Abstract
Understanding the extent to which evolution is predictable is fundamental to advancing research on biodiversity origins. We studied the evolution of ecotypes in Arctic charr in 18 freshwater lakes and two evolutionary lineages. We find significant phenotypic parallelism across replicates, with the degree of parallelism most pronounced in adaptive traits associated with foraging. Genome-wide analyses of the underlying population variation revealed a strong influence of historical demography and gene flow. Some divergent genomic regions were significantly shared across ecotype replicates, though there was a high proportion of population-specific variation. Functional molecular bases to ecotypes in gene expression and biological pathways were extensively and significantly shared across ecotype pairs. These results demonstrate a spectrum of predictability, from low parallelism at the nucleotide level to high parallelism in regulatory molecular bases and functional adaptive aspects of morphology, and advance our understanding of parallel evolution through divergent evolutionary routes.