Acoustic indices as proxies for biodiversity in certified and non-certified cocoa plantations in Indonesia

Acoustic indices allow time efficient analysis of large acoustic datasets obtained from passive acoustic monitoring, but results regarding their effectiveness in assessing biodiversity are inconsistent. We evaluated the efficacy of six acoustic indices (ACI, ADI, AEI, H, BI, NDSI) for studying bird and structural diversity in 51 cocoa plantations, 24 of which were certified by Rainforest Alliance, in Luwu Timur, Sulawesi, Indonesia. We used linear models to assess the correlation of index values with bird species richness, and linear mixed models to test the influence of canopy closure, shade tree basal area, distance to primary forest and tree cover in a 200-m buffer on index values. Bird species richness was positively correlated with BI (p = 0.02) and negatively with H (p = 0.03), yet predictive power was low (R² = 0.10 and 0.09, respectively). Acoustic indices did not differ significantly for certified cocoa plantations. Tree cover within the 200-m buffer moderately well predicted ACI values (marginal R² = 0.37) while for the other indices effect sizes were low or correlations were not significant. Comparing our results to other studies, acoustic indices may reflect biodiversity across land uses, but were of limited value for tracking subtle differences in cocoa plantations in Sulawesi. Future studies may include more land uses (i.e. rice paddies, secondary forest, oil palm) as well as more taxa (i.e. insects). More research is needed on the comparability of acoustic indices, as we found them to be influenced by recording equipment and calculation settings.