Evaluating factors affecting species detection using passive acoustic monitoring in neotropical forests: a playback experiment

Passive acoustic monitoring (PAM) has become a popular approach for terrestrial species monitoring. Studies determining detection ranges and predictors of detection probability are needed for accurate ecological modelling based on PAM data. We conducted a playback experiment to evaluate factors affecting call detectability in a neotropical forest. We used Geoffroys spider monkeys as a model species and broadcast their contact calls at seven distances from autonomous recording units (ARUs) positioned at different heights, in forest patches of different ages, and controlled for climate conditions during broadcasting trials. Using GLMMs (N = 1,342 data points), we found that call detectability increased the smaller the difference between the ARU and speaker height. We further found detection range to be wider the younger a surveyed forest patch is. To test whether these effects were related to structural signal characteristics, we included vocalisations from seven sympatric species in the experiment (N = 9,394). We found peak frequency to modulate the effect of ARU height difference, and peak frequency and bandwidth to modulate the effect of forest age on call detectability. Our study sheds light on factors influencing acoustic signal detectability and emphasises the need to consider climatic and habitat variables when designing PAM programmes.