The analyses of sequence data of 658 basepairs of the cytochrome oxidase I gene from 71 specimens supports the idea that the freshwater crab species occurring in the Malili lakes do not originate from one single common ancestor (Fig. 2).
One lineage (cluster I) contains the generalist and detritivore ecotypes of the lakes, Parathelphusa pantherina, Parathelphusa ferruginea and Nautilothelphusa zimmeri, and forms a well-supported clade in the analyses (BI 0.98; MP 76; ME 69). The molluscivore species Syntripsa matannensis and Syntripsa flavichela comprise the other lineage (cluster II; BI 1.0; MP 100; ME 100), which does not represent the sister clade to cluster I. Instead, the phylogenetic relationship of cluster II is either unresolved, as in the case of the MP and ME analyses (Fig. 2), or it groups together with the well-supported group (BI 0.92; MP 97; ME 95) of Parathelphusa possoensis and Parathelphusa sarasinorum from Lake Poso and tributaries in the BI (confidence of 0.85 not shown). Cluster I represents a sister taxon to Parathelphusa pallida, a riverine freshwater crab occurring in the Malili region (BI 0.97; MP 59; ME 65). The phylogenetic position of Parathelphusa celebensis, another riverine species from the south of Sulawesi, is very basal and clearly separated from the Malili lakes species.
Within the Malili lakes lineages, the phylogenetic pattern reported previously could be confirmed (Schubart and Ng, in press). The homogeneity of P. pantherina, the generalist from Lake Matano, is very well supported by our results (BI 1.0; MP 100; ME 100) and this species forms the sister taxon to the cluster of all P. ferruginea and N. zimmeri (BI 0.86; MP 73; ME 87). Within the latter cluster, the generalist P. ferruginea (from Lake Mahalona and Lake Towuti) groups together with the sympatric population of N. zimmeri (BI 0.99; MP 99; ME 98), whereas the Matano population of N. zimmeri is well separated from the first group (BI 0.99; MP 99; ME 100) (Fig. 2). In the minimum spanning network, N. zimmeri from Lake Matano is separated from the conspecific population in lakes Mahalona and Towuti and from P. ferruginea by at least 13 mutational steps (Fig. 3).
Within the cluster of the generalist (P. ferruginea morph) and the detritivore (N. zimmeri morph) forms from lakes Mahalona and Towuti, a weak separation of the two species can be observed, but not very distinct and consistent. Three haplotypes of N. zimmeri (18, 19, 20) are positioned within the P. ferruginea cluster, one of them is even shared by both species (Fig. 3). The pairwise FST value of 0.057 between the two species is low, yet highly significant (p<0.001), indicating a restriction of gene flow, but maybe very recent or incomplete. In comparison to that, the FST value between the Matano population and the Mahalona/Towuti population of N. zimmeri is much higher (FST = 0.462; p<0.001). This finding is surprising, since morpholgically and ecologically the two populations of N. zimmeri are very similar and clearly derived from the bodyplan of riverine species and those of lacustrine generalists as P. ferruginea.