New technologies like ecoacoustic surveys promise time and cost efficiency for biodiversity assessments, serve as a basis for effective conservation policies, and are particularly appealing for remote and highly diverse tropical areas. Acoustic indices facilitate the analysis of large acoustic datasets but no consensus on their performance has been reached yet. We evaluated the efficacy of four acoustic indices (Acoustic Complexity Index, Acoustic Diversity Index, Acoustic Evenness Index, Acoustic Entropy) for sound data analysis and biodiversity assessments inside a national park and the agricultural mosaic landscape of north-eastern Madagascar, a global biodiversity hotspot. We used self-built sound recorders to continuously record soundscapes on 80 plots across seven land-use types (old-growth forest, forest fragment, forest–derived and fallow-derived vanilla agroforest, herbaceous and woody fallow, rice paddy) and compared index values between land–use types, assessed the correlation with bird species richness as measured by point counts, and related the acoustic indices to plot- and landscape-scale parameters. The Acoustic Diversity Index, Acoustic Evenness Index (inverse) and Acoustic Entropy were highest in old-growth forest and lowest for rice paddies and fallow land. Index values for structurally similar land-use types did not differ significantly from each other. The correlation of the three acoustic indices with bird species richness was strongest during daytime $(R^2\ge 0.30)$. Differences in the index values were best explained by land-use type and vegetation density. Our results showed that all investigated indices except the Acoustic Complexity Index were suitable biodiversity indicators for a tropical, agricultural landscape. Soundscape diversity was positively affected by plot-scale vegetation structure, emphasizing the importance of forests and particularly old-growth forest for conservation. We demonstrated that acoustic indices and sound recordings are a useful tool for assessing biodiversity in tropical agricultural mosaic landscapes. To realize the full potential of ecoacoustics in conservation, sampling guidelines and user-friendly analysis packages will be key to facilitate a wider implementation.

生态声学调查等新技术为评估生物多样性提供了时间和成本效益，是制定有效保护政策的基础，并且特别适合偏远和高度多样化的热带地区。声学指数有助于分析大型声学数据集，但尚未就其性能达成共识。我们评估了四个声学指数（声学复杂度指数，声学多样性指数，声学均匀度指数，声学熵）对国家公园内的声音数据分析和生物多样性评估以及马达加斯加东北部（全球生物多样性热点）的农业马赛克景观的有效性。 。我们使用了自建的录音机来连续记录七种土地利用类型（旧林，森林碎片，源自森林和源自休闲的香草农用林，草本和木质休闲，稻田），并比较了土地利用类型之间的指数值，评估了通过点数测量的与鸟类物种丰富度的相关性，并将声学指数与图和景观尺度参数。原始森林的声学多样性指数，声学均匀度指数（倒数）和声学熵最高，而稻田和休耕地则最低。结构相似的土地利用类型的指数值没有显着差异。在白天，这三个声指数与鸟类物种丰富度的相关性最强 评估了通过点数测量的与鸟类物种丰富度的相关性，并将声学指数与地积和景观尺度参数相关联。原始森林的声学多样性指数，声学均匀度指数（倒数）和声学熵最高，而稻田和休耕地则最低。结构相似的土地利用类型的指数值没有显着差异。在白天，这三个声指数与鸟类物种丰富度的相关性最强 评估了通过点数测量与鸟类物种丰富度的相关性，并将声学指数与地块和景观尺度参数相关联。原始森林的声学多样性指数，声学均匀度指数（倒数）和声学熵最高，而稻田和休耕地则最低。结构相似的土地利用类型的指数值没有显着差异。在白天，这三个声指数与鸟类物种丰富度的相关性最强 结构相似的土地利用类型的指数值没有显着差异。白天，三个声学指数与鸟类物种丰富度的相关性最强 结构相似的土地利用类型的指数值没有显着差异。在白天，这三个声指数与鸟类物种丰富度的相关性最强$（{\mathrm{[R}}^2\ge 0.30）$。最好用土地利用类型和植被密度来解释指数值的差异。我们的结果表明，除声学复杂性指数外，所有调查指标均是热带农业景观的合适生物多样性指标。声景多样性受到地块规模植被结构的积极影响，强调了森林尤其是老龄森林对保护的重要性。我们证明了声学指数和录音是评估热带农业马赛克景观中生物多样性的有用工具。为了充分发挥生态声学在保护方面的潜力，采样指南和用户友好的分析工具包将是促进更广泛实施的关键。