Many biogeographical explanations are hampered by invoking simple notions of mechanism or process – dispersal and vicariance – or constraints, such as dispersal from a centre of origin, and, in so doing, dismiss more complex geological phenomena such as emergent volcanoes within island chains or composite areas as irrelevant. Moreover, they do not search for, therefore never discover, biogeographical patterns that may better explain the distribution of biota through time. Parenti & Ebach (2013, p. 813)
Documenting biogeographical patterns and testing hypotheses for the processes that generate patterns are challenging, as they require reconstruction of an area's geological and biotic history. Parenti & Ebach (2013) criticize two recent papers on the origins and biogeography of the biota from the Indo-Australian Archipelago (IAA) and from Sulawesi (respectively: Lohman et al., 2011; Stelbrink et al., 2012). Parenti & Ebach (2013) assert that the answers to why and how species distributions arise are solely hypotheses. Obviously, the answers to these questions result from actual events that took place, but our interpretation of these events is hypothesized because they were not directly observed. This is as trivial as it is true. They further state that ‘treating a hypothesis as if it were evidence or an empirical observation … favour[s] one explanation over another simply because no other explanation was ever considered’ (p. 813). This is a problem for all biogeographical analyses (Beaumont et al., 2010), but we believe that palaeogeographical reconstructions by geological experts (e.g. Hall, 2009) for the IAA and Sulawesi are the most suitable framework within which to test biogeographical hypotheses, as opposed to the use of hypothetical (see below) relationships of taxa (cladograms), by biologists, to reconstruct palaeogeography. Spurious ‘patterns’ can also result from failing to sample intermediate lineages between focal taxa (Turner et al., 2009), as we show here to be the case for one of the two data sets used by Parenti & Ebach (2013) to justify the recognition of Pandora (see Appendix S1 in Supporting Information). Thus, cladograms are also hypotheses (of species relationships), which are constantly revised as additional data are collected.
In our opinion, cladistic biogeographical analyses using ‘taxon area cladograms’, or more generally, ‘general area cladograms’, are weak tests of biogeographical hypotheses. Cladograms neglect temporal information (divergence times) preserved in biological data (molecular or other), and thus ignore linkages between distribution patterns, their potential causes, and their timing (e.g. geological events: separation of a tectonic fragment from its adjacent landmass, opening of a water strait, emergence of oceanic islands, etc.). Inconsistencies in timeframes result in so-called ‘pseudo-congruence’ (see Donoghue & Moore, 2003), and ignore potentially deviating biogeographical histories across lineages/clades due to potentially different causes (e.g. geological, biological, climatic). For example, geological changes on the island of Sulawesi of particular interest to biogeographers include the opening of the Makassar Strait and separation of the Sula Spur, which can be attributed to major geological changes during the middle Eocene and middle Miocene, respectively (Figure 1 in Stelbrink et al., 2012). Different organismal lineages may have different temporal histories associated with these geological events, and – while a molecular clock framework is a (working) hypothesis, as Parenti & Ebach (2013) correctly point out – the relative importance of each of these events can be segregated only if the timing of lineage evolution through space is understood, thus providing more predictive power than lumping all taxa – with potentially very different diversification histories – into a single analysis of area relationships. We therefore strongly emphasize that temporal information (estimates) of both geological events and divergence times are essential to test biogeographical patterns and their causes.
Parenti & Ebach (2013) believe it possible for organisms to colonize a volcanic island in situ and cite the example of Hawaiian gobies: ‘We hypothesize that Hawaii's endemic freshwater fish lineages ranged throughout the region where the Hawaiian Islands formed and, that the lineages of part of Hawaii's terrestrial biota ranged throughout that region, not necessarily as marine organisms, but on once emergent lands’ (p. 814). However, while this hypothesis is theoretically falsifiable, practically, it is not (see falsifiability and pseudo-science discussion in e.g. Popper, 1994), as we are unable to sample these unidentified hypothetical islands (geologically) or their ancestral biota (biologically) to test these ideas. Further, to the best of our knowledge, there have been extended periods (millions of years) of time when the Hawaiian Islands were completely submerged below sea level (Clague, 1996), although some panbiogeographers dismiss this evidence entirely (e.g. Heads, 2011). A counter-argument may be made that amphidromous gobies were able to disperse great distances across the sea (de Queiroz, 2005), a possibility not articulated by Parenti & Ebach (2013). This idea has been explicitly advanced by other ichthyologists (McDowall, 2007; Lindstrom et al., 2012).
The assumption of a vicariant distribution due to formation of a volcanic chain cannot be applied to the biota of Sulawesi based on our current state of geological knowledge. Two major vicariance events were revealed by the geological history of the composite island of Sulawesi: first, the opening of the Makassar Strait separating West Sulawesi from Borneo (Sunda Shelf), and secondly, the extension and westward movement of the Sula Spur from its former adjacent New Guinean/Australian landmass (Sahul Shelf; see discussion in Stelbrink et al., 2012). These represent ideal geological constraints with which to test whether vicariant scenarios can be excluded for terrestrial and freshwater taxa, through analysis comparing geological timing with estimated divergence times (see e.g. Figure 1 in Crisp et al., 2011). Parenti & Ebach (2013) note that ‘rejecting particular vicariance events to explain a distribution does not demonstrate dispersal as it ignores other vicariance events that may be invoked as an explanation.’ There might have been other vicariance events, but at the present time we cannot provide alternative explanations for Sulawesi's fauna, and neither do Parenti & Ebach (2013).
Parenti & Ebach (2013) consider Sulawesi to form part of a biogeographical subregion they call ‘Pandora’ (Parenti & Ebach, 2010), spanning an area including parts of Australia, Melanesia, Madagascar, Africa, Samoa and the Hawaiian Islands. Their evidence to erect this new subregion is based on uninterrupted distributions ‘across a large portion of the ancient southern continent, Gondwana’ (Parenti & Ebach, 2010, p. 311). The map they provide clearly shows these areas were highly disjunct even 30 Ma; however, these continents are known to have split apart far earlier (115 Ma according to Chakrabarty et al., 2012; one of the studies cited in Parenti & Ebach, 2013). Parenti & Ebach (2013) select two studies as tests for their newly proposed subregion: Austin (2000) and Chakrabarty et al. (2012). These are both molecular phylogenetic studies that use molecular clocks to seek a simple older than/younger than result for their conclusions, a practice criticized by Parenti & Ebach (2013). We re-examined these studies in detail. Austin (2000) found an ancient sister relationship between Malagasy and Melanesian boine snakes. However, Austin's (2000) study was made redundant by the more comprehensive study of Noonan & Chippindale (2006), which places the Malagasy boines with African taxa, rejecting the clade's former relationship with Melanesian taxa and overturning the sole basis for including these snakes as evidence for Pandora (Appendix S1). We re-analysed Chakrabarty et al.'s (2012) data set on troglobite fishes but excluded one of their four calibrations. This fossil was placed in the wrong part of the tree, was incompatible with the other calibration points, and significantly influenced the overall result (see Appendix S2). When these analyses were re-run, otherwise identically, the age estimates are significantly younger (see Figs S1 & S2 in Appendix S2), suggesting neither a Pandoran nor a Gondwanan explanation. Neither molecular study (Austin, 2000; Chakrabarty et al., 2012) provides support for this putative biogeographical subregion, casting doubt on Pandora's validity. We hope that the other studies used as raw data to derive Parenti & Ebach's (2010) Pandora areagrams are more robust.
In conclusion, we echo a recent call (Waters et al., 2013) to move beyond the ‘ancient vicariance’ argument to explain each and every biogeographical scenario. While area cladograms may provide useful (although potentially flawed – see Appendix S1) hypotheses for biogeography, as theory and methods have evolved, approaches that integrate geology and biology within a temporal framework (e.g. molecular dating) provide far more power for inferring biogeographical patterns and processes.