Would protecting tropical forest fragments provide carbon and biodiversity cobenefits under REDD+?

Tropical forests store vast amounts of carbon and are the most biodiverse terrestrial habitats, yet they are being converted and degraded at alarming rates. Given global shortfalls in the budgets required to prevent carbon and biodiversity loss, we need to seek solutions that simultaneously address both issues. Of particular interest are carbon‐based payments under the Reducing Emissions from Deforestation and Forest Degradation (REDD+) mechanism to also conserve biodiversity at no additional cost. One potential is for REDD+ to protect forest fragments, especially within biomes where contiguous forest cover has diminished dramatically, but we require empirical tests of the strength of any carbon and biodiversity cobenefits in such fragmented systems. Using the globally threatened Atlantic Forest landscape, we measured above‐ground carbon stocks within forest fragments spanning 13 to 23 442 ha in area and with different degrees of isolation. We related these stocks to tree community structure and to the richness and abundance of endemic and IUCN Red‐listed species. We found that increasing fragment size has a positive relationship with above‐ground carbon stock and with abundance of IUCN Red‐listed species and tree community structure. We also found negative relationships between distance from large forest block and tree community structure, endemic species richness and abundance, and IUCN Red‐listed species abundance. These resulted in positive congruence between carbon stocks and Red‐listed species, and the abundance and richness of endemic species, demonstrating vital cobenefits. As such, protecting forest fragments in hotspots of biodiversity, particularly larger fragments and those closest to sources, offers important carbon and biodiversity cobenefits. More generally, our results suggest that macroscale models of cobenefits under REDD+ have likely overlooked key benefits at small scales, indicating the necessity to apply models that include finer‐grained assessments in fragmented landscapes rather than using averaged coarse‐grained cells.