The Indo-Malay-Philippine (IMP) biodiversity hotspot, bounded by the Philippines, the Malay Peninsula and New Guinea, is the epicentre of marine biodiversity. Hypotheses to explain the source of the incredible number of species found there include the centre of overlap hypothesis, which proposes that in this region the distinct faunas of the Pacific and Indian Oceans overlap. Here we review the biogeographical evidence in support of this hypothesis. We examined tropical reef fish distributions, paying particular attention to sister species pairs that overlap in the IMP hotspot. We also review phylogeographical studies of wide-ranging species for evidence of lineage divergence and overlap in the IMP region. Our synthesis shows that a pattern of isolation between the Pacific and the Indian Ocean faunas is evident across a wide range of taxa. The occurrence of sister species, with one member in each ocean, indicates that the mechanism(s) of isolation has been in effect since at least the Miocene, while phylogeographical studies indicate more recent divergences in the Pleistocene. Divergence in isolation followed by population expansion has led to an overlap of closely related taxa or genetic lineages in the hotspot, contributing to diversity and species richness in the region. These findings are consistent with the centre of overlap hypothesis and highlight the importance of this process in generating biodiversity within the IMP.
The region bounded by the Philippines, the Malaysian peninsula, New Guinea and northern Australia has long been recognized as a biodiversity hotspot (Stehli & Wells, 1971; Veron, 1993; Randall, 1998; Bellwood et al., 2012; Briggs & Bowen, 2013). The geographical boundaries and shape of this hotspot differ with the viewpoint of the biogeographer, and there have been various names assigned to the region. The Coral Triangle defined by Veron et al. (2009) extends from the Philippines to the Solomon Islands but does not include the Coral Sea, while the Indo-Australian Archipelago defined by Bellwood & Hughes (2001) is inclusive of the Coral Triangle but extends considerably north and south (Fig. 1; see also Fig. 1C in Renema et al., 2008). Regardless of where the exact boundaries are drawn there is wide agreement that this region is the epicentre of marine biodiversity (Cowman & Bellwood, 2013), and herein we refer to it as the Indo-Malay-Philippine biodiversity hotspot (IMP hotspot; Fig. 1). Over 2600 species of reef fishes and 600 species of corals are found in the IMP region (Veron et al., 2009; Allen & Erdmann, 2012). Taxonomic diversity declines with distance from the area, both latitudinally and longitudinally: a pattern identified in corals by Stehli & Wells (1971) and since recognized across a broad array of taxa (Paulay, 1990; Veron, 1995; Briggs, 1999). The generality of this pattern has led many researchers to conclude that a common mechanism may be responsible for the high biodiversity in the region. Several hypotheses have been proposed to explain the species richness in the IMP hotspot and these can be grouped into four major categories: (1) centre of origin, (2) centre of accumulation, (3) centre of survival, and (4) centre of overlap. Contemporary species distributions (and most recently biogeographical modelling; Cowman & Bellwood, 2013) are the most common line of evidence used to test these hypotheses, and, amidst the jockeying of competing models, some have acquired vocal proponents while others receive comparatively little consideration.
First proposed by Ekman (1953), the hypothesis that has received the most attention is that of the centre of origin. Ekman (1953) and others who have followed (Briggs, 1999, 2003) suggest that the high number of species in the IMP hotspot is a product of an unusually high rate of speciation in the region, with new species radiating from this centre of origin. Mechanisms invoked to explain the elevated speciation rate include the fracturing of populations as a result of the geological complexity of the region and eustatic sea-level changes (McManus, 1985). Others have suggested that increased rates of speciation in the hotspot are driven by intense competition (Briggs, 2005; Bowen et al., 2013) and differing selection pressures in a highly heterogeneous environment (Rocha & Bowen, 2008). Several years after Ekman proposed the centre of origin hypothesis, Ladd (1960) countered with the centre of accumulation hypothesis. This model proposes speciation in peripheral locations with subsequent dispersal of novel taxa into the IMP hotspot. The long history of the Pacific archipelagos, some of which date to the Cretaceous, the possibility of isolation in these peripheral habitats, and ocean current and wind patterns that favour dispersal towards the IMP region have been suggested as mechanisms in support of this hypothesis (Ladd, 1960; Jokiel & Martinelli, 1992). Still others suggest that the habitat-rich IMP region is a centre of survival that provides a refuge from elevated extinction rates, which should be more common in peripheral locations (Paulay, 1990; Jackson et al., 1993; Bellwood & Hughes, 2001).
First proposed by Woodland (1983), the centre of overlap hypothesis maintains that the high species diversity in the IMP hotspot is due in part to the overlap of distinct faunas from the Pacific and Indian Oceans (Briggs, 1974; Woodland, 1983). Under both the centre of accumulation and the centre of overlap hypothesis, speciation occurs in locations outside the hotspot, with subsequent dispersal into the IMP region. The difference between these two hypotheses lies in the mechanism driving divergence and the resulting distribution of sister taxa. The centre of overlap hypothesis is based on the premise that the isolating mechanism is the shallow Sunda and Sahul shelves of Indonesia, Malaysia and Northern Australia, known as the Indo-Pacific Barrier (IPB; Fig. 2). Under this hypothesis, the faunas of the Pacific and Indian Oceans gained distinction during historical low sea-level stands when dispersal was restricted between ocean basins. Following sea-level rise, the geographical ranges of sister taxa formerly separated by the IPB expanded and eventually came to overlap in the IMP hotspot. Under the centre of accumulation hypothesis, speciation is also thought to occur in peripheral locations but no specific mechanism is implicated; sister taxa can reside anywhere in the range, and the overlap of closely related taxa at the IMP region is not required.
The literature is replete with examples that support one hypothesis over another. Efforts to compile the evidence have largely sought to uncover the prevailing mechanism responsible for the species richness of the IMP hotspot but have not elevated any single hypothesis (Connolly et al., 2003; Mora et al., 2003; Halas & Winterbottom, 2009). Emerging from decades of debate is a growing recognition that the competing hypotheses are not mutually exclusive but in fact are likely to be working in conjunction to create the species richness of the region (Bellwood & Hughes, 2001; Rocha et al., 2008; Bowen et al., 2013; Briggs & Bowen, 2013; Cowman & Bellwood, 2013). In an effort to contribute to the growing understanding of the complexity of this issue we present a review of the evidence in support of the centre of overlap (C-O) hypothesis. Here, we describe the mechanism of isolation and provide evidence of the partial, and at times nearly complete, isolation of the Pacific and Indian Ocean faunas since at least the Miocene.
The Mechanism of Isolation
Woodland (1983), in his work on rabbitfishes (family Siganidae), noticed pairs of sister species within which one had a distribution centred in the Pacific Ocean whilst the other's was centred in the Indian Ocean, with the tails of these ranges overlapping in the IMP region. He surmised that at least part of the species richness in the hotspot was due to the overlapping of distinct faunas from these two ocean basins. Since then, the list of sister species that share this distribution pattern has grown (Donaldson, 1986; Blum, 1989; Randall, 1998). Originally, Woodland (1983) suggested that Pleistocene low sea-level stands may have led to the isolation of populations in the Pacific and Indian Oceans (Fig. 2). However, subsequent phylogenetic work suggests an older date for most extant species (Renema et al., 2008), indicating that tectonic rearrangements in the Miocene may have restricted water flow between oceans even without dramatic changes in sea level.
During the Oligocene and early Miocene, the Pacific and Indian Oceans were connected via an extension of the Southern Equatorial Current known as the Indonesian Seaway (Hall, 2002; see also Figure 2 in Carpenter & Springer, 2005). Through this seaway, water could flow between oceans with little impediment. During the mid-Miocene (c. 16–8 Ma), however, the Australian and Eurasian plates collided, resulting in shallow seas and land barriers in the region that deflected equatorial currents and greatly reduced flow between ocean basins (Kennett et al., 1985). Today, after millennia of continued tectonic movement, water flow between the Pacific and Indian Oceans is largely confined to the Indonesian Throughflow (Fig. 2), which, while capable of carrying large volumes of water (10 million m3 s−1), must follow a narrow and restricted path through the Indonesian Archipelago (Gordon, 2005).
Accentuating the isolation of the Pacific and Indian Oceans were eustatic changes in sea level in the recent past. The shallow continental shelves of the IPB have been subjected to lower sea levels repeatedly since at least the Pliocene (Naish et al., 2009). During the Pleistocene, there were three to six glacial cycles that lowered sea level to as much as 130 m below present levels (Fig. 2; Chappell, 1981; Potts, 1983; Voris, 2000; Naish et al., 2009). During low sea-level stands, species on the continental shelves were extirpated, and presumably there was a major reduction in the already restricted gene flow between Pacific and Indian Ocean populations. Associated with the change in sea level were concomitant changes in oceanographic current patterns and altered freshwater and sediment discharge of local rivers, with corresponding changes in temperature and salinity (Tjia, 1966; van Andel et al., 1967). The narrow seaways that remained were likely under the influence of cooler upwelling, further limiting the availability of suitable habitat for tropical marine organisms (Galloway & Kemp, 1981; Potts, 1983; McManus, 1985; Fleminger, 1986; Voris, 2000). During these exceptionally low sea-level stands, population sizes were reduced (Bellwood & Wainwright, 2002) and isolation of the Pacific and Indian Oceans was nearly complete.
The restriction of flow between the Pacific and Indian Oceans since the mid-Miocene has played a role in isolating the fauna in the Pacific and Indian Oceans. However, perhaps of equal significance is the paucity of coral reef habitat in the north and central Indian Ocean. Notably, there is little coral development along the coasts of Bangladesh, India and Pakistan, owing to diminished habitat and large freshwater outflows. The Cocos-Keeling and Christmas Islands in the eastern Indian Ocean are known regions of faunal overlap and hybridization between Pacific and Indian Ocean reef fish fauna (Hobbs et al., 2009). West of here, a 2800-km stretch of open ocean separates Indonesia and Western Australia from the central Indian Ocean (Fig. 1). This oceanic desert is the western boundary of the Indo-Polynesia Province. Another 1000 km of open ocean separates the Saya de Malha banks in the Western Indian Ocean from the Indo-Polynesian Province (Fig. 1; Briggs & Bowen, 2012), a region that has an enormous influence on the biogeography of the Indian Ocean (Sheppard et al., 2012). Of course, the Pacific and Indian Ocean basins are not fully isolated, as evidenced by the large number of species whose distributions span both oceans (Randall, 1998). However, the presence of distinct faunas in the Western Indian Ocean and the Indo-Polynesian Province indicates restricted dispersal perhaps across the IPB and certainly across the central Indian Ocean, which is possibly amplified by ecological differences between these regions.
The C-O hypothesis predicts that isolation of intraspecific populations persisted long enough for speciation to occur. According to fossil and phylogenetic data, most extant reef fauna have a Pliocene (1.6–5.3 Ma) or Miocene (5.3–23.7 Ma) origin (Renema et al., 2008). Corresponding with these ages is the restriction of water flow between the Pacific and Indian Oceans that has been in place since the mid-Miocene, when the Asian and Australian plates collided and land barriers arose. While this may be sufficient to explain the origin of many recognized species, this should not be interpreted as implying that speciation did not occur during the Pleistocene. Instead, this finding may be, in part, a result of our tendency to name species only with clear and definable morphological characters (Rocha & Bowen, 2008; see ‘Intraspecific Comparisons’, below). Moreover, several lines of evidence indicate that isolation neither has to be absolute nor of long duration for speciation to occur (Gavrilets et al., 2007; Rocha & Bowen, 2008). Selection driven by ecological factors could accelerate divergence despite gene flow or continue generating divergence that began in isolation. Alternatively, sexual selection or assortative mating might reinforce isolating mechanisms in parapatric populations. Regardless of the mechanism of isolation, there is ample evidence of divergence across the IPB dating back to at least the late Miocene.
The strongest evidence comes from distributional patterns of pairs of sister species (Table 1). Here we define sister species as taxa that share a common ancestor and are each other's closest living relative. The C-O hypothesis predicts that one member of each pair will have a distribution centred in the Pacific Ocean with the other centred in the Indian Ocean. In some cases, species remain largely isolated in either ocean basin or exhibit asymmetric expansion into the IMP region (Fig. 3a). However, population expansion following speciation can lead to a pattern where closely related species overlap in the region (Fig. 3b). This pattern is the most convincing evidence for the C-O hypothesis (see Fig. 4); however, it might be considered an intermediate stage of population expansion. Given enough time and adequate dispersal ability the end result might be two closely related species with widely overlapping distributions (Fig. 3c; Lynch, 1989). It is important to note that these distributions are not inconsistent with other hypotheses. For instance, the pattern in Fig. 3(b) (and Fig. 4) could arise from the contraction of previously large widespread populations, or, conversely, the pattern in Fig. 3(c) could be the result of an expansion of populations historically restricted to the IMP region. However, the finding of numerous examples of sister taxa whose distributions overlap only in areas adjacent to the IPB and the lack of such a pattern in other areas provide evidence for the C-O hypothesis.
Table 1. Sister species pairs of fishes based on morphology and/or molecular data with distributions that overlap in the Indo-Malay-Philippine (IMP) biodiversity hotspot. More examples are given by Randall (1998), but we have restricted our list to sister species pairs within which one has a distribution centred in the Pacific Ocean and the other has one centred in the Indian Ocean, with the tails of these ranges overlapping in the IMP hotspot. Divergence times, as reported in source articles, are listed in millions of years ago (Ma). Divergence times in bold were calculated from mitochondrial cytochrome b sequences downloaded from GenBank, assuming a molecular clock of 2% per million years between lineages (Bowen et al., 2001; Lessios, 2008; Reece et al., 2010)
Divergence time (Ma)
Pterois miles (Bennett, 1828) – P. volitans (Linnaeus, 1758)
Most pairs of sister species have been defined based on taxonomic work (reviewed in Randall, 1998). In fishes, sister species are often delineated based on subtle differences in colour patterns, body measurements, or fin-ray counts. As a consequence, species pairs that are morphologically indistinguishable (cryptic species) and those that are morphologically more divergent than expected are omitted from consideration. Molecular phylogenies would be a useful tool to confirm the sister status of these pairs; however, few data are available for most groups, with few exceptions (e.g. butterflyfishes). Phylogenies, even for well-studied groups, are often incomplete, thus elevating the age of lineages and obscuring relationships (Alfaro et al., 2007). Only small groups, such as wrasses of the genus Anampses (family Labridae), have complete phylogenies (Hodge et al., 2012). Another complication is the lack of detailed species checklists for much of Indonesia and Malaysia where overlap is expected. Only a few areas in the region have been sampled adequately. Undoubtedly, the importance of the IPB in species formation will become clearer as phylogenetic data accumulate and biodiversity surveys become available.
Evolution is an ongoing process, with changes in allele frequencies and genetic divergence being an indication of evolutionary change at the population level. So while intraspecific patterns of genetic variation are not often directly employed to address questions concerning the generation of species, they provide a historical perspective that may not be apparent with contemporary species distributions (Palumbi, 1997; Avise, 2000). Pleistocene sea-level fluctuations resulted in repeated widespread extirpations on the continental shelves of the IMP region. The resulting interruptions in gene flow between Pacific and Indian Ocean populations are recorded in intraspecific patterns of genetic diversity in many taxa (Fig. 5). Since the end of the Last Glacial Maximum about 18,000 years ago, the land bridge that limited dispersal between the Pacific and Indian Oceans submerged and the rising sea level not only widened dispersal pathways but also was accompanied by an approximately 10-fold increase in suitable shallow reef habitat (Bellwood & Wainwright, 2002).
Each hypothesis concerning the origin of species richness in the IMP hotspot results in specific predictions about the geographical positioning of lineages within species (Palumbi, 1997; Drew & Barber, 2009). The centre of origin hypothesis predicts that the oldest and most diverse populations will be centred in the hotspot, with decreasing haplotype diversity emanating from the centre that parallels the observed decline in species richness (sensu Bowen et al., 1998). Similarly, the centre of overlap hypothesis predicts that the most diverse populations will be centred on the IMP hotspot. In this case, however, the high diversity is the result of the overlap of divergent lineages from peripheral regions (i.e. bimodal mismatch distributions).
Distinct genetic lineages that date to Pleistocene sea-level fluctuations and whose distributions are consistent with the C-O hypothesis have been detected in many coral reef organisms (reviewed in Gaither et al., 2010). In some taxa, effective migration between ocean basins is absent, as evidenced by a lack of shared haplotypes between oceans (e.g. Penaeus monodon, Duda & Palumbi, 1999; Chlorurus sordidus, Bay et al., 2004). These cases possibly reflect limited population expansion since the Pleistocene and perhaps the early stages of allopatric speciation. Other widespread taxa demonstrate strong evidence of historical isolation but lack contemporary spatial structure (e.g. Naso brevirostris, Horne et al., 2008; the Centropyge flavissima complex, DiBattista et al., 2012a) indicating extensive dispersal and introgression following sea-level rise. In other species we detect evidence of isolation across the IPB with population expansion and overlap in the IMP hotspot. As predicted by the C-O hypothesis, distinct lineages, one of which is widely distributed in the Pacific Ocean and the other in the Indian Ocean, demonstrate contemporary mixing in the IMP region (Fig. 5). Analyses of these patterns are limited in many cases owing to gaps in sampling, but the overall signal of isolation of Pacific and Indian Ocean lineages with overlap in the hotspot is supported across a variety of taxa, including two teleosts, an elasmobranch, an echinoderm and a gastropod (Fig. 5). Undoubtedly, more examples will emerge as studies across the region accumulate.
The region bounded by the Philippines, the Malay Peninsula, New Guinea and northern Australia is an epicentre of marine biodiversity (Ekman, 1953; Stehli & Wells, 1971; Veron, 1993; Randall, 1998; Bellwood et al., 2012; Briggs & Bowen, 2013). The explanation for the high number of species in the region has long been debated. There now seems to be an emerging consensus that no single mechanism is driving speciation, and instead the debate has shifted to the relative importance of the various processes (Bowen et al., 2013; Cowman & Bellwood, 2013). Species distributions and intraspecific molecular studies highlight the role of the IPB in contributing to the distinction of the Pacific and Indian Ocean faunas. Times of divergence across sister species pairs indicate that diminished gene flow across this barrier has been in effect since at least the Miocene and was accentuated during Pleistocene sea-level fluctuations. Population expansion following divergence has resulted in the sympatric distribution of sister species pairs and/or genetic lineages in the IMP region, adding to the species/genetic richness there. An understanding of the relative contribution of the IPB to biodiversity in the region is limited by gaps in our knowledge of the evolutionary relationships among coral reef species, by a lack of phylogeographical data from the Indian Ocean, as well as by a lack of detailed species checklists for many regions. However, as the data accumulate it is becoming clear that Woodland (1983) was correct: the overlap of distinct faunas from the Pacific and Indian Oceans has contributed to the species richness of the IMP biodiversity hotspot.
We thank Brian Bowen for lively discussions and intellectual input; Jean-Paul Hobbs and Shelley Jones for comments on the manuscript; and Kanesa Duncan and Toby Daly-Engel for sharing their data set. Jonathan B. Puritz, John E. Randall and Jeffery Williams allowed us to use their photographs. Suggestions by Michael Dawson, Cynthia Riginos and one anonymous referee greatly improved the final version. This work was supported by the California Academy of Sciences.
Michelle R. Gaither is a postdoctoral fellow in the Section of Ichthyology at the California Academy of Sciences. Her interests include introduced species, the biogeography of marine fishes, and the genomics of speciation.
Luiz A. Rocha is the curator of Ichthyology at the California Academy of Sciences. His interests are centred around the mechanisms that cause the diversification of coral reef fishes. His current research includes a genome-wide analysis of fishes that searches for signatures of different modes of speciation, the phylogeography of Red Sea reef fishes and the evolution of fishes from mesophotic coral reefs.