Animals show a rich diversity of signals and displays. Among the many selective forces driving the evolution of communication between individuals, one widely recognized factor is the structure of the environment in which signals are produced, transmitted and received. In particular, animals communicating by sounds often emit acoustic signals from specific locations, such as high up in the air, from the ground or in the water. The properties of these different display sites will impose different constraints on sound production and transmission and may therefore drive signal evolution. Here, we used comparative phylogenetic analyses to assess the relationship between the display site properties and the structure of reproductive calls from 161 frog species from the frog families Ranidae, Leptodactylidae and Hylidae. Specifically, we compared the dominant frequency of species that vocalize from aquatic versus non-aquatic sites, and its relation with body size. We found that the dominant frequency of frogs calling from the water was lower than that of species calling outside of the water, a trend that was consistent across the three families studied. Furthermore, phylogenetic path analysis revealed that the call site had both direct and indirect effects on the dominant frequency. Indirect effects were mediated by call site influencing male body size, which in turn was negatively associated to call dominant frequency. Our results suggest that properties of display sites can drive signal evolution, most likely through morphological constraints, in particular the ones imposed on the sound production mechanism. Also, variation in body size between calling sites explained some of the differences we found in call frequency, highlighting the relevance of the interplay between morphological adaptation and signal evolution. Changes of display site may therefore have important evolutionary consequences, as it may influence sexual selection processes and ultimately may even promote speciation.

For each family, we collected data on the snout–vent length, dominant frequency and calling site. Most of the information was obtained from the literature or other digital sources (see below). Personal measurements made by the authors of the present article were also included. If searched in the literature, body size and call dominant frequency were obtained from other comparatives studies and books. We restricted our search to body size of males and the dominant frequency of advertisement vocalizations. Information on calling sites was obtained mainly from verbal descriptions of frog vocal behaviour present in the literature, the specialized website AmphibiaWeb, and from the personal experience of the authors. Multimedia information available from AmphibiaWeb and YouTube, such as pictures and videos of calling males, was used to confirm ambiguous verbal descriptions. For a few species (19 of 175 species), vmultimedia information was used as the sole criterion for calling site assignment. Each species was assigned to one of three possible calling site categories: (1) aquatic, (2) nonaquatic and (3) mixed. Aquatic species included frogs that vocalize either standing in water, or floating on the water surface. The nonaquatic category included species that call from the ground, or from perched positions on trees or rocks without direct contact with water. Species calling from cavities dug in the ground or cavity-like structures on vegetation (e.g. the axils of bromeliads) were also included in the nonaquatic category. The few species for which both aquatic and nonaquatic calling was described were assigned to the mixed category.