Passive Acoustic Monitoring in Ranomafana National Park

Abstract

Vocally active species partition the soundscape to communicate on food availability, mate attraction and predation risk. Tropical forests, with high vegetation density, have many vocally active species, including many frugivores that help maintain the forest through seed dispersal or pollination. The tropical forests of Madagascar have been subjected to extreme anthropogenic disturbances leading to land degradation which threatens many native frugivorous species with extinction. To reforest and restore the tropical forests in Madagascar, more knowledge on the presence, behavior, and interaction of frugivorous species is needed. In this thesis I used a dataset gathered through Passive Acoustic Monitoring (PAM) in Ranomafana National Park (RNP) to characterize the soundscape and vocal activity of frugivorous bird species in RNP. I described daily and hourly variations in the soundscape with Sound Pressure Level (SPL) analysis in Kaleidoscope. I isolated the calls of individual frugivorous bird species using cluster analysis and pattern matching analysis in the programs Kaleidoscope and ARBIMON, respectively, and compared the acoustic niche space of different species using the activity package in R. I found a general increase in sound intensity between sunrise and sunset. However, the sound intensity variation across days was strongly influenced by geophysical sounds, especially rainfall. I successfully identified vocalizations using publicly available recordings of the four species Coracopsis nigra, C. vasa, Hypsipetes madagascariensis, and Philepitta castanea. The vocal activity of four frugivorous bird species showed little acoustic niche overlap in the temporal and spectral dimensions. With these results I showed that PAM can be used to observe frugivorous bird species in tropical forests to increase knowledge on these functionally important species. However, I recommend addressing some remaining technical and execution challenges to improve detection and study of tropical frugivores. First, a method to separate biological sounds from geophysical and anthropogenic sounds will help in observing vocal species. Second, there needs to be further study on how the location of the sensor influences sound quality and background noise of the recordings. Finally, I recommend collaboration with a local ornithologist to improve species identification based on vocalization. I have shown that particularly with these improvements, PAM represents a viable alternative to in-person observation of frugivore bird species in a dense tropical forest.