Potential Alternative To Standard Tagging Techniques Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Genetic tagging of individuals is emerging as a potential alternative to standard tagging techniques that allows researchers to ask questions in relation to contemporary patterns of genetic divergence, population size, and gene flow Palsboll 1999. Unlike many tagging methods, genetic markers (tags) have the added benefit of existing in all animals and being permanent (Palsboll et al. 1997), which is a vital assumption of capture-mark-recapture (CMR) techniques (Seber 1982). Thus, microsatellite data lend themselves to abundance estimation in a similar fashion to traditional identification methods (Palsboll et al. 1997).

Tissue samples are required prior to DNA extraction and analysis, which can be obtained either using intrusive or non-intrusive techniques. A number of intrusive techniques were detailed in Chapter 2 of this report; however, non-intrusive methods can also be used under certain conditions (e.g., Lukacs & Burnham 2005). Obtaining consistent samples using non-intrusive methods in the ocean can be problematic and often in these situations the DNA in samples may be degraded, which can in turn lead to analysis problems such as 'allelic drop out' (Palsboll 1999). Non-intrusive sampling may also lead to an insufficient quantity of DNA to carry out analysis (Bilgmann et al. 2007). To avoid these potential problems, the use of biopsies (Lambertsen 1987) have been adopted for tissue sample collection of large marine animals (Parsons et al. 2003) and are recommended for whale sharks (Chapter 2)

Tissue sampling and molecular analysis have not been attempted to date for whale sharks, therefore information pertaining to gene flow between whale shark populations and parent-offspring relationships are unknown. Additionally, microsatellite markers provide a validation technique for current photo-identification analyses, strengthening demographic estimates. The aim of this section is to describe the methods used to isolate microsatellite-containing fragments from whale sharks DNA collected at Ningaloo Reef.

3.2.2 Methods Construction of enriched microsatellite library

We employed the strategy used by (Kandpal et al. 1994) to isolate microsatellite-containing fragments (Figure 3.). In this procedure, genomic DNA fragments containing the desired repeats are hybridized to a repeat probe that has been biotinylated. These hybrid fragments are subsequently captured by a solid matrix to which avidin is covalently bound. Non-specifically DNA fragments are eliminated by washes, and the repeat-containing fragments are eluted and cloned to produce a library. This library should contain 20-90% repeat-containing fragments.

3.2.3 Results

First attempts to use this technique did not give any interpretable results (Figure 3.2). It appeared as if all the whale shark DNA, with or without any biotinylated probe, had been fixed non-specifically to the Vectrex avidin D matrix. For this reason, we decided not to clone PCR product and to make a second attempt with fresh DNA (Figure 3.3). We also decided to change the Vectrex avidin protocol and use a different buffer and try a second elution at 85°C. Thus, we extracted DNA from 4 individuals and pooled the DNA to obtain 5µg of fresh DNA.

Figure 3.2. Last PCR results before cloning P1, P2, P3: probe 1, 2 or 3 - : control without probe.

These were encouraging and we attempted enrichment for twelve different microsatellite motifs: CA-, AAC-, TACA-, TAGA-, GA-, ATG-, AAAC-, CATC-, AAG-, AAT-, AAAG- and CAGA-. The libraries yielded microsatellites as follows:

CA- seven out of eight sequences contained a microsatellite AAC- one out of nine sequences contained a microsatellite TACA- zero out of nine sequences contained a microsatellite TAGA- four out of nine sequences contained a microsatellite GA- six out of nine sequences contained a microsatellite ATG- one out of eight sequences contained a microsatellite AAAC- one out of nine sequences contained a microsatellite CATC- zero out of nine sequences contained a microsatellite AAG- one out of nine sequences contained a microsatellite AAT- zero out of nine sequences contained a microsatellite AAAG- one out of nine sequences contained a microsatellite CAGA- three out of eight sequences contained a microsatellite

Based on these results we obtained sequences from an additional set of 64 clones, drawn from three of the libraries that yielded microsatellites as follows: CA- 20 out of 20 sequences contained a microsatellite TAGA- 11 out of 21 sequences contained a microsatellite GA- 17 out of 22 sequences contained a microsatellite

Sequences were then examined to identify duplicates that might be present in opposite orientation, or which had not been noted upon examination of the electropherograms. In total, we identified 73 different microsatellite-containing clones from the three libraries. We designed PCR primers for 54 microsatellite-containing clones that were designed using DesignerPCR version 1.03 (Research Genetics, Inc.).

The final mirosatellites selected for the library and provided with design-tested primers are shown in Appendix 1.

3.2.4 Conclusion

Despite initial problems, the laboratory analysis protocols and microsatellite libraries have now been established for genetagging of whale sharks. It now remains to collect biopsy samples from large numbers of both the Ningaloo and other Indian Ocean populations firstly to validate photo-identification techniques and secondly, to begin to establish patterns genetic divergence, population size, and gene flow of whale sharks in the Indian Ocean region.


3.3.1 Introduction

The vastness of Earth's oceans may often conceal regional biological processes particularly for pelagic and highly migratory species. For example, many sharks and tunas mature and forage far from shore. Other species like pinnipeds and sea turtles may approach continental or island shores only occasionally to breed or rest. Moreover, many large marine vertebrates often have complex migratory behaviours that vary with age and sex (Brown et al. 1995, Craig & Herman 1997, Hughes et al. 1998, Bowen et al. 2005, James et al. 2005, Carlsson et al. 2007).

Though the natural histories of many pelagic migrants have become better known during the past few years, little is still known about the biology and biogeography of whale sharks (Rhincodon typus). Whale sharks appear to be widely distributed in tropical and warm temperate seas (30°N and 35°S) except, perhaps, in the Mediterranean (Compagno 2001). Most information about general distribution, however, is either from seasonal sightings in scattered locations or anecdotal observations (Colman 1997). Aggregations of whale sharks have been routinely reported off Ningaloo Reef (Australia), Gladden Spit (Belize), Yucatan peninsula, Baja California (Mexico), India, Taiwan, Japan, and the Philippines (Taylor 1996, Clark & Nelson 1997, Colman 1997, Taylor & Pearce 1999, Heyman et al. 2001, Wilson et al. 2001a, Stewart & Wilson 2005, Wilson et al. 2006). Some aggregations occur year-round while others may be associated with seasonal abundance of prey. Most known aggregations are immature sharks and segregation by size and sex may occur in some areas (Colman 1997, Compagno 2001). Even though recent studies have demonstrated the remarkable ability of this species to migrate long distances (e.g., Colman 1997, Compagno 2001, Eckert & Stewart 2001, Wilson et al. 2006) it is not clear whether whale shark populations are panmictic or composed of reproductively isolated subpopulations. Recent indication for tolerance of cold water when diving (Wilson et al. 2006) suggests that temperate and perhaps even sub-polar waters may not be impediments to movements of whale sharks across thermal boundaries. Here, we present the results of a study of the population genetics of this widely distributed marine megavertebrate using sequences from the mtDNA control region (CR) to assess the potential population relationships among ocean basins.

3.3.2 Materials and methods Sample collection and laboratory procedures

Skin biopsy samples were collected from 50 whale sharks (Figure 3.4) when they aggregated seasonally in the Gulf of California or Western Australia or were found stranded ashore from 1995 through 2005 at other sites and then preserved in either salt saturated DMSO solution or 95% ethanol and stored at room temperature.

We extracted total genomic DNA using a phenol-chloroform-isoamyl alcohol protocol (Sambrook et al. 1989) or 5% Chelex non-boiling protocol (Walsh et al. 1991)The mitochondrial CR was amplified using primers developed within the tRNAPro (WSCR1-F: 5′­TTGGCTCCCAAAGCCAAGATTCTTC-3′) and tRNAPhe (WSCR1-R: 5′­TTGTAACCAAAATTATACATGC-3′). Because of the large size of the CR (~1,100 - 1,325 nucleotides), two internal primers were designed to facilitate sequencing of the whole region. Primer WSCR2-R (5′-CTTAATATTTATTGTTCCTGGTTTCAGTT-3′) was paired with WSCR1-F, and primer WSCR2-F (5′-CTATAATTGATTTAAACTGACATTTG-3′) was paired with WSCR1-R producing two, overlapping fragments approximately 950 and 700 bp respectively. Amplification reactions were carried out in 50 µL volumes consisted of 1X Promega buffer (Promega, Madison, WI, USA), 1.25 U of IDProofTM DNA polymerase (ID Labs Inc., Ontario, Canada), 0.8 mM dNTPs, 2 mM MgCl2, 0.5 μM of each primer, 6.0 μg bovine serum albumen, and 1 - 3 μL of template. Cycling conditions for all primer pairs consisted of 95°C 1 min, 35-40 cycles of 95°C 45 sec, 58°C 60 sec, and 72°C 90 sec with a final extension at 72°C for 7 min. Amplicons were purified with QIAquick kit (Qiagen, Valencia, CA, USA) following the manufacturers instruction. Both strands were sequenced using an ABI 3730XL Genetic analyzer (Applied Biosystems, Inc., Foster City, CA, USA). Data analysis

Control region alignments were optimized in Sequencher 4.1 (Gene Codes, Ann Arbor, MI, USA) and gaps were introduced to maximize sequence similarity. Analyses were done both including and excluding ambiguous bases and missing data (i.e., gaps). In some analyses, contiguous gaps were treated as single events by omitting all but one of the gaped bases, and gaps were weighted as transitions. In the case of substitutions within gaps, variable positions were retained and gaps were weighted as transversions. The Akaike Information Criteria within ModelTest v3.06 (Posada & Crandall 1998) was used to determine the best-fit model of evolution. Phylogenetic analyses were done using PAUP* 4.0b10 (Swofford 2003). Gene tree reconstruction was performed using neighbor-joining algorithm (Saitou & Nei 1987), with the optimal distance model identified with ModelTest. Statistical support for the nodes was estimated with 100 non-parametric bootstrap replicates (Felsenstein 1985).

Summary statistics (number of haplotypes, haplotype frequencies, number of polymorphic sites, number of transition and transversions, and nucleotide composition) were estimated in ARLEQUIN 3.0 (Excoffier et al. 2005). Individuals were binned into five groups defined by geographical region: Gulf of Mexico/Florida (N = 17) in the northwestern Atlantic; South Africa (5) and Australia (12) in the Indian Ocean; Philippines/Taiwan (7) in the northwestern Pacific; and Gulf of California (8) in the northeastern Pacific. Genetic diversity within localities was measured as the number of DNA mitochondrial haplotypes, haplotype diversity (h), and nucleotide diversity (Ï€) estimated with Nei's corrected average genetic divergence (Nei 1987) incorporating Tamura & Nei's (1993) model of sequence evolution with ARLEQUIN.

We used mismatch distributions for each sample to distinguish between population growth models, especially those invoking past exponential growth and historical population stasis (Slatkin & Hudson 1991, Rogers & Harpending 1992). Population paramaters τ, θ0, and θ1 were obtained from ARLEQUIN, where τ is the mutational timescale, and θ0 and θ1 are the expected pairwise differences before and after a change in population size (growth or contraction), respectively (Harpending 1994). The mutational timescale is τ = 2µt, where t is measured in generations and µ is the mutation rate per generation for the entire sequence (µ = mTµ, where mT = number of nucleotides and µ = mutation rate per nucleotide). The expected pairwise differentiation is θ = 2Nfµ where Nf is the effective female population size. Tests for selection also can indicate population expansion and here we apply the algorithms of Tajima (1989) and Fu (1997).

Population subdivision and structure were estimated using an analysis of molecular variance (AMOVA, Excoffier et al. 1992), and pairwise population ΦST significance test (Cockerham & Weir 1993) as implemented in ARLEQUIN. Significance of ΦST was determined via nonparametric permutation (Excoffier et al. 1992) with 1,000 data permutations. For AMOVA analyses, we used the distance matrix generated by the model selected with ModelTest (HKY85+I). Population differentiation also was tested using the Raymond and Rousset test based on haplotype frequencies (Raymond & Rousset 1995).

3.3.3 Results

The mitochondrial CR from a total of 50 individuals ranged from 1,143 to 1,332 bp with a mean of 1,236 bp. Nearly all of this size variation was due to indels composed of repeated sequence blocks (Figure 3.5). Considering just the repeat unit structure (i.e., ignoring site substitutions) there were 11 different repeat motifs in the whale shark. Repeated blocks ranged in size from 9 bp (block A) to 64 bp (block E) long. All haplotypes had regions A1 to D1, E2, F2 E3, and F3 to J3 and this was the motif for the smallest haplotype, H18. The largest haplotype, H9, had all the common repeats, some less common ones, and was the only haplotype to have block I1. Haplotypes H10 and H11 were similar to H18 except they possessed blocks E1 and F1 (totaling 103 bp) making H10 and H11 the second largest haplotypes.

We also found substitutions between repeated blocks within the same sequence. For example, repeat A1 differed from A2 by a substitution of one nucleotide in haplotype H22. Other examples include substitutions shared between different haplotypes like block B, which was repeated twice in nearly all haplotypes. For some haplotypes these were perfect repeats whereas there were single transitional changes in others. Clearly, both larger indel changes and smaller substitutional changes are common in the evolution of whale shark CR.

To maximize sequence similarity among all sampled sharks, the complete DNA sequence alignment required multiple gaps of sizes ranging from 1 to 163 bp. There were 55 polymorphic sites, with 35 substitutions (32 transitions and 3 transversions) and 27 gaps resolving 28 haplotypes. Fifteen of those gaps were coded as single nucleotide transitions, while the other 12 were coded as transversions due to substitutions in those regions. Of the fifty-six evolution models tested by ModelTest using the Akaike Information Criteria (AIC), the HKY85+I model (Hasegawa et al. 1985) was selected as the best fit with the proportion of invariable sites I = 0.9292, and base frequencies of A: 0.3487, C: 0.1991, G: 0.1102, and T: 0.3421. Overall, the haplotype diversity (h), and nucleotide diversity (π) were relatively high with h = 0.90 - 1.0 and π = 0.007 - 0.016 (Table 3.1). Among the 28 observed haplotypes, only seven occurred in more than one shark (Table 3.2). Three of those shared haplotypes occurred in a single geographic region and four occurred in four of the regions. Except for some of the Gulf of Mexico haplotypes, there appears to be no phylogenetic clustering (Figure 3.6). AMOVA with HKY85+I distances assigned 87.05% of the genetic variability within and 12.95% among locations. There is statistically significant structure in whale shark populations, with overall ΦST = 0.13 (P < 0.005). The Atlantic population appears to be significantly different from all but the South Africa population (Table 3.3). Moreover, there appears to be divergence only between the Atlantic and the Australian and the Atlantic and northwestern Pacific populations using a test of exact population differentiation based on haplotype frequencies (Raymond & Rousset 1995).

The mutational timescale τ = 2µt can be used to estimate coalescence times for populations if generation time and mutation rate (µ) are available. Moreover, the initial and current effective population sizes (Nf0 and Nf1) can be estimated from the pairwise differences θ0 and θ1, if mutation rate is available or estimated. Based on the observation of an adolescent female with an osteological age estimate of 20 years (Wintner 2000), we provisionally apply a generation estimate of 25 years. The control region clock for hammerhead shark, Sphyrna lewini, is 0.8% divergence between lineages per million years (Duncan et al. 2006) and is similar to a rate derived from lemon sharks control regions (Negaprion brevirostris; J. Schultz, pers. comm.). In contrast, Keeney and Heist (2006) report a rate of 0.4% per million years for control region in the blacktip shark Carcharhinus limbatus. We provisionally apply both rates to whale sharks, with the caution that these three species are tens of millions of years divergent from R. typus. Results in Table 3.4 indicate coalescence times on the order of 630,000 - 1,250,000 years ago (early Pleistocene), founding effective population sizes of Nf0 = 9 - 17 individuals, and current effective population size Nf1 = 145,200 - 290,600 individuals.

3.3.4 Discussion

Our survey of whale sharks indicates unusual size polymorphism in the CR, significant population structure between Atlantic and Indian-Pacific ocean basins, and coalescence times on the order of 1 my. Before interpreting these results, we address two prominent caveats:

1) Sample size is small and lapses in coverage include the South Atlantic, Central Pacific, and South Pacific. Sample size clearly limits inference. Consequently, we have tempered our corresponding conclusions. There are no directed oceanic surveys for whale sharks, as there are for tunas, billfish, and sea turtles, and the species occurs at low density even in regional aggregates. The sample size of 50 represents ten years of directed effort on our part, and is the only genetic evaluation of this rare and enigmatic species. Nonetheless, though a larger sample size and more complete global sampling may increase the number and frequency of haplotypes, the sharing of haplotypes (H1 and H3) among multiple sharks at the extremes of the geographic range (NW Atlantic and NE Pacific) will not change.

2) Estimates of coalescence times and effective population sizes are based on tenuous calibrations of generation time and mutation rate, and the latter are derived from distantly-related sharks. The mutation rate and generation time are simple estimates based on few data, and should therefore not be interpreted quantitatively. Shark mtDNA evolution, however, appears to evolve about an order of magnitude slower than for bony fishes (Martin et al. 1992), which is consistent with our clock estimates used here. Consequently, we think that corresponding estimates are useful in a qualitative sense for determining whether (for example) population histories coalesce at 104, 105, or 106 years. Control region morphology

The CR in whale sharks (1,143 - 1,332 bp) is larger than observed in most elasmobranchs. Stoner et al. (2003) amplified this region in 52 species of elasmobranches and products were 1030-1050 bp long except for the barn door skate, Dipturus laevis, which was ~1200 bp long. Other studies revealed a CR smaller than the whale shark (Squalus acanthias - 1080 bp, Rasmussen & Arnason 1999; Mustelus manazo - 1,068 bp, Cao et al. 1998; Heterodontus francisi

- 1,068 bp, Arnason et al. 2001; Scyliorhinus canicula - 1,050 bp, Delarbre et al. 1998), or comparable to the smallest whale shark CR; Carcharodon carcharias - 1,146 bp, (Pardini et al. 2001). Variation in size in the CR of whale sharks is also higher than reported for other sharks (Kitamura et al. 1996, Pardini et al. 2001, Keeney et al. 2005), with a 189 bp difference between the largest and the smallest amplicon.

Variation in the size of the control region has been reported for a substantial number of bony fishes (Lee et al. 1995, Brown et al. 1996, Fujii & Nishida 1997, Bentzen et al. 1998, Hoarau et al. 2002, Rokas et al. 2003, Tsaousis et al. 2005). It is typically comprised of tandem repeats, as we observed in whale sharks (Figure 3.5). Our initial attempts to PCR amplify the CR of whale sharks using a variety of published shark primers failed, probably due to the highly duplicated nature of the CR. Because the rate and pattern of these mutations is unknown, most studies have not used size variants as population markers Insertions and deletions of repeat blocks may be relatively common, and homoplasy (convergence on the same number of repeats) is likely to confound any genealogical analysis. Genetic diversity and effective population size

Despite an apparent decline in both catch rates and sighting of whale sharks in various regions (e.g., Stewart & Wilson 2005, Theberge & Dearden 2006, Bradshaw et al. 2007), there is still relatively high genetic diversity in the species. Threatened and endangered species are expected, however, to retain historical levels of genetic diversity if the decline has occurred only recently (Roman & Palumbi 2003, Bowen et al. 2006). In the only other global surveys of shark CRs, the blacktip shark yielded h = 0.75 - 0.81 and π = 0.0020 - 0.0021 (Keeney et al. 2005), and the scalloped hammerhead sharks had h = 0.80 and π = 0.013 (Duncan et al. 2006), compared to h = 0.90 - 1.00 and π = 0.007 - 0.016 for whale sharks. These values are low compared to teleost fishes, but such low values of haplotype and nucleotide diversity are observed among many shark species and when using a variety of mtDNA assay methods (cf. Heist 1999, 2004).

The relatively high diversity in whale sharks is surprising, given that the other two globally distributed sharks are common and abundant coastal species, whereas whale shark aggregations are generally small and uncommon. Two general processes might contribute to the relatively high haplotype and nucleotide diversity observed in whale sharks: 1) secondary contact between divergent allopatric lineages or 2) large stable populations. Except perhaps for haplotype H9, the mtDNA phylogeny reveals no evidence of distinct evolutionary lineages that now occur in sympatry. Hence the inference of a large, historically stable population (Nf ~ 200,000) deserves special attention. Although the population size of whale sharks is unknown, though suspected to be declining, it is possible that whale sharks have maintained demographically stable populations until the active fishing for them began very recently. Our coalescence analysis, although tentative, indicates that the most recent common ancestor was around 1 my ago and that genetically effective population size of females was approximately an order of magnitude larger than the current estimate (Table 3.4). This outcome is further supported by the mismatch distribution indication of relatively stable, large populations. Moreover, new whale shark habitats continue to be discovered; in recent years a number of seasonal feeding aggregations have been documented near continental coastline and island habitats (e.g. Rowat & Gore 2006).

The large effective population size may mean that the transient surface feeding aggregations that are most often observed are not the principle habitats of adult whale sharks. Recent studies have demonstrated that at least some whale sharks spend most of each year distant from those coastal sites and often at relatively great depth in cold water (Wilson et al. 2006, Wilson et al. 2007). Although whale sharks appear to lack the anatomical, physiological and behavioral adaptations to conserve heat, the large body mass of adults may provide sufficient thermal inertia to allow extended cold-water exposure (Sims 2003, Wilson et al. 2006). Regardless of the extent of geographic and vertical population movements, it is clear that much of the habitat for this species is still unknown, and population sizes may indeed be considerably larger than expected (Nf = 22,000 - 67,200). Population structure

Recent satellite tracking has discovered substantial vagility in whale sharks (Gunn et al. 1999, Eckert & Stewart 2001, Eckert et al. 2002, Wilson et al. 2006). Like traditional tag-recapture studies, satellite tracking provides generally only short-term data and allows limited inference about movements, habitat range, and inter-population exchanges during the shark's life span and is not conclusive about the boundaries of stocks or evolutionary significant units (Moritz 1994, Vogler & Desalle 1994, Waples 1995). Assessing patterns of genetic variation can supplement, enhance, and extend an understanding of population movements, illuminate cryptic evolutionary partitions, and inform management plans. Our studies of mtDNA of whale sharks indicates a population partition between Atlantic and Indian-Pacific ocean basins that might not be easily discovered by electronic tracking of small numbers of sharks.

Our genetic studies indicate that whale shark aggregations from some ocean basins are substantially interconnected. Because our samples were collected from seasonal feeding aggregations, we cannot yet say, however, whether this pattern is due to interbreeding and gene flow among populations or just physical mixing of sharks from different populations in feeding areas. In any event, the high haplotype diversity that we detected is unexpected for multiple sampling of the same evolutionary unit.

Whale shark population structure is low, even against the standards of large migratory fishes. Bluefin tuna (Thunnus thynnus) show subtle (ΦST = 0.013) but significant population structure between western Atlantic (Gulf of Mexico) and the Mediterranean, separated by ~11,000 km (Carlsson et al. 2007). The sailfish (Istiophorus platypterus) also is divided among ocean basins with additional significant population structure also within the Pacific Ocean (Graves & McDowell 2003). Blue marlin lack subdivision within ocean basins, but are clearly divided among (ΦST = 0.217, Buonaccorsi et al. 2001). Marine mammals show similar patters of inter-ocean differentiation. Humpback (Megaptera novaeangliae, Baker et al. 1994), minke (Balaenoptera acutorostrata, van Pijlen et al. 1995), fin whales (Balaenoptera physalus Berube et al. 1998), and Cuvier's beaked whales (Ziphius cavirostris, Dalebout et al. 2005) all have pronounced inter-ocean subdivision and some division within an ocean basin between hemispheres. An interesting contrast to these examples is the sperm whale (Physeter macrocephalus); a true cosmopolitan species found in all ocean basins including polar regions. Population structuring in the sperm whale (GST = 0.03) is markedly less than that seen in the previously mentioned fish, whales, and whale shark and was only statistically significant among ocean basins. Interestingly, this was only true for the mtDNA but not for nuclear DNA presumably due to inter ocean migration by males. Barriers to movement between ocean basins generally appear to be stronger for marine mammals and large, pelagic fishes than for whale sharks. These comparisons indicate that large pelagic domains can be population barriers to many highly mobile fishes, whereas the only apparent barriers to whale sharks may be geographic and possibly thermal (see below). Marine phylogeography

In recent years there has been renewed interest in the biogeographic barrier between the Indian and Pacific Oceans, apparently due to substantially lower sea levels during glacial maxima (Barber et al. 2000). While this barrier is consistent with evolutionary separations in small marine invertebrates (Barber et al. 2002), it is a less substantial (albeit significant) population barrier to marine fishes (Bowen et al. 2001, Chenoweth & Hughes 2003, Craig et al. 2007), including sharks (Duncan et al. 2006, Keeney & Heist 2006). Whale shark dispersal ability appears to be unimpeded by this intermittent barrier. This suggests that migratory routes may flexible enough to accommodate newly-submerged habitats, or that connectivity can be quickly re-established after a barrier of several tens of thousands of years. Regardless of where they are going, whale sharks commonly migrate over large areas and reestablishment of connections across newly removed barriers is likely.

The last tropical connection between the Atlantic and Indo-Pacific ended with the rise of the Isthmus of Panama, about 3.5 MY ago (Coates & Obando 1996). In contemporary biogeography, the southern extensions of Africa and South America are regarded as formidable impediments to tropical connectivity. Yet tropical faunas of the Atlantic and Indo-Pacific, including whale sharks, share connections on a scale shorter than 3.5 MY, indicating dispersal around southern Africa (Bowen et al. 1997, Bowen et al. 2001). Recent research indicates that such events are rare, being measured on a scale of 105-106 years (Roberts et al. 2004, Rocha et al. 2005, Bowen et al. 2006).

The cold Benguela Current along western South Africa represents a formidable barrier to the dispersal of tropical fishes into the Atlantic (Gibbons & Thibault-Botha 2002). In a compilation of whale shark stranding and sightings in South Africa, Beckley et al. (1997) confirmed the occurrence of whale sharks along this frigid Atlantic coast. They suggest, however, that sharks arriving from the Indian Ocean succumb to the cold upwelling water and quickly perish. Here the observations on thermal tolerance are pertinent to discussions of inter-oceanic dispersal. Wilson et al. (2006) noted that whale sharks could inhabit cold water, but certainly not indefinitely. A deep cold-water grazing opportunity in the tropics can be balanced with a quick return to warm surface waters. In the Benguela upwelling system, however, surface waters are as cold as deep and no such relief from cold-water excursions is possible in this region, resulting death. Nonetheless, the sharing of haplotypes between Atlantic, Indian, and Pacific locations indicates a relatively recent connection. Whale sharks could have moved between Atlantic and Indian Ocean during hiatuses of Benguela upwelling that occurred between Pleistocene glacial epochs (Chang et al. 1999, Flores et al. 1999). Immediately following each ice age (100K to 400K years, but most recently 10K -20K years ago), tropical plankton appear in sediment cores off southwestern African, indicating an avenue of warm water into the South Atlantic (Peeters et al. 2004). Contemporary movement also is possible. Warm-core gyres from the Indian Ocean occasionally become entrained in the northward moving Benguela Current, feeding into the Central Atlantic (Flores et al. 1999, Penven et al. 2001). In either case, historical or ongoing gene flow is apparently limited, as indicated by the moderate and statistically significant global ΦST ≈ 0.13.

Finally, the sharing of haplotypes may simply be due to the retention of ancestral polymorphisms. We consider this unlikely, given the low phylogeographic signal, multiple shared haplotypes, and pattern of high connectivity. Even so, retention of ancestral polymorphisms is characteristic of large, stable populations, a possibility raised by coalescence analyses. Conservation implications

This first genetic survey of whale sharks indicates significant population structure throughout their global range. Management units for whale sharks may encompass 8,000 km in the Atlantic, and over 16,000 km in the Indian-Pacific ocean basins. Regardless of the potential for cryptic population subdivision, any management plan for whale sharks must consider that feeding aggregations drawn from a broad geographic range area in a single location. Unilateral management in any political jurisdiction will be inadequate for a highly mobile species that may travel through several political jurisdictions. Indeed, data from tracking studies of shark movements and our mtDNA survey both indicate that management plans for the Earth's largest fish will require ocean basin-wide cooperation. Multinational coordination on that scale has proven challenging for tunas and billfish, very difficult for whales, and will likely be very difficult for whale sharks. Given the increase in fishing pressure and the evidence for population declines, the only effective conservation measure may be threatened species status under IUCN guidelines.