Date of this Version
Accurate knowledge of species boundaries and species phylogeny are fundamental to testing hypotheses of recent evolutionary processes, but the estimation of these partitions is challenging due both to inherent confusion about what is being estimated as well as the data available to estimate them. Using multilocus data from mtDNA, microsatellites, and nuclear locus sequences of over 1100 individuals, we delimited eight separately evolving species of Cichla rather than the 15 described. Among species we found evidence of rare but widespread introgressive hybridization, while within these species we observed evidence of long-term gene exchange and constrained evolutionary trajectories. In most cases of hybridization, mtDNA was exchanged, while nuclear introgression was extensive to negligible. We estimated a phylogeny among Cichla species using sequences of 21 putatively unlinked and single copy nuclear loci. We observed minor discord among loci with respect to the concatenated tree, but when a phylogeny was estimated from separate locus genealogies while accounting for individual uncertainty, a congruent tree was reconstructed. We inferred that less variable loci contributed important phylogenetic information that altered the final phylogenetic estimate. We used extensive mtDNA datasets from three species of Cichla that are found in the Orinoco and Amazonas River basins to test the hypothesis that the Casiquiare River permits exchange of genes between populations in each basin. We inferred that either gene exchange has been occurring between these regions or has ceased very recently. Based on the inferred species relationships and using time- calibrated and coalescent-based analyses of unlinked loci, we inferred that the polyphyletic arrangement of mtDNA in C. orinocensis was not consistent with deep coalescence and most likely derived from a history of ancient introgression.