The conservation of tropical biodiversity depends not only on forest remnants, but also on anthropogenic land covers. Some shade crops are considered wildlife-friendly agroecosystems, but their conservation value is context- and taxon-dependent. Amphibians and reptiles have received less attention despite their high sensitivity to habitat disturbance. We determined the conservation value of lightly-harvested rustic cocoa plantations for herpetofauna in the Lacandona region, Mexico. We compared 12 environmental variables between habitats. Then, we compared the abundance, species number and composition of amphibian and reptile assemblages. Within each habitat, we explored the relationships between environmental variables and abundance and species number. Tree density, litter cover and litter depth were higher in cocoa. Abundance of reptiles and amphibians were higher in cocoa than forest; species number did not differ. Habitat explained some of the variation (8%) in assemblage composition. In cocoa, amphibian abundance was positively related to canopy height and the presence of a humus layer, while reptile abundance was negatively related to relative humidity. We conclude that lightly-used rustic cocoa plantations can be suitable habitat for forest herpetofauna. As long as cocoa plantations do not replace existing forest cover, they can play an important role in the design of wildlife-friendly tropical landscapes.

Habitat fragmentation and deforestation are thought to disrupt host–parasite interactions and increase the risk of epizootic outbreaks in wild vertebrates. A total of 220 individuals from three species of African rain-forest bird (Andropadus latirostris, Andropadus virens, Cyanomitra obscura), captured in two pristine and two agroforests in Cameroon, were screened for the presence of avian haemosporidian parasites (species of Plasmodium and Haemoproteus) to test whether habitat differences were associated with differences in the prevalence of infectious diseases in natural populations. Thirteen mitochondrial lineages, including 11 Plasmodium and two Haemoproteus lineages were identified. Whereas levels of Haemoproteus spp. infections were too low to permit analysis, the prevalence of infections with Plasmodium spp. reached significantly greater levels in undisturbed mature forests. Importantly however, the significant association between forest type and parasite prevalence was independent of host density effects, suggesting that the association did not reflect changes in host species composition and abundance between forest types. Our results illustrate how characterizing land-cover differences, and hence changes, may be a prerequisite to understanding and predicting patterns of parasite infections in natural populations of rain-forest birds.