1999). In contrast, the fragmented mainland populations contain relatively high levels of microsatellite diversity. We will examine if this pattern of diversity also occurs at a fitness related locus, the class II MHC locus, DAB, which is found exclusively in marsupials (Belov et al. 2004). Specifically, we will ask:
  1. 1.

    What are the levels of DAB diversity in mainland populations compared with an island population of the black-footed rock-wallaby?

     
  2. 2.

    How does the pattern of DAB diversity across island and mainland populations compare to the pattern of diversity at microsatellite loci?

     

Materials and methods

Populations and sampling

Tissue samples (blood or ear biopsy) were collected from remnant black-footed rock-wallaby populations between 1995 and 1998. Locations used in this study (Fig. 1a, b) were Exmouth, Barrow Island, and three localities in the Western Australian Wheat-belt: Mount Caroline Nature Reserve, Tutakin (Gundaring Nature Reserve), and Nangeen Hill Nature Reserve (see Table 1 for sample sizes). These populations are distributed across the majority of the range of the black-footed rock wallaby (Fig. 1a). The populations within the Wheat-belt are within 7 km of one another (Fig. 1b), and whilst fox predation limits dispersal and gene flow amongst them, Wheat-belt black-footed rock-wallabies are still successfully dispersing on rare occasions (Eldridge et al. 2001; Freegard and Orell 2005). Fox predation has been reduced by 1080 sodium monofluoroacetate baiting in the Mount Caroline and Nangeen Hill populations since 1982 (Kinnear et al. 1988), and in the Tutakin population since the early 1990s (Eldridge et al. 2004). Population size on Barrow Island is difficult to measure due to trapping difficulties (Burbidge 2008), but was estimated at 116–154 animals in 1993 (Hall et al. 1993). Across all sites in the Wheat-belt, 426 animals where trapped in 1998 (J. E. Kinnear, unpublished data, cited in Eldridge et al. 2004), compared to a population size of 75–100 measured in a 1979–1982 survey (Kinnear et al. 1988). Between 1982 and 1990, due to fox baiting, the population census size at Mount Caroline increased from 13 wallabies to 50 and at Nangeen Hill increased from 29 to 116 (Kinnear et al. 1998). The Tutakin population size was 7–9 individuals for the period of 1980–1990, after which fox control of Tutakin was commenced (Kinnear et al. 1998). The Wheat-belt populations were completely homozygous at all 20 allozyme loci genotyped by Kinnear et al. (unpublished data, cited in Hopper et al. 1982) prior to the commencement of fox control.
Table 1

Sample size, allelic diversity, and mean number of alleles per individual for each population for DAB β1 and eleven microsatellite loci assessed by Eldridge et al. (1999)

Population

Barrow Is

Exmouth

Mt. Caroline

Nangeen Hill

Tutakin

Sample sizes

 DABβ1

24

11

17

9

15

 Microsatellites

29

15

32

30

19

Mean number of DABβ1 alleles per individual (range of alleles per individual in brackets)

1.29 (1–2)

2.9 (2–4)

3 (1–4)

2.1 (1–4)

3 (3)

Number of alleles per population

 DAB β1

2

7

4

6

4

 Microsatellitesa

1.2

3.4

3.3

2.6

3.9

Gene diversity (expected heterozygosity)

 DABβ1

0.395

 Microsatellitesa

0.053

0.620

0.508

0.404

0.604

aAveraged across all polymorphic microsatellite loci for each population

Molecular procedures

DNA extractions, prepared according to Sunnucks and Hales (1996), and were drawn from samples used in two previous studies (Eldridge et al. 1999, 2004). DNA extractions were deposited with the Australian Museum Tissue Collection, but had not yet been given accession numbers at the time of publication. We used primers that amplified the second exon (β1 domain) of the DAB MHC class II gene: DAB β1 42F (5′-AGT GTG ACT TTG TGA ACG GG) and DAB β1 253R (5′-TCT CGT AGT TGT GCC TGC AGT). These primers were designed by Browning (2009) based on an alignment of the DAB sequences available in Genbank from the red-necked wallaby (Macropus rufogriseus), the brushtail possum (Trichosurus vulpecula), the gray short-tailed opossum (Monodelphis domestica), and the tammar wallaby (Macropus eugenii).

Single-strand conformation polymorphism (SSCP) assays were used to detect different DAB β1 alleles in each individual. Amplification reactions contained 10 ng of template DNA, 1× QIAGEN™ buffer, 1× QIAGEN™ Q-Solution, 3.5 mM MgCl2, 320 μM of each dNTP, 0.4 μM of each primer, 0.5 U of QIAGEN Taq polymerase, and (for initial radioactive SSCP experiments) 0.05 μl of 33P labelled dATP. Amplification conditions were 94°C (5 min); 30 cycles of 94°C (30 s), 58–60°C (30 s), 72°C (30 s); 72°C (10 min). Amplicons were denatured then electrophoresed on an 8% non-denaturing acrylamide gel for 4 h 30 min at 15 W and 4°C. Radioactive SSCP gels were auto-radiographed according to Sunnucks et al. (2000). We later switched to using non-radioactive SSCP gels, which were stained with SybrGold (Molecular Probes) and photographed under UV light. Interpretation of SSCP banding patterns was aided by the occurrence of heteroduplex banding patterns in heterozygotes on the SSCP gels (Wallace 2002).

During SSCP, the forward and reverse strands of an allele appear as two bands of different electrophoretic mobilities on the SSCP gel. For each unique two-band allele pattern identified by SSCP, 1–7 individuals that contained that band pattern were sequenced. If an allele occurred in homozygous individuals (i.e. an individual that had only two bands), the DAB β1 fragment was amplified by PCR, purified, and sequenced directly. If an allele only occurred in heterozygous individuals (i.e. an individual that had 4 or more bands), the DAB β1 fragment was amplified by PCR, then cloned before sequencing.

Purification of PCR products was performed with the Ultraclean™ 15 DNA Purification Kit (MO BIO Laboratories Inc, Carlsbad, California), or ExoSAP-IT® (USB, Cleveland, Ohio). Products were cloned into the pCR®4-TOPO® vector, using the TOPO TA® Cloning Kit (Invitrogen, Australia). Plasmids and PCR products were sequenced using BigDye (version 3) chemistry on ABI 377 and ABI 3730 instrumentation (Applied Biosystems). Allele sequences were deposited in Genbank as the accession numbers EU919635–EU919649.

Statistical tests

Since more than one DAB gene copy co-amplified in the mainland populations we were unable to assign alleles to loci and so we could not analyse our data as genotypic data or calculate allele frequencies. We applied the approach of Ekblom et al. (2007) and Louiseau et al. (2009) for statistical analysis of an MHC gene where more than one gene copy is amplified, using the software ARLEQUIN 3.1 (Excoffier et al. 2005). Alleles were treated as haplotypic data: the DNA sequence of each allele was input in a haplotype list, and then for each population sample, the name of each sequence in the population and number of individuals containing that sequence were listed. We obtained the data for 10 microsatellite loci for each population from Eldridge et al. (1999, 2001), which was in the format of diploid genotypes. ARLEQUIN was used to calculate global and pair-wise estimates of FST for both DAB β1 and microsatellite data. Isolation by distance for DAB β1 and microsatellites was tested in the ISOLDE function of GENEPOP (Raymond and Rousset 1995), which calculated a one-tailed P-value for the Spearman Rank correlation co-efficient. The total number of DAB β1 alleles per individual was compared between the island and the mainland (data from all mainland populations combined) with a two-sample t-test assuming unequal variances in Microsoft EXCEL. Alignment of DAB β1 sequences and calculation of number of nucleotide and amino acid differences between alleles were performed in MEGA version 4 (Tamura et al. 2007). A neighbour-joining phylogeny of the nucleotide sequences was constructed in PAUP* version 4.0 b10 (Swofford 2002) using Kimura-2-parameter distances, with MHC class II sequences from other marsupials for comparison.

Detection of positive selection and gene conversion

Where positive selection has acted at a locus, the ratio (ω) of non-synonymous (dN) to synonymous (dS) nucleotide substitutions is expected to be greater than 1. MEGA version 4 was used to calculate ω across the 15 sequenced allele segments, using the Nei–Gojobori method with the Jukes–Cantor modification (Nei and Kumar 2000). The Codon-based Z-test of selection was used to determine if dN was statistically significantly greater dS.

Within the peptide binding region, there is expected to be stronger positive selection, and therefore a greater ω, at the peptide binding site (PBS) codons compared to the non-PBS codons (Ford 2002; Nei and Kumar 2000). These sites have not yet been determined in DAB β1 as extensive structural investigations of this molecule have not been performed.

We used CODEML in PAML version 4.2b (Yang 2007) to detect sites of the DAB β1 at which positive selection has acted. At each codon, CODEML calculated Bayes Empirical Bayes probabilities for two comparisons: the first compared M1a (nearly neutral) versus M2a (positive selection), while the second compared two indices, M7 versus M8, which provides a more robust indication that positive selection is occurring at individual codons.

To further characterize the black-footed rock-wallaby DAB β1 alleles, we used the program GENECONV version 1.81a (Sawyer 1999) to detect if gene conversion had occurred between the ancestors of any of the black-footed rock-wallaby DAB β1 sequences.

Results

Overall DAB β1 variation

In total, 15 alleles were sequenced from across the five populations, and were named with the prefix Pela-DAB as per MHC nomenclature conventions (Klein et al. 1990). The average number of pair-wise nucleotide and amino acid differences between alleles was 25.267 (SE = 2.67) nucleotides (average percent difference = 14.7%) and 15.333 (SE = 2.132) amino acids (average percent difference = 27%) respectively. Thirteen of these fifteen alleles were amplified from two or more independent PCR reactions, as per the MHC allele verification criteria of Kennedy et al. (2002). Twenty sequences differed by three or less nucleotides from other alleles sequenced from the same individual, and were not amplified in independent PCRs. According to the criteria of Kennedy et al. (2002), these sequences were not considered to be unique alleles. They are probably the result of single nucleotide Taq DNA polymerase mis-incorporation errors during amplification of a true DAB allele. Two sequences that were not verified in independent PCRs still differed substantially enough from other alleles that we considered them as true alleles for the purposes of this study. The sequences, Pela-DAB*04 and Pela-DAB*09, differed by >8 and >25 nucleotides respectively from other verified alleles (Table 2).
Table 2

Number of nucleotide differences (above diagonal) and amino acid differences (below diagonal, bold) between alleles (with pair-wise rather than complete deletion of alignment gaps)

 

Pela-DAB*01

Pela-DAB*02

Pela-DAB*03

Pela-DAB*04

Pela-DAB*05

Pela-DAB*06

Pela-DAB*07

Pela-DAB*08

Pela-DAB*09

Pela-DAB*10

Pela-DAB*11

Pela-DAB*12

Pela-DAB*13

Pela-DAB*14

Pela-DAB*15

Pela-DAB*01

 

6

29

16

29

17

17

15

33

22

22

27

24

24

21

Pela-DAB*02

3

 

29

16

34

18

13

15

35

22

23

32

26

25

23

Pela-DAB*03

16

16

 

32

23

32

30

25

34

32

22

28

26

23

24

Pela-DAB*04

10

9

20

 

38

26

23

17

40

8

29

34

27

26

25

Pela-DAB*05

15

17

13

22

 

34

34

27

30

35

25

13

18

26

25

Pela-DAB*06

11

12

16

19

17

 

17

22

35

32

20

36

27

26

24

Pela-DAB*07

10

8

17

15

19

11

 

21

34

27

22

33

26

20

21

Pela-DAB*08

10

10

15

13

15

14

13

 

33

21

22

27

21

21

19

Pela-DAB*09

19

20

20

22

19

21

21

20

 

36

27

26

32

32

25

Pela-DAB*10

13

12

20

5

20

22

17

14

20

 

29

30

25

28

21

Pela-DAB*11

15

15

14

18

16

13

15

13

17

19

 

26

25

23

23

Pela-DAB*12

14

16

17

21

8

20

19

17

16

18

17

 

29

26

26

Pela-DAB*13

12

13

15

16

9

17

16

11

20

16

15

17

 

16

17

Pela-DAB*14

14

15

15

16

16

17

13

14

19

17

15

18

12

 

20

Pela-DAB*15

12

14

15

18

15

15

15

11

15

15

17

17

11

13

 

Several individuals contained an SSCP banding pattern representing a possible sixteenth allele. Direct sequencing of an individual containing only this banding pattern produced the sequence of Pela-DAB*13, and cloning of individuals that contained this banding pattern together with the banding pattern of Pela-DAB*13 also produced only Pela-DAB*13 sequence. Variant banding patterns of the same sequence can appear on SSCP gels (Wallace 2002), so this sixteenth allele was likely a variant SSCP conformation of Pela-DAB*13.

Identified alleles were of three sizes, resulting from in-frame insertion/deletion events (indels): 169, 172 (one additional amino acid) and 178 bp (three additional amino acids). Indels in DAB β1 alleles have been noted in the tammar wallaby, and because were in-frame they did not appear to make the alleles non-functional (Browning 2009).

DAB β1 variation in the island compared to mainland populations

Only two alleles were identified in the Barrow Island population (Pela-DAB*01, frequency 0.73 and Pela-DAB*02, frequency 0.27), neither of which were detected in the mainland populations. Observed heterozygosity (Ho) was 0.29, and gene diversity (expected heterozygosity: He) was 0.395. Genotype frequencies for Barrow Island (Appendix 1 in Electronic supplementary material) did not differ significantly from Hardy–Weinberg Equilibrium (HWE) expectations (χ2 = 0.07, P = 0.9656). Given that no individual contained more than 2 alleles, and that the Barrow Island DAB β1 behaved according to HWE expectations if treated as a single locus, it is likely that only one DAB gene copy was present in the Barrow Island population.

Thirteen alleles were found in the mainland populations (Pela-DAB*03 to Pela-DAB*15). Each population contained from four to seven alleles (Table 1). The average number of alleles per individual ranging from 2.1 to 3 among the populations (Table 1) and only three individuals at Nangeen Hill and two at Mount Caroline possessed one allele. In contrast, on Barrow Island, over 70% of individuals had only one allele. The number of DAB alleles per individual was significantly lower (P = 4.208 × 10−14) in the island compared to the mainland population (all mainland populations grouped as one) (island average = 1.29, mainland average = 2.77).

Only one allele (Pela-DAB*12) was shared between the two mainland regions (the Wheat-belt and Exmouth) (Table 3). Within the Wheat-belt, a number of alleles were shared among the populations, as could be expected based on their close geographic proximity (Table 3).
Table 3

Amino acid sequences of alleles and the presence (+) or absence () of each allele in each population

https://static-content.springer.com/image/art%3A10.1007%2Fs10592-009-9993-y/MediaObjects/10592_2009_9993_Tab3_HTML.gif

Evidence for multiple gene copies of DAB in the mainland populations was found. Nine of 11 individuals in Exmouth, 10 of 17 at Mt. Caroline, two of nine at Nangeen Hill and all at Tutakin had three or four alleles and therefore at least two DAB gene copies (Appendix 1). In addition, some individuals in Mt. Caroline and Nangeen Hill appeared to possess only one gene copy as they had only one DAB β1 allele.

There were only 3 amino acid (aa) differences between the two Barrow Island DAB β1 alleles. In contrast, there was high amino acid sequence divergence between alleles in each of the mainland populations. Average pair-wise amino acid divergence between alleles was 16.5 aa at Exmouth (N = 21 pair-wise comparisons), 16.7 aa at Mount Caroline (N = 6), 15.3 aa at Nangeen Hill (N = 15) and 18 aa at Tutakin (N = 6).

Overall, the mainland populations contained greater DAB β1 diversity than the Barrow Island population in three ways: a greater number of alleles, a greater degree of sequence divergence between alleles, and multiple DAB gene copies.

Genetic variation and population differentiation at DAB β1 versus microsatellites

There were a greater number of DAB β1 alleles than microsatellite alleles (averaged across all polymorphic microsatellite loci) for all populations except Tutakin (Table 1). In addition, gene diversity was far higher at DAB β1 than at microsatellites for the Barrow Island population (Table 1). This comparison could not be performed in the mainland populations because the presence of multiple, co-amplifying DAB gene copies prevented the calculation of gene diversity for any individual DAB gene copy in these populations.

At DAB β1 the fixation index, FST, was greater than 0.4 between Barrow Island and every one of the four mainland populations, indicating a high level of population differentiation between island and mainland (Table 4). FST ranged from 0.12 to 0.2 when Exmouth was compared with each of the three Wheat-belt populations, indicating moderate population differentiation between the northern vs southern mainland populations. FST ranged from 0.067 to 0.12 amongst the three Wheat-belt populations, indicating modest population differentiation, as expected by their proximity.
Table 4

Pair-wise FST values for all population comparisons for MHC Class II DABβ1 locus (below diagonal, bold) and for microsatellites (above diagonal)

 

Barrow Island

Exmouth

Mt Caroline

Nangeen Hill

Tutakin

Barrow Island

 

0.67760*

0.66795*

0.72898*

0.67241*

Exmouth

0.50651*

 

0.37992*

0.44431*

0.33304*

Mt Caroline

0.42442*

0.20631*

 

0.22726*

0.17473*

Nangeen Hill

0.52470*

0.12375

0.07901

 

0.25469*

Tutakin

0.45457*

0.20233*

0.06797

0.12164

 

The P-value for the computation of every FST value: * P > 0.00001,  P > 0.01,  P > 0.02

FST was greater at microsatellite loci than at DAB for each pair-wise population comparison (Table 4), but followed the same pattern of decreasing FST with decreasing distance between populations. FST ranged from 0.67 to 0.72 between Barrow Island and the mainland populations, 0.33–0.44 between Exmouth and the Wheat-belt populations, and 0.17–0.25 amongst the Wheat-belt populations.

Isolation by distance was statistically significant for DAB β1 (P = 0.0480) and for microsatellites (P = 0.0489) (Fig. 3a, b). Visually, the slope of the line was far more gradual (indicating reduced isolation by distance) at DAB compared to microsatellites, however we were unable to statistically test the difference in slope with the analysis software we used.

Evidence for positive selection and gene conversion at DAB β1

The 15 sequenced DAB β1 alleles had an overall ω ratio of 1.764. The Codon-based Z-test of selection rejected the hypothesis that dN = dS, in favour of the alternative hypothesis that dN > dS, averaged over the 15 alleles of DABβ1 (P = 0.04). A ω ratio of greater than one indicates the presence of positive selection at a locus over long evolutionary time periods (Hedrick et al. 2001a). Thus, the large degree of diversity present at the black-footed rock-wallaby DAB β1 is likely the result of a long period of positive selection. Both comparisons performed by CODEML detected statistically significant positive selection at codons for 10 amino acid sites (Table 5). GENECONV detected gene conversion between the ancestral sequences of 5 pairs of sequenced alleles, and between Pela-DAB*03 and an unknown allele (Table 6). We note that likelihood methods of detecting positive selection, such as CODEML, may generate moderately higher levels of false-positive results if gene conversion has occurred (Casola and Hahn 2009; Anisimova et al. 2003).
Table 5

CODEML detection of amino acid sites under position selection

Codon number

Amino acid

Probabilitya ω > 1 (M1a vM2a)

Mean ω (M1a v M2a)

Probabilitya ω > 1 (M7 v M8)

Mean ω (M1a v M2a)

6

Leucine (L)

0.981*

6.975 ± 1.362

0.993**

6.977 ± 1.218

8

Leucine (L)

1.000**

7.099 ± 1.084

1.000**

7.021 ± 1.114

10

Tyrosine (Y)

0.998**

7.085 ± 1.118

1.000**

7.018 ± 1.121

12

Tyrosine (Y)

0.999**

7.095 ± 1.095

1.000**

7.020 ± 1.116

17

Tyrosine (Y)

1.000**

7.099 ± 1.084

1.000**

7.021 ± 1.113

39

Serine (S)

1.000**

7.098 ± 1.086

1.000**

7.021 ± 1.114

42

Tyrosine (Y)

0.999**

7.096 ± 1.093

1.000**

7.020 ± 1.116

49

Isoleucine (I)

1.000**

7.097 ± 1.089

1.000**

7.021 ± 1.115

52

Glutamine (Q)

0.958*

6.820 ± 1.604

0.989*

6.945 ± 1.276

53

Arginine (R)

1.000**

7.098 ± 1.086

1.000**

7.021 ± 1.114

aBayes empirical Bayes Posterior Probabilities: * P > 95%, ** P > 99%. ω = ratio of non-synonymous to synonymous nucleotide substitutions

Table 6

Gene conversion events detected through GENECONV program

Allele pairs

Simulated P-value

Nucleotide positions of DAB β1 sequence involved in gene conversion event

Pela-DAB*04;Pela-DAB*13

0.0087

111–144

Pela-DAB*01;Pela-DAB*15

0.0136

128–177

Pela-DAB*14;Pela-DAB*15

0.0180

76–143

Pela-DAB*10;Pela-DAB*13

0.0292

111–144

Pela-DAB*04;Pela-DAB*10

NSa

25–154

Pela-DAB*03;unknown allele

0.0052

148–154

aValue indicated as not significant for simulated P-value, but is significant for Bonferroni-corrected pairwise permutation P-value (P = 0.0210)

Phylogeny of the DAB β1 alleles

The phylogeny of the 15 sequenced black-footed rock-wallaby DAB β1 alleles, including DAB β1 and DBB β1 alleles from other marsupials, is displayed in Fig. 2. The DAB sequences from the other marsupials clustered in among the black-footed rock-wallaby alleles, suggesting that the origin of DAB predates the divergence of the possum and wallaby lineages (55 million years before present: Westerman et al. 1990; Kirsch et al. 1997). The alleles from separate mainland populations did not cluster together. This is expected, given that the divergence of DAB alleles is very likely to have occurred prior to the isolation of each population. However, the Barrow Island alleles clustered together, and so appeared to have been recently derived.
https://static-content.springer.com/image/art%3A10.1007%2Fs10592-009-9993-y/MediaObjects/10592_2009_9993_Fig2_HTML.gif
Fig. 2

Neighbour-joining tree of black-footed rock-wallaby DAB β1 alleles (indicated as Pela-DAB*01-Pela-DAB*15) aligned with other marsupial DAB β1 and DBB β1 alleles. DAB sequences from different marsupial species cluster together, demonstrating that many polymorphisms arose prior to the divergence of these species. Genbank numbers for other species’ sequences are: Macropus eugenii: AY438042 (Maeu-DAB*01), AY438038 (Maeu-DBB*01), AY438039 (Maeu-DBB*02), AY438040 (Maeu-DBB*03), AY438041 (Maeu-DBB*04); Macropus rufogriseus: M81624 (Maru-DAB*01), M81626 (Maru-DAB*02), M81625 (Maru-DBB*01); Monodelphis domestica: NM 001032991 (Modo-DAB*01); Sarcophilus harrisii: EF591102 (Saha-DAB*01), EF591105 (Saha-DAB*05); EF591107 (Saha-DAB*06); Trichosurus vulpecula: AF312029 (TrvuMHC HLA-DRB*02), AF312030 (TrvuMHC HLA-DRB*03). The T. vulpecula sequences are labelled DRB, but since their naming, it has been established that they belong to the DAB family (Belov et al. 2004). All bootstrap values over 65 have been included

Discussion

Fifteen DAB β1 alleles were detected amongst the five black-footed rock-wallaby populations, with moderately high nucleotide sequence diversity amongst alleles (average = 14.7%), and 1–4 alleles per individual. Positive selection was detected at 10 specific amino acid sites. Though total DAB β1 diversity in the black-footed rock-wallaby is moderately high when compared with other eutherian mammals (Bernatchez and Landry 2003), diversity in each individual mainland population (4–7 alleles per population) is moderately low and diversity in the island population is very low.

MHC diversity has only recently been assessed in marsupials, and compared to these the black-footed rock-wallaby again has moderate overall diversity but low within-population diversity. Tammar wallabies on Kangaroo Island possess 5–8 DBB β1 alleles per individual, with 44 alleles identified in total (Cheng et al. 2009). Introduced brushtail possums in New Zealand, examined at four MHC class II loci, contained 20 DAB β1 alleles, 5 DAA β1 alleles (2–4 per population), 12 DBA β1 alleles (3–7 per population) and 5 DBB β1 alleles (2–4 per population) (Holland et al. 2008). Tasmanian devils (Trichosurus vulpecula) contain 26 MHC class I α1 alleles and 16 class I α2 alleles, however these alleles have very low amino acid diversities and many fewer substitutions at peptide binding sites than a severely bottlenecked population of lions (Siddle et al. 2007). Forty-four DAB β1 alleles (average percent difference between allele sequences: 14.5%) were found in the Brazilian gracile mouse opossum (Gracilinanus microtarsus), whilst only 8 DAB β1 alleles were found in the gray slender mouse opossum (Marmosops incanus) (average percent difference: 13.1%) (Meyer-Lucht et al. 2008).

Island compared to mainland diversity

The Barrow Island black-footed rock-wallaby population possesses a lower level of diversity at DAB than the mainland populations, in three ways: a lower number of alleles, a lower level of sequence divergence between alleles, and fewer DAB gene copies. As the two Barrow Island DAB β1 alleles differ at only three protein residues, there may be significant overlap in the sets of peptides that each allele can bind. These effects combine to severely limit the number of antigenic peptides from extracellular pathogens that individuals in the Barrow Island population can recognise.

As the Barrow Island alleles are very similar to one another in sequence, this suggests that they diverged in the recent past. Therefore, a very plausible explanation for the pattern of DAB diversity on Barrow Island is that DAB became fixed for one allele in the past—either through genetic drift due to small population size, or strong directional selection due to an epidemic—and the two alleles we see today diverged from this allele. The Barrow Island population has been isolated from the mainland for at least 8,000 years (Dortch and Morse 1984), equating to 1,600 generations (Eldridge et al. 1999). Eldridge et al. (1999) estimated that all neutral variation would have been lost in this population within 300 generations, and that the expected heterozygosity maintained by drift mutation equilibrium would be 0.055. Genetic drift could have plausibly forced the fixation of the DAB locus at some time in the past.

The few studies of MHC diversity in both insular and mainland populations of the same species concur with this study in finding that island populations have lower MHC diversity than mainland populations. Island populations of the Australian bush rat (Rattus fuscipes greyii) have lower MHC locus allelic diversity and heterozygosity than mainland populations (Seddon and Baverstock 1999). Likewise, island populations of the woodmouse (Apodemus sylvaticus) have lower MHC locus allelic diversity than mainland populations (de Bellocq et al. 2005). Populations of African green monkeys (Cercopithecus aethiops sabaeus) and desert bighorn sheep (Ovis canadensis mexicana) introduced to islands by humans (and having experienced a bottleneck during establishment) also displayed greatly reduced MHC diversity compared to their mainland source populations (Hedrick et al. 2001b; Lekutis and Letvin 1995). Two other taxa endemic to islands (Galapagos penguin Spheniscus mendiculus and Chatham Island Robins Petroica traversi) have low MHC class II diversity than closely related taxa existing on larger landmasses (Bollmer et al. 2007; Miller and Lambert 2004). An assessment of the quantity of parasites and pathogens encountered in island versus mainland populations may help explain whether there is a lessening of selection pressure at MHC loci in island populations, or whether these patterns are only the result of low population size and genetic drift.

The greater level of MHC diversity in the Wheat-belt versus Barrow Island populations of the black-footed rock-wallaby was interesting, given that the Wheat-belt populations have been fragmented and have suffered declines. Small mainland populations are expected to suffer from the same afflictions that island populations suffer: genetic drift, inbreeding, and loss of genetic diversity (Frankham 1996). Mount Caroline, Tutakin and Nangeen Hill experienced population bottlenecks of 10–30 individuals before fox baiting began and population numbers recovered (Hall and Kinnear 1991; Kinnear et al. 1998). The Wheat-belt black-footed rock-wallaby populations have been fragmented within the last 140 years, whereas the Barrow Island population has been isolated for the past 8,000 years. Possibly, population fragmentation and bottlenecks of the Wheat-belt populations have not endured for long enough for genetic erosion at DAB β1 to have reached the level seen in the Barrow Island population. Alternatively, limited dispersal between the Wheat-belt populations may have been enough to prevent erosion of genetic diversity, even though dispersal is partly inhibited by fox predation (Kinnear et al. 1988; Eldridge et al. 2001). In addition, the mainland populations may be exposed to more pathogens than the isolated island population, ensuring that a broader range of MHC alleles is selected for.

Many individuals in the mainland populations probably have two DAB gene copies. The presence of several copies of an MHC gene is evolutionarily favorable, as it increases the number of alleles that can be carried and therefore the antigen presentation capacity of an individual (Doherty and Zinkernagel 1975). It is not unusual for individuals within a population to vary in the number of class II MHC gene copies, and has been observed in the Madagascan mouse lemur (Schad et al. 2004), the Gila Topminnow (Hedrick et al. 2001a), the Madagascan rodent Eliurrus myoxinus (Sommer et al. 2002), humans (Zhang et al. 1990), and domestic cats (Kennedy et al. 2003), amongst others. This is due to the existence of MHC haplotypes with varying numbers of MHC gene copies (Zhang et al. 1990, Kennedy et al. 2003).

DAB β1 diversity versus microsatellite diversity

The level of MHC diversity in the Barrow Island black-footed rock-wallaby population appears to be higher than at microsatellite loci (Table 1). Both FST and isolation by distance appear to be greater at the microsatellite loci than at DAB β1. Positive selection (detected in MEGA and CODEML) has maintained greater diversity at DAB β1 than would be expected under neutrality, at least partly explaining why DAB diversity appears to be greater than at the microsatellite loci. The difference between MHC and microsatellite diversity is most acute in the Barrow Island population, where gene diversity is over seven times higher at DAB β1 compared to the single microsatellite locus that is polymorphic in that population (10 out of the 11 microsatellite loci are homozygous at Barrow Island) (Eldridge et al. 1999).

A similar situation is seen in the San Nicholas Island fox (Aguilar et al. 2004), which is homozygous at microsatellite loci but has observed heterozygosity of 0.36 at the class II locus DRB (Aguilar et al. 2004). However, other researchers have reported greater levels of genetic diversity at neutral loci than MHC loci in small populations. Many island populations of bush rat contain higher levels of microsatellite diversity than diversity at the class II locus DQ-α (which is homozygous in most populations) (Hinten et al. 2003; Seddon and Baverstock 1999). A similar situation also exists in populations of foxes on San Miguel Island and San Clemente Island (Aguilar et al. 2004). Microsatellites have a far higher mutation rate (Weber and Wong 1993) than MHC loci and other functional genes (Melvold et al. 1997; Satta et al. 1993). Plausibly, if a population became fixed at many loci due to a bottleneck during founding, the greater microsatellite mutation rate may generate new microsatellite alleles relatively quickly, resulting in higher levels of microsatellite diversity than MHC diversity.

Variation in selection pressure across an environment should lead to greater population differentiation at MHC loci compared to neutral loci, whilst balancing selection occurring in the absence of local adaptation will lead to lower population differentiation at MHC compared to neutral loci (Charbonnel and Pemberton 2005). The black-footed rock-wallaby displays lower population differentiation at DAB β1 compared to microsatellites according to F-statistics and isolation by distance (Fig. 3, Table 4). However, very few DAB β1 alleles are actually shared between Barrow Island, Exmouth and the Wheat-belt, and possibly the lower level of population differentiation at the DAB β1 compared to neutral loci is a statistical artifact of the method we used (Ekblom et al. 2007; Louiseau et al. 2009) to analyse MHC alleles amplified from multiple loci. If it is true that there is lower population differentiation at DAB β1 than at microsatellites, then the selection pressures operating on DAB β1 are likely to be similar in the examined black-footed rock-wallaby populations, so that less population differentiation is present at DAB β1 than would be expected if only drift and gene flow were operating on this locus. Similar selection pressures in different populations of the black-footed rock-wallaby would be unusual as the populations in the three regions (Barrow Island, Wheat-belt, Exmouth) are so far apart.
https://static-content.springer.com/image/art%3A10.1007%2Fs10592-009-9993-y/MediaObjects/10592_2009_9993_Fig3_HTML.gif
Fig. 3

Isolation by distance (FST/1 − FST) for DAB β1 (a), microsatellites (b) for each pair-wise population comparison. Note that, for all populations, isolation by distance is greater at the microsatellite loci than at DAB β1

In some other species, population differentiation of MHC loci versus neutral loci is dependent on geographic distance between populations because the greater geographic distance places populations in different selection environments. Populations of the house sparrows (Passer domesticus) that are geographically close (and so have similar selection regimes) display weaker levels of population differentiation at MHC loci compared to microsatellite loci, whilst populations at greater distances from each other (and so under different selection regimes) display stronger levels of population differentiation (Louiseau et al. 2009). Similarly, populations of great snipe (Gallinago media) that are in ecologically distinct areas display significant MHC differentiation (after controlling for variation at neutral loci) yet populations that are close by display no significant population differentiation by this measure (i.e. their MHC differentiation is not greater than microsatellite differentiation) (Ekblom et al. 2007). Soay sheep (Ovis aries), which live within several environments on one small island, display greater polymorphism at MHC compared to neutral loci in some populations, possibly because selection pressure (from parasite infection) varies within the different environments on the island over time (Charbonnel and Pemberton 2005).

Whilst diversity at DAB β1 is greater than at microsatellite loci, the pattern of diversity in island compared to mainland populations of the black-footed rock-wallaby is the same at all loci. Both DAB β1 and microsatellite diversity are far lower in the island population compared to all the mainland populations.

Conservation implications

The evidence presented here suggests that MHC diversity is considerably lower in the island population than mainland populations of the black-footed rock-wallaby. If this situation is the same at other fitness-related loci, then the island population has a reduced value as a reservoir population. However, the remote island population does contain a unique assemblage of MHC class II alleles, so has a role in the preservation of overall genetic diversity in the black-footed rock-wallaby.

This study supports the assertion of Eldridge et al. (2004) of the importance of preserving fragmented mainland populations, even where island populations exist. As mainland populations contain greater levels of MHC diversity, and potentially overall fitness-related diversity, they are the most useful as sources of individuals for re-introduction efforts, or for supplementing the genetic diversity of other mainland and island populations. Establishing new populations by translocating individuals from other populations may benefit the most from a strategy of pooling individuals from all populations, both mainland and island, to encapsulate a spectrum of the genetic diversity of the species in the new founding population. Importantly, island populations should not be used as the sole source of individuals for re-introduction, as their low levels of genetic diversity may severely hinder the ability of the founding population to adapt to a new environment. Studies of other fitness associated loci need to be completed in island populations to assess if a lack of fitness-related diversity, as observed at the MHC, is the norm.

Acknowledgments

We thank Juliet King (University of Western Australia), Jack Kinnear, Graham Hall and staff of the Department of Conservation and Land Management, Western Australia for sample collection, as well as Cushla Metcalfe, Kathy Belov, Dave Briscoe and Jenny Donald, and two anonymous reviewers for helpful discussion or comments on the manuscript. This research was supported by a 2003 Australian Academy of Science Award for Research on the Conservation of Endangered Australian Vertebrate Species, awarded to M. D. B. Eldridge.

Supplementary material

10592_2009_9993_MOESM1_ESM.doc (38 kb)
Supplementary material 1 (DOC 38 kb)

Copyright information

© Springer Science+Business Media B.V. 2009