A molecular phylogeny of Asian barbets: Speciation and extinction in the tropics

Highlights

• Not all named species are monophyletic.

• Five clades raised to species level.

• Most species originate before the Pleistocene.

• Phylogenetic evidence for extinction in island lineages.

• Support for “museum hypothesis” origin of tropical diversity.

Abstract

We reconstruct the phylogeny of all recognized species of the tropical forest associated Asian barbets based on mitochondrial and nuclear sequence data and test for the monophyly of species and genera. Tropical regions are well known for their extraordinarily high levels of biodiversity, but we still have a poor understanding of how this richness was generated and maintained through evolutionary time. Multiple theoretical frameworks have been developed to explain this diversity, including the Pleistocene pump hypothesis and the museum hypothesis. We use our phylogeny of the Asian barbets to test these hypotheses. Our data do not find an increase in speciation in the Pleistocene as predicted by the Pleistocene pump hypothesis. We do find evidence of extinctions, which apparently contradicts the museum hypothesis. However, the extinctions are only in a part of the phylogeny that is distributed mainly across Sundaland (the Malay peninsula and the islands off southeast Asia). The theory of island biogeography predicts a higher rate of extinction on islands than on mainland areas. The data from the part of the phylogeny primarily distributed on the mainland best fit a pure birth model of speciation, and thus supports the museum hypothesis.

Introduction

Tropical regions contain a disproportionate amount of biodiversity (Myers et al., 2000), and are also under particular threat due to extraction of resources for export and increasing local human populations (Cincotta et al., 2000, Davies et al., 2006, Woodruff, 2010). The Asian barbets (Aves: Megalaimidae) are a bird family closely related to woodpeckers and the African and South American barbets (Moyle, 2004, Hackett et al., 2008), and may be representative of many tropical forest taxa for several reasons. They are non-migratory, forest dependent, non-passerine birds with relatively poor dispersal (Short and Horne, 2001, Horne and Short, 2002). Only one species, the widespread coppersmith barbet (Megalaima haemacephala), expanded across Huxley’s line and reached the Philippines (Dickinson et al., 1991) (Fig. 1). Barbets and the closely related toucans inhabit tropical Africa, Asia and America, but the species in each of these three regions have been shown to form monophyletic clades (Moyle, 2004).

The clade of current Asian barbets originated around 31 million years ago (MYA) (Moore and Miglia, 2009) and contains around 30 species (Short and Horne, 2001, Collar, 2006, Feinstein et al., 2008) that are mostly territorial in their breeding behavior (Short and Horne, 2001). Some of the species are widespread and contain morphologically distinct subspecies. In the last few years, some subspecies have been raised to species based in part on molecular data (Rasmussen and Anderton, 2005, Collar, 2006, Feinstein et al., 2008), and it is possible that more data will identify more divergent units that should be raised to species level.

Three general processes govern the level of biodiversity or species richness in any area: speciation, extinction and dispersal (Wiens and Donoghue, 2004). How the extraordinary level of tropical diversity originated is still debated. Two basic hypotheses are the ‘museum hypothesis’ and the ‘Pleistocene pump’ hypothesis. The museum hypothesis explains the high level of biodiversity through low levels of extinction (Mittelbach et al., 2007). The Pleistocene pump hypothesis explains the diversity by high levels of speciation during the Pleistocene as populations diverged and speciated in allopatry and then expanded into sympatry repeatedly as the climate fluctuated.

The Pleistocene pump hypothesis was not formulated specifically for tropical Southeast (SE) Asia (Haffer, 1969), but it seems particularly appropriate for this region (Sodhi et al., 2004, Woodruff, 2010). Unlike the other major tropical regions, SE Asia contains both large areas of mainland and many islands of varying size including Borneo and Sumatra. These islands are on a shallow continental shelf called the Sunda shelf, and were connected with each other and the mainland of SE Asia by large expanses of dry land periodically through the end of the Pliocene and the Pleistocene (Heaney, 1986, Heaney, 1991, Hall, 1998, Sathiamurthy and Voris, 2006). This cyclical pattern may generate diversity through a pump in which populations of widespread taxa diverge in isolation when sea level is high, and then expand out of their refugia as reproductively isolated species that colonize the other regions when the landmasses are rejoined in times of lower sea levels (Gorog et al., 2004, Lim et al., 2011; Woodruff, 2010). This Pleistocene pump hypothesis predicts an increase in diversification rate starting when the sea level fluctuations start at the end of the Pliocene or as they intensify in the Pleistocene. This can be tested by constructing a phylogeny of all species in the group, and dating the nodes representing speciation.

The museum hypothesis predicts very low levels of extinction, so to reject the museum hypothesis it would be necessary to demonstrate high levels of extinction in tropical taxa. Although the fossil record makes it perfectly clear that the history of life on earth is primarily the history of extinctions (Raup, 1986), actually testing for extinctions using phylogenies is quite difficult (Purvis, 2008, Rabosky, 2010). A serious issue hindering this is that the pattern of temporal diversification yielded by extinctions can also be generated by an increase in diversification rate affecting the whole tree or a rate shift among clades in the phylogenetic tree (Chan and Moore, 2005, Rabosky, 2010). So, in order to demonstrate extinctions it must first be demonstrated that there has not been a change in diversification in the phylogeny. Empirical data from birds most commonly yield a density dependent pattern of speciation (Phillimore and Price, 2008). However, recently developed temporal and topological tree tests now enable changes in rates of cladogenesis along phylogenies to be tested (Rabosky, 2006a, Chan and Moore, 2005). Therefore, if a lineage through time analysis yields a pattern consistent with extinctions, a change in diversification rate can be tested for, and potentially rejected. Topological tree tests can be used to test if there has been a significant rate shift among clades and even identify which branch(es) are affected (Moore et al., 2004, Chan and Moore, 2005) and therefore distinguish between a birth death pattern and among clade rate diversification.

Here we construct the complete phylogeny of the Asian barbet species with extensive subspecies sampling, so we are able to test some taxonomic hypotheses within the group. Further, we will use that phylogeny to test two hypotheses that may explain tropical biodiversity. First, the ‘Pleistocene Pump’ hypothesis that predicts increased diversification at the onset of increased climatic fluctuations (glacial cycles) and associated sea level fluctuations (Lisiecki and Raymo, 2005) around 2.5–3 MYA. Secondly, we test the ‘museum’ hypothesis, that explains high tropical biodiversity by low rates of extinction in relatively stable environments, by comparing (i) if a rate constant model fits the observed data better than a rate variable model and (ii) which model of rate constant diversification (one with extinction and one without extinction) fits the observed pattern of temporal diversification better.