Acoustic indices fail to represent different facets of biodiversity

• 11 acoustic indices were calculated in three habitat types in Italy. • 3 diversity metrics were calculated from recordings for acoustically active birds. • Relationships were fitted between the diversity metrics and the acoustic indices. • Relationships were inconsistent and often contrary to what is theoretically expected. • Acoustic indices were not good proxies of avian diversity metrics. Acoustic indices are cost-efficient tools for monitoring acoustically active taxa. Despite their increasing popularity, their suitability as biodiversity proxies remains debated due to theoretical and practical issues. Existing research has predominantly focused on assessing their capacity to capture species richness and abundance of sounds, with limited attention given to other biodiversity facets like phylogenetic and functional diversity. Yet, phylogenetic distance and trait diversity may influence acoustic communication, so acoustic indices may be better suited to capture their changes than changes in the mere number of species. Here, we investigated the ability of acoustic indices to represent the species, phylogenetic, and functional diversity of acoustically active avian species across three habitat types in Italy: urban, riparian forest and deciduous woodland. Recordings were collected at dawn by a trained ornithologist, who also identified acoustically active species captured, first in the field and secondarily from recordings. We then applied eleven widely used acoustic indices for each recording and tested their ability to capture differences in species, functional and phylogenetic diversity across sites (i.e., listening points) within each habitat type using a linear mixed effect model. We then reduced the acoustic indices dimensionality to a few axes and repeated the analysis to test if compound indices delivered comparable information. Species richness and phylogenetic diversity correlated with at least one acoustic index across the three habitat types, while functional diversity correlated with an index only in the urban habitat. However, the nature of the relationships identified was often opposite to what expected by the indices’ formulation. Furthermore, across the three habitat types the interrelations among indices varied, as well as the observed relationships with diversity indices. Compound indices also exhibited inconsistent relationships across sites and contrary to what expected based on their correlation with acoustic indices. Our results cast doubts on the suitability of acoustic indices as proxies for biodiversity for biodiversity monitoring. Additionally, the inconsistent relationship among indices across the three areas suggests they are unsuitable for geographic comparisons. Temporal comparisons are possible but require local validation and may need adjustments over time of the community composition changes. Despite the attractiveness of acoustic indices as easy and cheap tools for biodiversity monitoring, we urge caution on their application.