Patterns and repeatability of multi-ecotype assemblages of sympatric salmonids

Aim: High repeatability among assemblages of closely related but ecologically distinct ecotypes implies predictability in evolution and assembly of communities. The conditions under which ecotype assemblages form predictably, and the reasons, have been little investigated. Here, we test whether repeatability declines as the number of ecotypes builds.Location: Postglacial lakes with a circumboreal distribution.Time Period: Data were extracted from studies published between 1982 and 2019.Major Taxa Studied: Ecotype assemblages from two Salmonid genera - Salvelinus and Coregonus. Fish in postglacial lakes commonly occur as pairs of ecotypes, typically with a pelagic and a littoral/benthic form, but in Salvelinus and Coregonus, assemblages commonly contain multiple sympatric ecotypes.Methods: We used a meta-analysis of Salvelinus and Coregonus to empirically assess how repeatability varies across assemblages of two to seven ecotypes. We examined repeatability of use of broad niche categories as well as underlying phenotypic traits.Results: Within Coregonus, repeatability across multi-ecotype assemblages did not break down with the addition of a third or fourth ecotype. However, in Salvelinus, repeatability was largely absent and independent of the number of ecotypes. Repeatability of trait frequency distributions was absent in both genera, yet associations between trait means and niche categories were evident, especially in Coregonus.Main Conclusions: These results show that repeatability can vary greatly between lineages; that repeatability need not break down as the number of ecotypes builds; and that high repeatability of broad niche categories may result despite marked differences in the underlying frequency distribution of trait means. These findings not only affirm the presence of repeatable ecotype assembly and early stages of divergence in postglacial fishes at a global scale, but also highlight variability among taxa and underlying phenotypic traits.