Optimal Auditory Sensitivity Under Variable Background Noise Conditions: A Theoretical Model

Abstract

Sound signals relevant for mating and survival are very often masked by background noise, which makes their detection and recognition by organisms difficult (i.e., communication often takes place under partially masked conditions; e.g., Zwicker and Fastl 1990). Ambient noise (AN) varies in level and shape among different habitats, but remarkable variations in time and space also occur within the same habitat. Variable AN conditions mask hearing thresholds of the receiver in complex and unpredictable ways, thereby causing distortions in sound perception. For instance, sound and speech recognition in animal and human subjects quickly deteriorates with decreasing signal-to-noise ratio under nonstationary noise conditions. Furthermore, no sound with energy lower than the noise can be heard. These observations suggest that when communication takes place in a noisy environment, a highly sensitive system may confer no advantage to the receiver compared with a less sensitive one (Hawkins and Myrberg 1983). Fishes live in all types of underwater habitats differing widely for AN conditions, from quiet deep oceans and shallow ponds to noisy coastal waters and stony streams. Notably, they show an impressive variety of audiograms that differ in shape, level, and frequency range. Lugli et al. (2003) showed that the best hearing range and the dominant frequencies of sounds of the two freshwater gobies (Padogobius bonelli and Gobius nigricans) fit within a relatively quiet window in the low-frequency spectrum of the stream AN. Amoser and Ladich (2005) noted that teleosts with the best hearing (e.g., Cyprinids) live in quieter habitats than those with poor hearing abilities (e.g., Percids). These findings suggest that AN may be an important selective factor in the evolution of hearing sensitivity of a species. The way it would select for the level of hearing sensitivity of a species is unclear, however. Here I describe a simple fitness model for the detection and discrimination (and recognition) of sounds under variable AN conditions. I assume that noise masking significantly impairs all the above functions. The model predicts high sensitivity (i.e., low hearing thresholds) as the best strategy for species living in quiet habitats and low sensitivity (i.e., high hearing thresholds) as the best strategy for those living in noisy ­habitats, provided the cost of incorrect discrimination is not low.