Abstract
Rod spectral sensitivity data ( λ max ), measured by microspectrophotometry, were compiled for 403 species of ray‐finned fishes in order to examine four hypothesized predictors of rod spectral sensitivity (depth, habitat, diet and temperature). From this database, a subset of species that were known to be adults and available on a published phylogeny ( n = 210) were included in analysis, indicating rod λ max values averaging 503 nm and ranging from 477 to 541 nm. Linear models that corrected for phylogenetic relatedness showed that variation in rod sensitivity was best predicted by habitat and depth, with shorter wavelength λ max values occurring in fishes found offshore or in the deep sea. Neither diet, nor the interaction of diet and habitat, had significant explanatory power. Although temperature significantly correlated with rod sensitivity, in that fishes in temperate latitudes had longer wavelength rod λ max values than those in tropical latitudes, sampling inequity and other confounds require the role of the temperature to be studied further. Together, these findings indicate that fish rod λ max is influenced by several ecological factors, suggesting that selection can act on even small differences in fish spectral sensitivity.