Explaining why some lineages diversify while others do not and how are key objectives in evolutionary biology. Young radiations of closely related species derived from the same source population provide an excellent opportunity to disentangle the relative contributions of possible drivers of diversification. In these settings, lineage-specific effects are shared and can be ruled out. Moreover, the relevant demographic and ecological parameters can be estimated accurately. Midas cichlid fish in Nicaragua have repeatedly colonized several crater lakes, diverged from the same source populations, and, interestingly, diversified in some of them but not others. Here, using the most comprehensive molecular and geometric morphometric data set on Midas cichlids to date (∼20,000 SNPs, 12 landmarks, ∼700 individuals), we aim to understand why and how crater lake populations diverge and why some of them are more prone to diversify in sympatry than others. Taking ancestor-descendant relationships into account, we find that Midas cichlids diverged in parallel from their source population mostly-but not exclusively-by evolving more slender body shapes in all six investigated crater lakes. Admixture among crater lakes has possibly facilitated this process in one case, but overall, admixture and secondary waves of colonization cannot predict morphological divergence and intralacustrine diversification. Instead, morphological divergence is larger the more dissimilar a crater lake is compared to the source lake and happens rapidly after colonization followed by a slow-down with time. Our data also provide some evidence that founder effects may positively contribute to divergence. The depth of a crater lake is positively associated with variation in body shapes (and number of species), presumably by providing more ecological opportunities. In conclusion, we find that parallel morphological divergence in allopatry and the propensity for diversification in sympatry across the entire Midas cichlid fish radiation is partly predictable and mostly driven by ecology.
Keywords: Admixture; RADseq; colonization history; demographic inference; ecological opportunity; evolutionary rate; founder effect; geometric morphometrics; parallel evolution; phenotypic trajectory.