Fragmentation, disturbance and tree diversity conservation in tropical montane forests

Abstract

Summary Loss in forest cover associated with human activities leads to forest islands within a fragmented landscape. Anthropogenic disturbance may also directly alter tree species diversity. Habitat fragmentation and disturbance may have implications for biodiversity conservation and can affect a variety of population and community processes over a range of temporal and spatial scales. Effects are manifest both within and between fragments. The complexity of this process challenges the predictive value of simple models based on island biogeographical theory. We analysed data on tree species diversity from 195 field plots in 16 tropical montane forest fragments in the Highlands of Chiapas, Mexico. Within this area we identified five community types. Variation in species diversity between fragments in each community was analysed by fitting linear models in which area, core area, edge/area ratio, and mean proximity index were related to mean plot diversity. Variations within fragments were then analysed using linear mixed‐effects models in which the fixed effects were considered to be proximity to edge, canopy closure and a degradation index, whereas fragment‐level variation was modelled as a random effect. Effects of fragmentation per se, defined as the reduction in area and connectivity between fragments, were not detectable. Within‐fragment effects, however, were evident. Disturbance was associated with lower tree diversity at a local scale, and a notable impact on late successional species. While not statistically detectable, we expected fragmentation per se to be acting to reduce diversity in the long term. We therefore complemented our analysis by calculating the theoretical loss in species when the number of tree stems in a fragment was halved for each community type using resampled accumulation curves. These models predicted a maximum loss in transitional forest (12 species lost) and evergreen cloud forest fragments (seven to nine species lost) and a minimum loss in oak forest fragments (three species lost). Synthesis and applications. Our results call into question the utility of island theory when setting conservation priorities for tree species. In our study region, and elsewhere, tree diversity is most immediately threatened by the effects of within‐fragment disturbance. The few remaining areas of intact native forests should be prioritized for conservation regardless of their size and connectivity. If diversity is conserved within these fragments, the short‐term effects of landscape‐scale change may be minimized. They may then be reversed if long‐term restoration initiatives can be implemented.