Hierarchical divergence across genomic, phenotypic, and microbiome dimensions in two annual killifish species from Malawi
Divergence among populations and species commonly occurs across multiple biological levels, yet the extent to which genomic, phenotypic, and ecological dimensions are coupled remains poorly understood. We integrated whole-genome sequencing, multivariate morphology, population-level phylogenetic structure, and gut microbiome composition to evaluate divergence in two allopatric annual killifish species across spatially structured populations in Malawi. Geographic and hydrological structure emerged as the primary axis of divergence, with strong differentiation between species and among populations. Genome-wide analyses revealed consistent clustering among populations in ordination and phylogenetic analyses, indicating pronounced spatial structuring of genomic variation. Although genomic divergence was widespread, exon-level enrichment analyses revealed distinct signatures across evolutionary scales. Divergence within N. kirki was associated primarily with translation-related functions, whereas divergence within N. wattersi involved ATP biosynthesis and physiological homeostasis. Interspecific divergence was enriched for transcription factor activity and transcription factor binding, implicating regulatory evolution as a major component of species differentiation. Morphological variation was likewise strongly structured among populations and aligned with drainage systems and geographic regions but was not correlated with genome-wide genetic differentiation, indicating partial decoupling between genotype and multivariate phenotype. Gut microbiome composition represented a more environmentally responsive layer of divergence, broadly reflecting host species and drainage structure while exhibiting greater overlap among populations. These results support a hierarchical model in which geographic and hydrological structure organize stable genomic divergence, whereas phenotypic and microbiome variation represent increasingly context-dependent and only partially aligned biological layers. Our findings highlight the value of integrating multiple biological levels to understand how evolutionary processes shape divergence in natural populations.