MORPHOLOGICAL STASIS IN AN ON-GOING RADIATION OF BELLAMYA GASTROPODS FROM ANCIENT LAKE MALAWI

Evolution via natural selection can proceed rapidly, even under weak selection. No rates of evolution have been calculated for morphological changes observed in freshwater mollusks of the East African Rift. This is unfortunate, because several mollusk groups occupying ancient rift lakes have diversified with an accompanying increase in morphological disparity. The monophyletic group of Bellamya gastropods from Lake Malawi, in which currently four species are considered taxonomically valid, is a good example. Molecular analyses arguably indicate that these extant species are very young, and the strong demographic and spatial expansion these analyses revealed suggests that the group is diversifying since Late Pleistocene (135-70 ka) lake level drops of Lake Malawi. We studied morphological evolution via semi-landmark analyses of 912 Bellamya specimens from a high-resolution sequence of 21 early to middle Holocene fossil assemblages and 283 modern specimens of the four extant species. These data were combined with shell height measurements and information on the number of whorls, as these features are considered diagnostically important in Bellamya. Principal component analysis indicated that the fossil shells are intermediate in the abovementioned features to two extant nominal species, but the fossils also occupy an area of morphospace that is not represented in the modern fauna. Clustering using normal mixture models suggests that the fossil specimens belong to a single evolutionary lineage. The evolution in this lineage is best fit by a model of evolutionary stasis, hence slower than anticipated. The obtained rate parameters for changes in the number of whorls and in shell shape, however, are still remarkably high compared to those obtained in rate simulations along randomly resolved phylogenies. Evolutionary divergence towards the extant descendants could have occurred since the middle Holocene if morphological evolution for these features would have proceeded at rates comparable to the paleontological ones calculated for the fossil time series. For shell height, the simulated phylogenetic rates coincided best with the paleontological rates if divergence started some 50±20 ka ago.