JEFFERY, J.E.; BININDA-EMONDS, O.R.P.; COATES, M.I.; RICHARDSON, ,M.K.; International School of Amsterdam, The Netherlands; Tech. Univ. of Munich; Univ. of Chicago, USA; Univ. of Leiden, The Netherlands: Optimising Heterochrony onto Phylogenetic Trees

Event-pairing is a powerful method for encoding developmental timing data, in order to optimise them on phylogenetic trees. For each species the method makes a series of pair-wise comparisons of the developmental events, giving a numerical scores based on their relative timing. Differences in �event-pair� (EP) scores between species can reveal heterochrony (evolutionary changes in developmental timing). The process of pair-wise comparisons introduces a large degree of interdependence to the final data for each species. Each event in the primary data set will share exactly one EP character with every other event in the final data set. However, tree optimisation procedures (e.g. ACCTRAN and DELTRAN) usually assume that each character is independent. Thus, at internal nodes, scores would be reconstructed for each EP character without regard for the interdependence of all EPs. In exceptional cases, equivocal changes (where a particular EP score change could have occurred at any one of several different nodes) can lead to illogical reconstructions. These are reconstructions that imply impossible timing relationships between the original events (e.g. the triplet �A occurs before B�, �B occurs before C�, and �A occurs after C�). This problem may be exacerbated if there is a large amount of missing data. We have been studying this phenomenon with a view to constraining the optimisation to only logical reconstructions. This requires methods of parsing a reconstruction rapidly, to assess it for any illogicalities. If several score changes are equivocal, a method is required to investigate different optimisation schemes. This is because making the reconstruction at one node logical could create illogicalities at another.