Increasing urbanization has led to large scale land-use changes, exposing persistent populations to drastically altered environments. Sensory pollutants, including low-frequency anthropogenic noise and artificial light at night (ALAN), are typically associated with urban environments and known to impact animal populations in a variety of ways. Both ALAN and anthropogenic noise can alter behavioral and physiological processes important for survival and reproduction, including communication and circadian rhythms. Although noise and light pollution typically co-occur in urbanized areas, few studies have addressed their combined impact on species’ behavior. Here we assessed how anthropogenic noise and ALAN can influence spatial and temporal variation in breeding activity of a wild frog population. By exposing artificial breeding sites inside a tropical rainforest to multiple sensory environments, we found that both anthropogenic noise and ALAN impact breeding behavior of túngara frogs (Engystomops pustulosus), albeit in different ways. Males arrived later in the night at their breeding sites in response to anthropogenic noise. ALAN, on the other hand, led to an increase in calling effort. We found no evidence that noise or light pollution either attracted frogs to or repelled frogs from breeding sites. Thus, anthropogenic noise may negatively affect calling males by shifting the timing of sexual signaling. Conversely, ALAN may increase the attractiveness of calling males. These changes in breeding behavior highlight the complex ways that urban multisensory pollution can influence behavior and suggest that such changes may have important ecological implications for the wildlife that are becoming increasingly exposed to urban multisensory pollution.

Datasets were collected by placing aritificial breeding sites within the tropical rainforest around Gamboa, Panamá. Using audio and video data, we collected data examining the presence/absence of frogs at breeding sites, the timing of breeding behavior, and multiple calling characteristics. All raw data was further processed in R, using the code provided. Additional information concerning methods can be found within the Methods section of our manuscript.