1. Positive allometry has been considered a hallmark of sexual selection whereby larger males of superior condition develop disproportionately larger ornaments for their body size compared to smaller males of poorer condition.

2. Yet many structures known to be sexually selected often exhibit other allometric patterns. This has led to controversy over the utility of allometry in adequately capturing the signature of sexual selection, particularly if static (within population) and evolutionary (across species) allometries are functionally constrained by stabilising natural selection.

3. To investigate this, we evaluated the allometries of ornamental head crests and dorsal fins across multiple species of blenny fish. In particular, we compared species that occupied an aquatic environment—where swimming performance was expected to have constrained ornament size—with species that have transitioned onto land where such biomechanical constraints on ornament size have been removed.

4. Static allometries of both head crest and dorsal fin ornaments were found to be positive in males, but less so in females, across all species examined. This was consistent with the allometric theory of sexual selection that predicts positive allometry specifically in male ornamentation. Nevertheless, male allometric slopes were constrained in aquatic species whereas males of terrestrial species were free to exaggerate the size of their ornaments. Natural selection therefore appears to suppress ornament size in aquatics because of the biomechanical constraints associated with swimming. These differences in within-population static allometry between aquatic and terrestrial species in turn manifested in a greater across-species evolutionary allometric intercept, but not slope, for terrestrial species relative to aquatic and species.

5. These findings indicate that the study of ornament allometries can provide useful insights into the role of sexual selection on ornament elaboration and also help reveal the presence of opposing natural selection that might result in alternative allometric patterns. The relationship between static and evolutionary allometries remains complex, and our results caution against the interpretation of evolutionary allometry in the absence of a clear understanding of the underlying static allometries associated with it.