Wasiu (AJ) Akanni

I started my PhD in Dr. Davide Pisani’s evolutionary biology lab (National University of Ireland, Maynooth) in October 2010 after completing my Bsc. in Bioinformatics at the same university.

The aim of my research project is  testing co-evolution between plants and inserts. The basis of this research comes from a seminal paper published in 1964, in which Ehrlich and Raven proposed, based on fossil evidence and their own study of the diet of closely related butterflies larvae, that insect diversity resulted from a coevolutionary process: an arm race between the angiosperms and herbivorous insects.

In its simplicity, this hypothesis is very appealing, and it is perhaps unsurprising that, despite it has never been adequately tested, it is generally considered to be well established.

In my project I plan to test Eilrich and Raven’s (1964) coevolutionary hypothesis using multiple insect groups, large phylogenies, modern molecular clock methods, and the comparative method.

In so doing I shall gain significant insights in the reality of the most prominent model ever proposed to explain biodiversity accumulation (i.e. arm race driven coevolution).

Secondary aims of this project are the development of new strategies to select calibration points for molecular clock analyses, and the development of new supertree methods.

Supertrees are particularly important for this project because to address the proposed problem I first need to generate large trees for plants and insects.

To do so, during the first year of my PhD, I have been working on developing and implementing new Maximum Likelihood (ML) and Bayesian supertree methods (in conjusction with Peter G. Foster, The Natural History Museum, London).

Supertrees are generalisation of standard consensus methods allowing the amalgamation of partially overlapping trees – thus allowing the generation of very large phylogenies from heterogeneous data. However, most available supertree methods, e.g. the widely used Matrix Representation with Parsimony method (MRP) are ad hoc.

Recently, Steel and  Rodrigo (2008)  proposed that ML supertrees could be developed using an exponential model to model errors in input trees. Availability of ML-based supertrees will allow afundamental shift in the theory of supertree reconstruction, as will allow the use of non ad hoc methods.

Together with Dr Foster I wrote Python software to build Bayesian and ML supertrees. However, these programs have not been publicly released yet, as we are still testing them using some classic data set e.g. the eukaryotic dataset of pisani et al. (2007) and the bilaterian dataset of Holton and Pisani (2010).

These supertree methods will be used to build the large phylogenies we will need to test co-evolution of insects and plants.