Higher speciation and lower extinction rates influence mammal diversity gradients in Asia

Background Little is known about the patterns and correlates of mammal diversity gradients in Asia. In this study, we examine patterns of species distributions and phylogenetic diversity in Asia and investigate if the observed diversity patterns are associated with differences in diversification rates between the tropical and non-tropical regions. We used species distribution maps and phylogenetic trees to generate species and phylogenetic diversity measures for 1° × 1° cells across mainland Asia. We constructed lineage-through-time plots and estimated diversification shift-times to examine the temporal patterns of diversifications across orders. Finally, we tested if the observed gradients in Asia could be associated with geographical differences in diversification rates across the tropical and non-tropical biomes. We estimated speciation, extinction and dispersal rates across these two regions for mammals, both globally and for Asian mammals. Results Our results demonstrate strong latitudinal and longitudinal gradients of species and phylogenetic diversity with Southeast Asia and the Himalayas showing highest diversity. Importantly, our results demonstrate that differences in diversification (speciation, extinction and dispersal) rates between the tropical and the non-tropical biomes influence the observed diversity gradients globally and in Asia. For the first time, we demonstrate that Asian tropics act as both cradles and museums of mammalian diversity. Conclusions Temporal and spatial variation in diversification rates across different lineages of mammals is an important correlate of species diversity gradients observed in Asia. Electronic supplementary material The online version of this article (doi:10.1186/s12862-015-0289-1) contains supplementary material, which is available to authorized users.


How Asian are the Asian species?
There are 457 mammalian genera that have distribution in Asia. There are 1863 species of mammals in Asia [1]. If all species in each of these genera, irrespective of their geographic distribution is considered, there are 2549 species. Thus overall, Asian genera have 25% of their species outside the Asian boundaries as defined by this study.
The figure shows the distribution of different proportions of Asian species across the 457 genera. Most genera have species that are restricted in Asia, while few have more than 50% of their species distributed outside this study's definition of Asia. Thus, we consider this dataset to be reflective of processes occurring in Asia and to be an accurate representation of the same.

How do missing species affect the patterns?
We have three analysis that rely on the phylogeny heavily -Phylogenetic diversity, LTT and Stadler's method for temporal patterns and GeoSSE for spatial patterns. Both Stadler's algorithm and the GeoSSE method can account for missing species while estimating diversification rates.
In total, 405 mammal species are not represented in the supertree. Of these, 275 have geographical ranges described by the IUCN (which is the source of range-data in our study).
The remaining 129 have no range data described in the IUCN. Of the 275 species whose ranges are described in the IUCN, 75 species have ranges in Asia.
The 75 species include 1 Monotreme, 6 marsupials, 2 primates, 3 lagomorphs, 2 Erinaceomorphs, 14 soricomorphs, 21 bats, 6 Artiodactyls and 20 rodents. Most of these species are from South and Southeast Asia and Western mountains of the Himalayan system (Supplementary Figure 2). Of the 129 species with no range data in IUCN, 9 are from Asia (whose locations fall within our study area). Most of these species have gone extinct.

How does unrecognized or unknown diversity impact the patterns?
Previous studies have predicted that the highest number of undiscovered mammal species will be from the tropical forests [2]. Given this, and the fact that there have been new species of rodents and shrews described from tropical forests in Asia, it is conceivable that a high proportion of unsampled diversity is hidden tropical regions. Some of the recently discovered mammal species from these regions include -Leonastus aenigmamus (striped rabbit) [3], Saxatilomys pailinae [4], Tonkinomys daovantiens [5], Hylomys megalotis, Chodsigoa caovansunga [6], Crocidura kegoensis, among others [7]. Accounting for this 'hidden' diversity will only accentuate the patterns we observe.
Most this hidden diversity is non-randomly distributed across clades and space. Small mammal orders -such as Chiroptera, Rodentia and Sorcimorpha are witnessing high species discovery.
Apart from genetic species, new morpho-species are also being described. Many regions of tropical Asia still remain poorly surveyed and the diversity poorly documented. Species and phylogenetic diversity of this region will increase, and the estimates of diversification rates may subsequently increase for the tropical biome. Given that for Rodentia, Chiroptera and Soricomorpha we recovered higher speciation rates in the tropical biomes, discovery of new species will only increase the estimates of speciation rates.

What is the relationship between species richness (SR) and Phylogenetic diversity
(PD)?

Pull of the present and LTT plots:
Inferring speciation and extinction rates solely from LTT plots may be incorrect as they suffer from the 'pull of the present' effects. The pull of the present refers to the apparent increase in diversification rate towards the present, arising from the fact that lineages that have arisen recently are less likely to go extinct and thus are represented in the phylogeny. In our study, we do not estimate diversification rate changes from the LTT, and use the LTT plots as indicators of the overall temporal patterns of lineage diversifications. The diversification shift-times were estimated using the method developed by Stadler et al 2011, which is unaffected by the 'pull of the present'.  The inclusion of all species in a genus even if they were not distributed in Asia was inevitable to take into consideration the entire evolutionary history of the genus. If we ignored some species in every genus, it would be an underestimate of the number of evolutionary events that have occurred in the genus, thus resulting in lower speciation estimates. But this results in a decrease in the sampling frequency fed into the algorithm, leading the algorithm to assume that the missing species are also found in the corresponding biomes in Asia. However, given that most sister-species pairs show similar niches, it is conceivable that the species outside of Asia also belong to similar biomes as their sister species in Asia. Thus, the diversifications rates estimated by our method should be an accurate representation of the true trends, while the estimates themselves may have some inaccuracies.