Inference of Cichlid Speciation Patterns is Dependent on Spatial Covariance and Species Delimitation
Macroevolutionary analyses typically treat species as discrete units and account for shared evolutionary history. However, speciation is a continuous process and taxa are often spatially clustered, potentially biasing inferences of diversification. Here, we investigate how species delimitation and spatial non-independence influence speciation dynamics and inferred drivers using cichlid fishes as a model system. Using a phylogeny and trait dataset of 1,712 species, we first generated a reduced dataset of 820 species by removing incipient species based on known breeding compatibilities. We then fit phylogenetic and spatiophylogenetic models using an integrated nested Laplace approximation framework to jointly account for phylogenetic and spatial covariance. We find that the treatment of incipient species and spatial non-independence both alter speciation patterns and inferred drivers. Analyses of the full phylogeny identified strong trait associations and spatial hotspots driven by young adaptive radiations in Lake Victoria and Lake Malawi, whereas removing incipient species and accounting for spatial non-independence reduced extreme speciation rates, weakened or removed trait effects, and largely eliminated spatial hotspots. These results demonstrate that macroevolutionary inference is sensitive to species delimitation and spatial structure, highlighting the need to consider the influence of incipient species and spatial covariance in comparative analyses.