Simultaneous studies of both nuclear and mitochondrial markers were undertaken in two widespread Indo-West Pacific (IWP) marine invertebrates to compare and contrast the ability of these markers to resolve genetic structure. In particular, we were interested in the resolution of a genetic break between the Indian and Pacific Oceans due to historical isolation. Sequence variation from the nuclear gene encoding myosin heavy chain (MyHC) and the mitochondrial gene cytochrome oxidase I (COI) were examined for the snapping shrimp Alpheus lottini from wide-ranging populations throughout the Indian and Pacific Oceans. A previously identified genetic break between oceans based on COI sequences appears to have been an artifact caused by the inadvertent inclusion of pseudogene sequences; our new COI data provide evidence only of a break between IWP and East Pacific populations. Distribution of a single nucleotide polymorphism in MyHC, on the other hand, shows evidence of a cline between Indian and Pacific Oceans. New allozyme and mtDNA sequence data were also obtained for the starfish Linckia laevigata. Allozyme data show a clear genetic break between Indian Ocean populations and Pacific (including western Australian) populations, whereas the distribution of mtDNA haplotypes shows a region of overlap in the central IWP. Comparisons of our data for both Alpheus and Linckia with data from other population genetic studies in the IWP suggest that nuclear markers (allozymes, sequence data and morphological characters) may in some instances reveal historical patterns of genetic population structure whereas mtDNA variation better reflects present day patterns of gene flow.
At present, there is a continuous route for gene exchange between tropical organisms in the Indian and Pacific Oceans. The main pathway is via the Indonesian throughflow, which is responsible for a large transfer of warm surface water from the tropical north-west Pacific along the Makassar Strait, into the Flores and Banda Sea before entering the Timor Sea and the Indian Ocean (Gordon and Fine, 1996). However, for much of the last 3 million years, sea levels are thought to have been substantially lower than at present. For instance, during the height of the last glaciation about 18,000 years ago, sea levels are thought to have been about 130 m lower than present day levels (Chappell and Shackleton, 1986). At such times the Indonesian throughflow is thought to have been greatly reduced, and the Torres Strait (which provides a tropical marine connection between western and eastern Australia) was completely closed by a land bridge between New Guinea and Australia, thereby greatly decreasing the opportunity for genetic exchange between the two oceans. Multiple glaciations resulted in repeated periods of isolation between tropical marine faunas in the Pacific and Indian Oceans, providing an important mechanism for population differentiation and incipient speciation.
The distribution of many sister species is consistent with this idea, with one species found only in the Pacific Ocean and the other only in the Indian Ocean, often with a boundary near the Indonesian archipelago (e.g.,Woodland, 1983; McManus, 1985; McMillan and Palumbi, 1995; Lessios et al., 2001). Since this is a common pattern in many marine organisms, it is consistent with the possibility that a vicariant event has been influential in their evolution (Avise, 1994). Moreover, molecular genetic analyses of some species that are widely and continuously distributed throughout the Indian and Pacific Oceans have revealed evidence of population structure that mirrors the same biogeographic pattern described above, with a genetic break between populations in the Indian Ocean and the Pacific Oceans. Several of these broadscale molecular studies have focussed on using mitochondrial DNA which, although variable, must be considered a single locus (e.g., deep sea fish, Miya and Nishida, 1997; snapping shrimp, Alpheus lottini, Williams et al., 1999; mud crab, Gopurenko et al., 1999; sea urchin Eucidaris metularia, Lessios et al., 1999; two species of bonefish, Colburn et al., 2001; two species of Diadema sea urchin, Lessios et al., 2001). Others have used only nuclear genes, most commonly allozyme markers (e.g., damselfish, Lacson and Clark, 1995; coconut crab, Lavery et al., 1995; crown-of-thorns starfish, Benzie, 1999; mangrove tree, Duke et al., 1998; tiger prawn, Duda and Palumbi, 1999).
To date, broadscale genetic structure has been elucidated combining both nuclear and mtDNA variation for only two organisms that show an IWP break (coconut crab, Lavery et al., 1995, 1996; starfish Linckia laevigata, Williams and Benzie, 1998). The use of multiple loci in population studies is important since the pattern observed at any one locus may not truly reflect the species genetic structure. For instance, reliance on mtDNA sequences alone may be hazardous because inadvertent incorporation of mtDNA pseudogenes into analyses could result in incorrect inferences about gene flow. Moreover, mitochondrial and nuclear genes might be expected to differ in their resolution of past and present gene flow because of differences in the rates at which they come to genetic equilibrium.
Here we provide new comparisons of nuclear and mitochondrial loci for two previously analyzed and widely distributed Indo-Pacific species, the snapping shrimp Alpheus lottini and the starfish Linckia laevigata. Previously, the snapping shrimp Alpheus lottini showed a fixed difference at a single nucleotide in the COI gene between populations in the Indian and Pacific Oceans (Williams et al., 1999). However, it is now clear that pseudogene sequences are pervasive and difficult to detect in many Alpheus species (Williams and Knowlton, 2001). Therefore, we have tested for the possibility of pseudogenes in this A. lottini study by obtaining new COI sequences for Pacific samples. The genetic structure observed from the mtDNA is also compared with that observed using new sequence data from the nuclear myosin heavy chain gene. In the case of the starfish Linckia laevigata, a combination of morphological (color), nuclear (allozymes) and mitochondrial data (COI sequence, and PCR-RFLP of a portion of mtDNA including the control region) were previously shown to be consistent with an IWP break (Williams and Benzie, 1998; Williams, 2000). In this study we extend these earlier findings by incorporating additional data to determine whether new samples from Indonesia, western Australia and the Seychelles fit the expected biogeographic pattern. Finally, we also discuss the implications of differences between the degree of structure observed using mtDNA and nuclear loci.
MATERIALS AND METHODS
Alpheus lottiniGuérin-Méneville, 1829
Tissue samples of Alpheus lottinisensu stricto used to extract DNA in this study were those used in two previous studies (Knowlton and Weigt, 1997; Williams et al., 1999) with the exception of newly collected samples from Clipperton Atoll and Panama. Shrimp from Clipperton Atoll and Panama were taken from Pocillopora corals and frozen in liquid nitrogen. Total genomic DNA was obtained from all Pacific samples using a PureGene™ DNA isolation kit, following the manufacturer's protocols, except that samples were digested overnight or for several days at 65°C with 6–10 μL of Proteinase K (20 mg/ml).
New COI sequences were obtained for Pacific sites (Pohnpei, Palau, Clipperton and Panama) from new DNA extractions and used in conjunction with previously published COI sequences from the Indian Ocean obtained by S.T.W. previously (Williams et al., 1999; GenBank accession numbers: AF107049–AF107068). Some of the new sequences obtained for the Pacific were from the same individuals sequenced previously (Knowlton and Weigt, 1997, samples from Pohnpei and Palau, GenBank accession numbers U76428–U76440). One COI sequence from a shrimp from Panama was obtained previously by RT-PCR of cDNA (Williams et al., 2001, GenBank accession number AF310752).
The nuclear myosin heavy chain gene codes for a contractile protein found in many tissues, usually in multiple forms (e.g.,Mahdavi et al., 1982). In vertebrates these different isoforms arise from multiple copies within a gene family. However, in some animals, copies arise from differential splicing of an RNA arising from single gene (e.g.,Drosophila, Rozek and Davidson, 1986). Initial studies showed that the latter is also the case for Alpheus. MyHC amplified from cDNA resulted in a variety of different gene products differing in length and with rearrangements in the order of pieces of sequences. Amplification from the same animal using the same primers and amplification conditions, but using gDNA, resulted in a single gene product in Alpheus lottini (and one or two in other Alpheus species, unpublished data, S.T.W.). This suggests that MyHC is a single or low copy nuclear gene in Alpheus and as such may be a potentially valuable marker for population and phylogenetic studies.
MyHC sequences homologous to sequence spanning exons 15 and 16 in Drosophila melanogaster (GenBank # X53155) were obtained from Indian Ocean sites within the Chagos archipelago, the Red Sea, the Seychelles and from north-western Australia, using DNA extractions from a previous study (Williams et al., 1999). MyHC sequences were also obtained from additional samples from north-western Australia and samples from five sites in the Pacific Ocean: Pohnpei, Palau, Moorea, Clipperton Atoll and the Pacific coast of Panama. Initially about 830 bp (including a variable size intron of about 155 bp) were amplified for 31 individuals representing a sub-sample of animals from all populations. These amplifications were cloned and sequenced in order to investigate allelic variation, following the protocol in Williams and Knowlton (2001). Sample sizes for each population were then increased by amplifying a smaller portion of MyHC of the 5′ exon (which did not show size variation) and sequencing the polymerase chain reaction (PCR) product directly (see Results).
Diluted total DNA (15–80 ng) was used in a PCR to amplify approximately 615 bp of the cytochrome oxidase I (CO I) gene, and approximately 830 bp (long sequences) or 339 bp (short sequences) of MyHC. Amplification reactions for MyHC (long sequences) were performed in 50 μl volumes containing 0.1 μM of MyHC1124 and MyHC1806 (sequences in Table 1), 200 μM of each dNTP, 1.5 mM magnesium chloride, 2.5 units of AmpliTaq Gold™ DNA polymerase and 5 μL of AmpliTaq™ buffer (10 ×). Short sequences were amplified in 30 μl reactions using primers MyHC1124 and MyHClottR (Table 1). Thermal cycling for all MyHC PCRs was performed as follows: initial “hot-start” denaturation for 8 minutes at 95°C to process the polymerase, followed by 40 cycles of 1 minute at 94°C, one minute at 56°C and 80 sec at 72°C with a final extension of 30 minutes at 72°C. Amplification reactions for COI were performed in 30 μl volumes, with the same concentrations as for MyHC but using “normal” Taq instead of Taq Gold. Two primer pairs were used to amplify COI: either COIF with COI(10) or COIF with H7188 (Table 1). Thermal cycling was performed as follows: initial denaturation for 2 minutes at 95°C, followed by six cycles of 15 sec at 95°C, 15 sec at 45°C, 1 minute at 72°C, then 30 cycles of 15 sec at 95°C, 15 sec at 48°C and 1 minute at 95°C with a final extension of 3 minutes at 72°C.
Those MyHC PCR reactions from clones with the correctly sized insert were prepared for sequencing by incubation with eight units of exonuclease I and 1.5 units of shrimp alkaline phosphatase for 2 hours at 37°C followed by 15 minutes at 80°C to inactivate the enzymes. At least eight clones were sequenced from each individual. All COI and short MyHC (exon sequence only) amplicons were enzyme purified and sequenced directly without cloning. Automated sequencing was performed using a dRhodamine Kit (PE-ABI). Protocols for purification and cycle sequencing followed manufacturer's instructions, and the products of sequencing reactions were run on a 377 Applied Biosystems automated sequencer. Final sequence data were verified by comparisons between forward and reverse sequences.
Trees were constructed from COI and long MyHC nucleotide sequences using a neighbor joining (NJ) algorithm in the computer program PAUP* 4.0b3a (Swofford, 2001). Genetic distances were estimated using Kimura's 2-parameter model (Kimura, 1980). Trees were rooted with sequence data from Alpheus saxidomus, which is the closest, available sister taxon to A. lottini (Williams et al., 2001).
Linckia laevigata(Linnaeus, 1758)
A total of 100 new individuals of the starfish Linckia laevigata were screened for allozyme variation: 91 were obtained from three sites in Indonesia (Melintang, Bali, Lombok) and nine from Ningaloo in Western Australia. Variation was screened at seven polymorphic loci: enolase (E.C. 18.104.22.168; ENO); glucose-phosphate isomerase (E.C. 22.214.171.124; GPI); hexokinase (E.C. 126.96.36.199; HK); peptidase using leucylglycylglycine substrate (E.C. 3.4.11/13; LGG); peptidase using leucylproline substrate (E.C. 3.4.11/13; LP); peptidase using leucyltyrosine (E.C. 3.4.11/13; LT) and superoxide dismutase (E.C. 188.8.131.52; SOD). Electrophoresis was carried out in horizontal starch gels using a discontinuous LiOH buffer (gel buffer pH 8.4 and electrode buffer pH 8.15) or on cellulose acetate gels (Cellogel™, Chemetron, Milan) using citrate phosphate buffer (pH 6.5) as in Williams and Benzie (1996).
Calculations of allele frequencies and population genetic statistics were carried out using the Biosys-1 package (v1.7, Swofford and Selander, 1981) and Genepop (v3.3, Raymond and Rousset, 1995). Measures of genetic variability (average heterozygosity, calculated as the actual proportion of heterozygous individuals in the sample [“direct count”]), and the average number of alleles per locus) were calculated for all four new sites and the combined Indonesian sites. The distribution of genotypes across Indonesian sites did not vary significantly, therefore they were combined in all further analyses. The distribution of genotypes did not differ significantly between Ningaloo and other Western Australian populations. However, Ningaloo was kept separate, despite its small sample size, because it is located on the western coast of western Australia and other populations are from the northern coast of western Australia. Chi-square goodness-of-fit tests were performed to test conformance of populations to Hardy-Weinberg equilibrium (HWE). Exact significance probabilities were used to avoid the small cell sizes associated with using chi-square tests when sample sizes and/or some genotypic frequencies were small. F-statistics were calculated using the equations of Weir and Cockerham (1984), whose formulae include corrections for small or unequal sample sizes among the populations tested.
Approximately 690 bp of COI sequence was amplified from four new samples from the Seychelles. DNA was extracted as above for A. lottini, from samples of pyloric caecae preserved in DMSO. Amplification reactions were performed in 25 μl volumes containing 0.4 μM of COIEF and COIER (Table 1), 240 μM of each dNTP, 5 units of TTH DNA polymerase and 2.5 μl of buffer (10×). Cycle parameters were 3 minutes at 95°C, and 40 cycles of 1 minute at 95°C, 1 minute at 50°C, 80 sec at 72°C. PCR products were sequenced directly following the protocol above for A. lottini.
Previous analyses of COI have shown that L. laevigata haplotypes separate into two main clades identified by four fixed differences (Williams, 2000). In order to determine to which clade the new sequences belonged, we constructed a phylogeny as described above for Alpheus lottini. The new sequences were analyzed together with 31 L. laevigata sequences from Williams (2000) (GenBank AF 18792–AF 187950).
Myosin heavy chain
Nineteen unique alleles were identified among 31 shrimp from throughout the Indo-Pacific in 803 bp of MyHC nucleotide sequence (of an amplicon including a variable sized intron of about 155 bp) (GenBank AJ493168–AJ493186). Phylogenetic analysis of these 19 alleles revealed that 50 sites are variable, of which only two are phylogenetically informative for A. lottini and its sister taxon outgroup. Only one of these is phylogenetically informative within A. lottini: a silent change in the exon (base number 284 out of 803) which divides the alleles nearly evenly into two groups (data not shown). The other base substitution (number 328, also in the exon) is a “C” in 18 alleles and a “T” in the other, which occurred only once in the 31 A. lottini and also in the outgroup species. Since there was only one phylogenetically informative site within A. lottini, we decided to amplify 339 bp of the exon spanning position 284. By avoiding amplification of the intron we avoided length variations and were able to sequence PCRs directly. Heterozygotes were unambiguous and could be clearly identified as double peaks occurring in both forward and reverse chromatograms. MyHC variation for the phylogenetically informative site at base number 284 was scored for a total of 106 animals, Allele frequencies are listed in Appendix 1a and allele distribution in Indo-Pacific populations is shown in Figure 1a. Allele “A” was much more common in all Indian Ocean sites than Pacific sites, occurring about 85% of the time or more in the Red Sea, Seychelles and Chagos. Allele “A” occurred about 25% of the time in Clipperton and Panama increasing to 33% in Moorea. Unfortunately around the Indo-Malay region the sample size is small, both in the number of individuals sampled and in the number of populations, but the frequencies of alleles in the ‘A’ clade increase moving from Pohnpei to western Australia, suggesting mixing between the two oceans. This pattern is similar to that seen in the distribution of mtDNA clades in the starfish Linckia laevigata.
Cytochrome oxidase I
New Pacific Ocean sequences were not the same as previous Pacific sequences; new samples had a “C” at position 90, rather than an “A” or “G” identified in a previous study (Knowlton and Weigt, 1997). Although some of the individuals from the Pacific Ocean are the same as those used in a previous study (Knowlton and Weigt, 1997), the most similar pairs of old and new sequences from the same location varied by at least 2 bp. We believe that the data set in the present study is the correct one and that the previous data set from Knowlton and Weigt (1997) contains pseudogene sequences. In support of this, one of the new sequences is identical to a sequence obtained from cDNA for a single individual from Panama (Fig. 3, Williams et al., 2001). This sequence also has a “C” at position 90. We cannot, however, rule out the possibility that some of the variation in our data set may also be due to pseudogenes. In order to reduce the possibility of including pseudogenes, all chromatograms were carefully compared, and sequences with any ambiguous base calls that occurred in both forward and reverse chromatograms were excluded (all new COI sequences, GenBank AJ493473–AJ493488).
Within the Pacific Ocean, the most common COI haplotype (occurring in 6 individuals out of 26) was found in Panama (3/10) and Clipperton Atoll (3/8). One COI sequence from a shrimp from Panama was identical to a sequence obtained previously by RT–PCR of cDNA (Williams et al., 2001, GenBank accession number AF309910). Other identical haplotypes were also shared between Palau and Pohnpei and between Clipperton and Palau, and one haplotype was shared between oceans, occurring in both the Red Sea and Pohnpei (GenBank accession AF107061). Of 564 bp of sequence analyzed for 33 unique Indo-Pacific A. lottini sequences, 119 sites were variable and 21 were phylogenetically informative.
Neighbor-joining analysis of K2P genetic distances among unique Indo-Pacific haplotypes of COI sequences revealed the same two major clades as found in a previous study (Williams et al., 1999). Haplotypes from Clipperton Island and Panama formed an East Pacific clade (Clade II) which differed by 7 fixed transitions from the IWP haplotypes in Clade I (data not shown). The distribution of Clade I and Clade II haplotypes is shown in Figure 1b. All haplotypes found in the Indian Ocean and Pohnpei are from Clade I, whereas those in Panama and Clipperton are from Clade II. Only Palau has haplotypes from both clades.
Measures of genetic variability indicated high levels of diversity within both Indonesian (H = 0.378, mean number of alleles = 4.9, all sites combined) and Ningaloo (H = 0.397, mean number of alleles = 2.4) populations of L. laevigata. Observed average heterozygosities were close to those expected under conditions of HWE, with all expected values falling within one standard error of the observed values. Chi-square tests using exact significance probabilities showed no significant deviations of genotype frequencies from those expected under conditions of HWE.
Samples from Indonesia and Ningaloo both clustered with Pacific and West Australian samples; pairwise genetic distances for both Indonesia and Ningaloo were highest when comparisons were with South Africa and Thailand. The new specimens from Ningaloo were all blue in color like those from the southwest Pacific. Samples from Indonesia showed a similar range in color morphs to those found in other Indo-Malay populations. These included a mixture of blue and gray-blue morphs, some of which had orange under-surfaces, and more rarely pink or apricot on both surfaces. Fst values were consistent with those found in previous studies, suggesting a break between the Indian and Pacific oceans with both Ningaloo and Indonesia more similar to Pacific populations than to either Thailand or South Africa (Table 2). The locus which showed the greatest difference between oceans was LT-1, as in previous studies. Allele distribution at this locus in Indonesia and Ningaloo samples was similar to that observed in Pacific populations, although no rare alleles were found in the Ningaloo sample, probably because of the small sample size (Fig. 2a). Allele frequencies for new samples are listed in Appendix 1b.
The new COI sequences from the Seychelles (GenBank AJ493187–AJ493190) clearly belong to the same clade as sequences from South Africa and Thailand (Clade 2 in Williams, 2000) (data not shown). The distribution of haplotypes belonging to Clades 1 and 2 is shown in Figure 2b. Only haplotypes from Clade 1 are found in South Africa, Thailand and the Seychelles, whereas only haplotypes from Clade 2 are found in eastern Australia and Fiji. Haplotypes from both clades are found in western Australia and north-west Pacific populations.
In contrast to a previous study (Williams et al., 1999) our new COI data for the snapping shrimp Alpheus lottini show no evidence of a break between Indian and West Pacific populations. These two ocean populations were thought to be distinguished by a single nucleotide: Indian Ocean samples have a “C” at position 90 whereas old Pacific samples have an “A” or “G” (Williams et al., 1999). However, all new samples from widespread locations throughout the Pacific have a “C” at position 90 like Indian Ocean samples. A large number of COI pseudogenes have been found to occur in several species of Alpheus (Williams and Knowlton, 2001), and some of the pseudogenes observed in Alpheus are so similar to the functional mitochondrial sequence that it has not been possible to distinguish them a priori. Since the sequences in this study were not the same as those in Knowlton and Weigt (1997), even though some of the same individuals were used in both studies, we believe that one data set is in error. In particular, the presence of a “fixed substitution” between Indian Ocean and Pacific Ocean populations appears to be doubtful. We believe that the sequences, and therefore the genetic pattern observed in the present study (i.e., no evidence of an IWP break), are correct for two reasons. Firstly, sequence chromatograms used in this study did not show any of the ambiguities associated with observed Alpheus pseudogenes. Secondly, Williams and Knowlton (2001) were able to successfully amplify functional copies of COI from cDNA, and one COI haplotype from this study was identical to that obtained previously from a Panamanian sample of cDNA (with a “C” in position 90, GenBank accession AF309910, Williams et al., 2001). An analysis of the entire COI data set, including the putative pseudogenes, shows that they are not obviously differentiated from the new sequences, falling into the expected clades based on their geographic location.
Indo-West Pacific breaks in Alpheus lottini and Linckia laevigata
Over the past decade there have been a growing number of broadscale studies looking at intraspecific genetic variation in a variety of widespread marine species that occur throughout the IWP, and many have revealed evidence of a genetic break between Indian and Pacific Ocean populations (e.g., damselfish, Lacson and Clark, 1995; coconut crab, Lavery et al., 1995, 1996; deep sea fish, Miya and Nishida, 1997; mangrove tree, Duke et al., 1998; starfish Linckia laevigata, Williams and Benzie, 1998; crown-of-thorns starfish, Benzie, 1999; tiger prawn, Duda and Palumbi, 1999; mud crab, Gopurenko et al., 1999 [based on haplotype distribution]; sea urchin Eucidaris metularia, Lessios et al., 1999; two species of bonefish, Colburn et al., 2001; two species of Diadema sea urchin, Lessios et al., 2001). Some studies on a smaller scale have also shown congruent genetic breaks consistent with an effect of lowered sea level (e.g., tiger prawn, Benzie et al., 1992; sea perch, Elliott, 1996; teleost fish, Chenoweth et al., 1998; mantis shrimp, Barber et al., 2000, 2002; tiger prawn, Brooker et al., 2000; water snake, Karns et al., 2000; giant clam Kittiwattanawong et al., 2001). These within-species patterns mirror the geographic segregation of sister species whose ranges abut at or near the Indonesian archipelago (e.g., siganid fish, Woodland, 1983; strombid mollusks, McManus, 1985; butterfly fish, McMillan and Palumbi, 1995; sea urchin Diadema paucispinum species A and B, Lessios et al., 2001). Many authors have suggested that these concordant patterns may be the result of repeated periods of lowered sea level during the Pleistocene that prevented or restricted dispersal between the Indian and Pacific Oceans for tens of thousands of years (e.g.,Potts, 1984).
The large number of congruent biogeographic patterns in several different taxa is convincing evidence of a vicariant event (Neigel and Avise, 1993, Avise, 1994). However, as the number of wide-scale population genetic surveys has increased, contrary evidence has emerged with some studies showing no indication of a break between the Indian and Pacific Oceans. In some cases this may simply reflect small sample sizes. For example, the goatfish Mulloidichthys vanicolensis has been suggested to show no significant differentiation between Indian and Pacific Oceans (Stepien et al., 1994), but the authors note that three loci have alleles that were found only in the Indian Ocean. Given that only five individuals were sampled from the Indian Ocean, it is possible that a larger sample might show a significant frequency shift between oceans. Studies that show no genetic differentiation between oceans most frequently include those of highly vagile species (e.g., swordfish, Chow et al., 1997, 2000; bigeye tuna Alvarado Bremer et al., 1998) but also at least one benthic invertebrate (sea urchin Diadema savignyi, Lessios et al., 2001).
The two species considered in this study show different degrees of genetic differentiation between the two oceans. Sequence variation from the nuclear gene encoding myosin heavy chain (MyHC) and the mitochondrial COI gene were examined for the snapping shrimp Alpheus lottini from wide-ranging populations throughout the Indian and Pacific Oceans. COI variation was consistent with high gene flow throughout the IWP. Distribution of a single nucleotide polymorphism (SNP) in MyHC, on the other hand, is consistent with a cline between oceans. It is possible to speculate that such a cline may have arisen as a result of recent dispersal between the Indian and West Pacific Oceans after an IWP break between oceans. The fact that mtDNA variation shows no evidence of isolation by distance over the same range, provides some support for this idea, since it is consistent with the idea that present-day gene flow is high in the IWP. However the present sample size is small and additional samples for both data sets are required from the south-west Pacific and Indo-Malay region to determine if this interpretation is correct.
New allozyme data for Linckia laevigata confirm previous findings of a break between populations from the Indian Ocean and those from the Pacific Ocean, the Indo-Malay region and western Australia (Williams and Benzie, 1998). Ningaloo and north-western Australian populations are genetically very similar to each other and to Pacific populations, despite being located in the Indian Ocean. It is likely that the reefs at Ningaloo (but not those of the Rowley Shoals and other reefs on the north-west continental slope) were completely exposed during low sea-level stands (Hatcher, 1991). This suggests that L. laevigata may have only existed in Ningaloo since the last glaciation. The genetic similarity between Ningaloo and north-western Australian populations is explained by the presence of the Leeuwin current which brings tropical water southward from the north-west corner of western Australia down the western coast of Australia (Cresswell and Golding, 1980, Pearce, 1991). Like the allozyme data, COI sequence data are consistent with differentiated populations in each ocean but unlike the allozyme data, suggest that there has been recent gene flow in the Indo-Malay region. Two clades of COI haplotypes are distinguished; one clade is found in the Indian Ocean and the other in the Pacific Ocean with a region of overlap in north-west Australia, the Indo-Malay region and north-west Pacific where both clades co-occur (Fig. 2b). Since any reduction in nuclear gene flow would also have resulted in a reduction in mtDNA gene flow (as evidenced by the presence of two distinct clades), the observed overlap in the Indo-Malay region must be a result of gene flow after the removal of that barrier.
Nuclear vs. mitochondrial evidence for past and ongoing events in the Indo-West Pacific
Mitochondrial DNA is often used to examine variation among populations because it can be highly variable and more importantly, owing to its smaller effective population size, it may show a greater degree of population structure due to genetic drift than nuclear genes. The effects of drift may be overcome by gene flow, but since the effective population size of mtDNA is smaller than that of nuclear DNA, mtDNA will show the effects of either an increase or decrease in gene flow faster than nuclear DNA. Therefore mtDNA is likely to have reached an equilibrium with present-day gene flow such that variation in marine organisms is likely to reflect gene flow along ocean currents (e.g.,Williams and Benzie, 1997).
There are many examples where nuclear and mitochondrial markers do not show the same pattern of genetic structure (Grosberg and Cunningham, 2001). In some cases this is because the nuclear markers, in particular many allozyme loci, are affected by selective pressures, whereas mitochondrial markers are apparently neutral and therefore better represent true patterns of gene flow. For example, the American oyster showed little population structure at allozyme loci (Buroker, 1983) because of balancing selection (Karl and Avise, 1992), but strong population structure in two neutral markers (mtDNA, Reeb and Avise, 1990; anonymous nuclear gene, Karl and Avise, 1992). Another reason for differing patterns of genetic structure is that nuclear markers take longer to reach an equilibrium between drift and migration than mitochondrial markers (Neigel, 1994). For example, allozyme variation has sometimes been shown to reflect patterns that are not consistent with present day ocean currents, but may instead reflect past patterns of dispersal along ocean currents that are now non-existent (e.g.,Benzie and Williams, 1995, 1997). Equally, nuclear genes, represented by allozymes, nuclear sequences and morphological differences, may still be showing some evidence of past historical barriers to dispersal which have not yet come to equilibrium with present-day gene flow (although possibly only at a limited number of loci, Palumbi et al., 2001; Hare, 2001). In the IWP there might be a greater degree of structure at some nuclear markers than mtDNA markers if there is nothing to obstruct present-day gene flow between oceans, but there had been in the past. Such patterns may exist in only a few species because the difference in time to equilibrium is less when gene flow is high (Neigel, 1994). Therefore in highly dispersive organisms the pattern at both markers would be similar.
The data for Linckia laevigata suggest that this may be one example where a difference in genetic patterns reflects differences in the rate at which markers have come to equilibrium. Allozyme variation showed evidence of a significant genetic break between populations in the Pacific Ocean (including the Indo-Malay region and western Australia) and the Indian Ocean, notably at a single nuclear locus (LT-1) (Williams and Benzie, 1998; this study). MtDNA variation, on the other hand, showed evidence of recent dispersal between Indian and west Pacific populations, with populations in the Indo-Malay region and north-west Australia showing a mixture of both Indian and Pacific Ocean haplotypes consistent with recent gene flow across the IWP ‘break’ (Williams and Benzie, 1998; Williams, 2000, this study). Williams and Benzie (1998) discuss the possibility that selection may be affecting the distribution of allozyme alleles, but dismiss it as unlikely given that the alleles are distributed more or less evenly within each ocean and separate into two groups at a recognized biogeographic boundary.
In Alpheus lottini there was also weak evidence of greater genetic structure in the IWP in nuclear rather than mitochondrial DNA. Sequence from COI suggests that there is high gene flow throughout the IWP, but variation at a single nuclear gene shows a cline across the Indo-Malay region (albeit with small samples). Populations of yellowfin tuna are also significantly structured between Indian and West Pacific Oceans at a single allozyme locus and show congruent, but more limited, structure using mtDNA variation (Ward et al., 1997). Some smaller-scale studies along the northern coast of Australia (which was divided by Pleistocene land bridges between Australia and New Guinea) have also noted the fact that allozymes differentiated populations more strongly than mtDNA variation (Elliott, 1996).
Morphological differences between populations in the two oceans have been noted for several species (e.g.,Suzuki, 1962; Stepien et al., 1994; Lacson and Clark, 1995; Williams and Benzie, 1998; Benzie, 1999, this study). Since any genetic change in color or morphology must require at least one and possibly a suite of nuclear genes, these distribution patterns may in some cases reflect genetic differentiation between oceans, especially if populations within one ocean are similar, and differences are limited to comparisons between oceans. Morphological differences between oceans can be found both in those organisms which show evidence of population genetic structure, as well as those that do not. For instance, bigeye tuna show no evidence of population structuring using mtDNA variation (Alvarado Bremer et al., 1998; Chow et al., 2000) but show some differences between oceans when blood types (Suzuki, 1962) and morphology (Kume et al., 1971) are considered. Equally, Alpheus lottini from Chagos were pale relative to Pacific morphs (N.K. unpub. data), although genetically not distinguished by COI sequence data. The distribution of color morphs in populations of both L. laevigata (Williams and Benzie, 1998) and the crown-of-thorns starfish (Benzie, 1999) are consistent with genetic differentiation between the oceans.
There are still only a limited number of broadscale IWP studies using both nuclear and mtDNA genes, but in some cases mitochondrial and nuclear variation show the same pattern of population structure. In the coconut crab, fixed differences between Indian and Pacific Oceans were found using both allozymes and mtDNA sequences, and mtDNA also showed reciprocal monophyly of haplotypes (Lavery et al., 1995, 1996). This suggests that present day gene flow is still very low between oceans, possibly because of reduced population size resulting from overfishing. Populations of swordfish, on the other hand, showed no variation using either mtDNA (Rosel and Block, 1996; Chow et al., 1997; Chow and Takeyama, 2000) or nuclear markers (Chow and Takeyama, 2000). This suggests that present-day gene flow in this species has been sufficiently high to bring both nuclear and mtDNA markers to a new equilibrium, or that swordfish dispersal was not affected by lowered sea levels.
Location and age of boundary between Indian and Pacific Oceans
An intriguing difference has been found in the location of biogeographic boundaries for those species that showed a break between oceans. Western Australian populations were sometimes found to have strong affinities with those in the Pacific and other times to be either a mixture of Pacific and Indian Ocean variation, or more like populations from the Indian Ocean. For example, all three echinoderm species collected from north-western Australia that showed an IWP break had strong affinities with Pacific Ocean samples (Linckia laevigata, Williams and Benzie, 1998; crown-of-thorns starfish, Benzie, 1999; Diadema setosum, Lessios et al., 2001).
Links between western Australia and the south-west Pacific were greater in the past than they are at present. For instance, 9 MYA the Indonesian-Australian seaway was wider, allowing for free interchange between the Pacific and Indian Oceans (George, 1997). Later, as New Guinea moved north about 5 MYA, the seaway closed and the source of the Indonesian throughflow changed from the south-west to the north-west Pacific (Cane and Molnar, 2001) limiting direct dispersal between the Great Barrier Reef and other south-west Pacific populations and western Australia. However dispersal was still possible between the Indo-Malay region and Western Australia throughout the entire Pleistocene, albeit at a reduced rate, via the Indonesian throughflow (Williams and Benzie, 1997, 1998; Benzie, 1999; Lessios et al., 2001).
Although the Indonesian throughflow continued throughout the Pleistocene, it is thought to have been reduced during periods of glaciation and this in turn may have weakened the Leeuwin current, preventing it from warming the western coast of Australia (Wells and Wells, 1994; Godfrey, 1996; Takahashi and Okada, 2000). Patches of cold surface-water also resulted from upwellings off southern Java (Martínez et al., 1999; Takahashi and Okada, 2000) and north-western Australia (Wells and Wells, 1994; Takahashi and Okada, 2000) during Pleistocene glacial maxima. The combined effects of these upwellings and changes to ocean currents may have been to restrict larval dispersal of temperature-sensitive animals between oceans (Fleminger, 1986; Williams and Benzie, 1998). This explains the affinity some west Australian populations have with the Pacific, since the cold water would have separated north-western Australia from the rest of the Indian Ocean, but not the Indo-Malay region (Williams and Benzie, 1998).
Different biogeographic boundaries between the Indian and Pacific Oceans may have arisen in different species if Pleistocene cold water upwellings varied seasonally in intensity. Those species with reproductive seasons co-incident with the period of greatest upwellings would have been the most affected by these barriers (e.g., copepods, Fleminger, 1986), whereas other species may have been little affected. Alternately some of the differences in biogeographic patterns observed among species may simply reflect different vicariant events. For instance, the magnitude of DNA divergence between Indian and Pacific Oceans is not the same for two species of Diadema sea urchins that show an IWP break (Lessios et al., 2001). The authors suggest that, whilst both species show divergences that date within the Pleistocene, the population genetic structure of each species was affected by different low-sea level stands.
The biogeography of the IWP is, in part, the product of many different glaciations and tectonic events. Not only were there many periods of lowered sea level throughout the Pleistocene, but the movement of continental plates throughout the Miocene, in particular the collision of the Australia/New Guinea plate with the Asian plate, altered oceanographic circulation patterns in the Indo-Malay region (George, 1997). These changes would have had a profound effect on genetic connections among marine populations. So far, most molecular studies have suggested that IWP intra-specific genetic variation is the result of Pleistocene events. However many (but not all) IWP speciation events have been placed prior to the Pleistocene (e.g., spiny lobsters, George, 1997; bony fish, Colborn et al., 2001; periwinkles, unpublished data, S. Williams, D. Reid and T. Littlewood) (see also Avise and Walker, 1998; Avise et al., 1998). It is possible that isolation between oceans due to Pleistocene glaciations were of sufficiently limited duration as to result most often in changes only at the population level and less often to have resulted in allopatric speciation. Earlier geological events may have had more profound effects on species evolution resulting in speciation events.
From the Symposium Integrated Approaches to Biogeography: Patterns and Processes on Land and in the Sea presented at the Annual Meeting of the Society for Integrative and Comparative Biology, 2–6 January 2002, at Anaheim, California.
Samples of Alpheus lottini were provided by J. Maté (Clipperton Atoll) and P. Davis (Cartier Reef, Timor Sea; Queensland Museum accession numbers W17600 and W17634). Samples of Linckia laevigata were provided by J. Benzie (Bali and Lombok, Indonesia), B. Burnett (Seychelles), C. Christie, W. Oxley and A. Robertson from the Australian Institute of Marine Science (AIMS) monitoring team (Ningaloo) and S. Seddon (Melintang, Indonesia). We are especially grateful to J. Benzie and M. Johnson for the use of their laboratories for allozyme electrophoretic work and DNA extractions respectively. Thanks to B. Ballment and L. Peplow at AIMS for arranging shipment of specimens to Panama. D. Reid, M. Hare and one anonymous reviewer offered valuable criticisms of the manuscript. The Smithsonian Institution provided much of the support for this project.