Assessing tropical forests' climatic sensitivities with long-term data.

摘要

Analyses relating long-term records of tree growth to interannual climatic variation at La Selva, Costa Rica have revealed marked forest sensitivities to both temperature and dry-season intensity (Clark et al. 2010). The tropical-forest biome is certain to become warmer, and many areas may become drier. Testing the generality of the La Selva findings with similar analyses of field data from diverse forests across the biome will be a valuable next step. Based on our experiences during the La Selva studies, we propose that such assessments will need to address three issues. One is the number of repeat forest measurements. Short series of re-censuses can be an unreliable basis for assessing climatic sensitivities. For some key climatic factors (e.g., temperature), records consisting of fewer than 10-12 re-censuses can span limited climatic ranges, producing erratic and largely nonsignificant correlations. Multiyear census intervals exacerbate these data limitations. Second, different types of forest-growth data call for different analysis approaches. Cohort and tree-ring records need to be adjusted for ontogenetic growth changes, while stand-level data require taking into account potentially confounding influences from forest compositional changes, as from succession. Third, a reliable meteorological record is critical. Poor-quality or internally inconsistent climatic records can fatally corrupt assessments of forest sensitivities. To be usable in such analyses, the meteorological record requires data quality control, gap filling, and adjustments to maintain the record's internal consistency in the face of commonly occurring methods changes (instruments, siting). We illustrate these issues using analyses of the long-term La Selva records.Digital Object Identifier http://dx.doi.org/10.1111/j.1744-7429.2010.00654.x