Sample coverage affects diversity measures of bird communities along a natural recovery gradient of abandoned agriculture in tropical lowland forests

Tropical old‐growth forests continue to decline worldwide, resulting in a huge loss of biodiversity. The extent to which the expansion of second‐growth forests can counteract biodiversity loss is context‐dependent and controversial. To test the recovery of bird communities along a gradient from active pastures and cacao plantations, through regenerating forest on land last used for agriculture between 1 and 38 years ago, to old‐growth forest, we sampled simultaneous audio recordings from 66 plots, from which an expert identified all bird species detected at fixed time points throughout the day. The study area is characterized by typical small‐scale agriculture with remnant trees in the Ecuadorian Chocó Forest. To quantify different aspects of biodiversity, we used incidence‐based Hill numbers focusing on infrequent, frequent and highly frequent species in taxonomic, functional and phylogenetic diversity, considering sample coverage (an objective measure of sample completeness). Bird community composition changed with the regrowth gradient represented on the first axis of the ordination. Differences in bird communities were also very robust to changes in sample coverage. The sample coverage decreased significantly along the recovery gradient and affected the different measures of alpha diversity. Although the results controlled by sample coverage showed no change in taxonomic and phylogenetic diversity, the functional diversity of infrequent, frequent and highly frequent species decreased along the recovery gradient. Cacao plantations exhibited particularly high diversity values, highlighting the potential of these patches to support woodland and shrubland species in agriculture. Furthermore, several forest species regularly used the agricultural areas, attracted by remnant trees characteristic of the small‐scale agricultural landscape in our study region. Synthesis and applications. Our results highlight the importance of standardizing biodiversity measures and incorporating beta diversity in biodiversity monitoring. We demonstrate that taxonomic, phylogenetic and functional bird diversity can be high in secondary forests within smallholder agricultural landscapes. This underscores the potential for natural forest recovery, particularly when recovery patches are embedded within a forest matrix that includes old‐growth stands.