The pattern and timing of diversification of Philippine endemic rodents: Evidence from mitochondrial and nuclear gene sequences

The 22 genera and 64 species of rodents (Muridae: Murinae) distributed in the Philippine Islands provide a unique opportunity to study patterns and processes of diversification in island systems. Over 90% of these rodent species are endemic to the archipelago, but the relative importance of dispersal from the mainland, dispersal within the archipelago, and in situ differentiation as explanations of this diversity remains unclear, as no phylogenetic hypothesis for these species and relevant mainland forms is currently available. Here we report the results of phylogenetic analyses of the endemic Philippine murines and a wide sampling of murine diversity from outside the archipelago, based on the mitochondrial cytochrome b gene and the nuclear-encoded IRBP exon 1. Analysis of our combined gene data set consistently identified five clades comprising endemic Philippine genera, suggesting multiple invasions of the archipelago. Molecular dating analyses using parametric and semiparametric methods suggest that colonization occurred in at least two stages, one ca. 15 Mya, and another 8 to 12 million years later, consistent with the previous recognition of Old and New endemic rodent faunas. Ancestral area analysis suggests that the Old Endemics invaded landmasses that are now part of the island of Luzon, whereas the three New Endemic clades may have colonized through either Mindanao, Luzon, or both. Further, our results suggest that most of the diversification of Philippine murines took place within the archipelago. Despite heterogeneity between nuclear and mitochondrial genes in most model parameters, combined analysis of the two data sets using both parsimony and likelihood increased phylogenetic resolution; however, the effect of data combination on support for resolved nodes was method dependent. In contrast, our results suggest that combination of mitochondrial and nuclear data to estimate relatively ancient divergence times can severely compromise those estimates, even when specific methods that account for rate heterogeneity among genes are employed.