1. Global tropical forests have been modified and fragmented by commodity agroforests, leading to significant alterations in ecological communities. Nevertheless, these production landscapes offer secondary habitats that support and sustain local biodiversity. In this study, we assess community level and species-specific responses of amphibians to land management in areca, coffee and rubber, three of the largest commodity agroforests in the Western Ghats.
2. A total of 106 agroforests across a 30,000 km² landscape were surveyed for amphibians using a combination of visual and auditory encounter surveys. We used a Bayesian multi-species occupancy modeling framework to examine patterns of species richness, beta diversity, dominance structure and individual species occupancies. The influence of biogeographic variables such as elevation and latitude as well as microhabitat availability of streams, ponds and unpaved plantation roads were tested on amphibian species occupancy. 
3. Coffee agroforests had the highest species richness and lowest dominance when compared to areca and rubber. Beta diversity was highest in areca for within agroforest measures. Compared across agroforests, coffee had highest beta diversity with areca and rubber. Both elevation and latitude showed an overall positive association with amphibian occupancy although species-specific responses varied considerably. 
4. Microhabitat availability was one of the strongest predictors of amphibian occupancy, with mean community response being positive with presence of water bodies and roads. Pond presence increased species richness per site by 37% (species-specific responses in occupancy ranged from -2.7% to 327%). Stream presence alone did not change species richness but species-specific response ranged from -59% to 273%). Presence of plantation roads also increased species richness by 21.5% (species-specific response range -82% to 656%). Being unpaved with little vehicular traffic, plantation roads seem to provide additional habitats for amphibians. Presence of all three microhabitats at a site increased species richness by 75%.
5. Our study highlights the importance of land management strategies that maintain diverse native canopy and freshwater bodies and other microhabitats in sustaining amphibian fauna. Market driven land-use change from coffee to other agroforest types will have detrimental effects on amphibian communities and their long-term sustainability in the Western Ghats.

We surveyed 106 unique agroforests (33 areca, 33 coffee, and 40 rubber). Surveys were conducted between June and September, in the monsoon season, when amphibians are most active. Three teams of two trained observers conducted systematic, time-constrained searches between 19:00 and 23:00 hours. The length of the survey varied based on the size of the site from 30 minutes to 300 minutes, with an average survey time of 93 minutes per site. The observers walked at a consistent pace, in a single direction, along foot paths, roads, cultivated areas, and water bodies. Observers recorded all frogs detected using visual and auditory encounter surveys (VES/ AES). Encounter of a species by either method was considered as a detection and mid to high canopy species were only observed through auditory surveys. There were no spatial constraints set on auditory detections. We attempted to identify every individual encountered to the species-level based on external morphology and acoustic characteristics. This data consists of 26 species which were reliably identified in the field.