Long-term global tropical forest trends are limited in their capacity to assess ecosystem changes due to human impacts and climatic events. Two main limitations for national inventories, the basis for global scale forest assessments such as those compiled by United Nations Food and Agriculture Organization (FAO), are: (1) very few tropical countries have complete inventories of their forest resources; and (2) national forest cover assessments for some countries are derived from large scale remotely sensed forest cover analyses, which limit our understanding of biodiversity at local and regional spatial scales. My work addresses the use of natural forest management plans (NFMPs) as a potential framework to assess tree biodiversity in the Central Volcanic Cordillera Conservation Area in Costa Rica (ACCVC). NFMPs are the basis for selective logging practices throughout the tropics. They summarize information about tree species abundance and spatial distribution in forest units based on a forest inventory compiled for a specific diameter threshold (e.g. 30 and 60 cm in Costa Rica). In this work, I have created a preliminary Forestry Geographic Information System (FGIS) that encompasses tree species abundance and distribution based on NFMPs between 1995 and 2005. The FGIS analysis shows that management trends were driven by government incentives during the first period while a sudden decline in the second period was perhaps due to social groups' critiques on non-sustainable management practices. In addition, forest management is being carried out in only a small fraction of the total forest area, with a primary focus on the largest forest remnants in the ACCVC (21 out of 630 fragments). When I compared tree diversity assessments for eight management units to a taxonomically developed independent tree inventory (≥ 10 cm diameter) I found that by using the 30 cm diameter threshold, NFMPs lack sufficient sample size to characterize species richness. Nonetheless, with NFMP data it is possible to assess local diversity for trees ≥ 30 cm diameter and potentially regional diversity for trees > 60 cm diameter. Moreover, the use of hyperspectral data shows that units logged at different times are distinguishable from imagery collected during a single time frame.
