“You: on the earth’s surface. so young. so dynamic. full of life and suggesting a world of possibility. Me: subterranean. really old. fossilized, almost. intriguing but slightly inscrutable. We brushed past each other in Rabosky and Matute (2013). I thought there was something there, but in the blink of a p-value you were gone.”
One of the perennial questions in evolutionary biology is “What factors determine how many species are on earth?” Researchers take numerous approaches to get at this very big question. One is to look for correlations between attributes of organisms, the environments they inhabit, or geologic history and rates of species diversification. This the study of macroevolution, and it is based on the idea that the discovery of these correlations on large scales (often using datasets with hundreds to thousands of species with deep histories spanning tens of millions of years) would be a powerful indicator of the factors governing species richness. Another approach is to study speciation on a small scale, to examine sets of closely related populations currently in the process of diverging. The thought is that if we can observe the forces driving divergence in contemporary populations, we can use those observations to develop a more general understanding.
One of the tenets of modern evolutionary biology that underlies the latter approach is that microevolutionary processes operating at the population level (mutation, selection, drift, migration) that are observable on short (-ish) time scales and among only a few populations are responsible, over tens to hundreds of millions of years for generating global-scale patterns of diversity. Though this belief is widely held, it is quite difficult to make direct connections between microevolutionary processes and macroevolutionary patterns. This is not necessarily because they aren’t there, but because the data required to address such questions are difficult to obtain, and once in hand require extremely sophisticated statistics to be properly analyzed. In recent years, however, large datasets have begun to be assembled and there have been major advances in statistical methods available to analyze them.
A recent paper by Rabosky and Matute (2013) draws on two large datasets and develops a novel statistical method to test one possible connection between macro pattern and micro process. They ask if the rate of evolution of reproductive isolation (RI) among species is correlated with the rate of species diversification (DR). RI refers to the inability of two members of different populations or species to produce fertile offspring. RI is often thought to be important in diversification because some theory predicts that even low levels of intermating between populations can prevent divergence from occurring and because hybridization between divergent populations can cause them to homogenize, or cause one population to become extinct. If these factors commonly prevent speciation or cause incipient species to go extinct, one might expect a positive correlation between the rate of evolution of RI and DR. This paper is the first test of this prediction.
Rabosky and Matute bring two datasets to bear on the question, a Drosophila dataset including 94 species and a bird datset consisting of 6,670 species. These datasets include measures of RI as well as phylogenetic information that can be used to estimate DRs. They look at two types of RI, premating and postzygotic, and measure them using a statistic of their own devising. They fit a battery of models to the RI data to ascertain how RI changes over time and whether its pace of evolution varies among subgroups within both the birds and Drosophila. They then estimate DRs for those subgroups and ask if RI is correlated with DR. They take a couple approaches to this question using variants of their models and surprisingly, they find no correlation between the rate of evolution of RI and DR. Their strongest evidence is for no relationship between postzygotic RI and DR, as they only had data on premating RI for Drosophila, and power analyses indicate the tests using those data had weak discriminatory ability.
This finding suggests that the rate of evolution of postzygotic RI does not play a very large role in the formation or persistence of evolutionary lineages. Instead, other factors, such as the ability to colonize distant habitats or adapt to new environments may be more important in governing species diversification.
I think a flaw of this type of approach, (which is acknowledged by the authors) is that the issue of assigning species status is mostly sidestepped. The authors note that using the most commonly held species concept, the Biological Species Concept, in which species status is only attained when RI between sister populations is complete, would introduce a circularity to the analyses that would all but guarantee a positive correlation. Instead, the authors simply assume that species are most commonly delimited on the basis of genetic and phenotypic data that may only be weakly related to RI and accept the taxonomic judgements of a diverse group of experts. There is no realistic way around this, as there is currently no way to redefine the taxonomy of thousands of organisms on the fly using a consistent approach in the context of a single paper (although, with massive phenotypic and genetic databases being assembled, this may eventually be possible). I think it is reasonable to suspect that the approach taken to delimit species (both operationally and conceptually) may have a profound affect on the results of these types of studies, but there hasn’t been much work done to test it.
So, in summary: cool datasets, fancy new statistical methods (that could merit their own blog post), surprising result, I think species delimitation is important. Good paper.
Rabosky, Daniel L., and Daniel R. Matute. “Macroevolutionary speciation rates are decoupled from the evolution of intrinsic reproductive isolation in Drosophila and birds.” Proceedings of the National Academy of Sciences 110.38 (2013): 15354-15359.