Forest diversity and conservation in the western Amazon based on tree inventory and remote sensing data

This dissertation contributes to debates in conservation biogeography by examining the spatial heterogeneity of local and regional tree diversity feature using ground and remotely sensed data, and by taking approaches to design a spatially explicit landscape zonation map for future conservation planning in western Amazon, one of the most biodiverse regions on Earth. Fine scale tree diversity and conservation-related studies took place in tropical rainforests in southeastern Ecuador, whereas coarse scale tree diversity research was conducted using data from eastern Ecuador and northern Peru.

The lack of species assemblages within three 1-ha tree inventory plots in southeastern Ecuador and the weak correlations with biophysical environment implied that neutral processes may contribute to species diversity. In contrast, differences in species assemblages between plots corresponded to relative geographic locations of the plots, indicating that geographic distance or dispersal limitation may play an important role influencing diversity patterns at a regional scale.

Species of high local abundance was found in 1-ha tree inventory plots in western Amazon. Changes in density of locally abundant species between western and eastern plots indicated that some species may have limited distributions. Shifts in species dominance and the significant relationship between floristic variation and geographic distances between plots implied dispersal limitation. Variation in rainfall showed significant relationship with species composition. Therefore, dispersal limitation and precipitation seasonality are potentially the most significant factors that contribute to spatial differences in tree diversity in western Amazon.

Characteristics of canopy shadows and palm stem density based on fine-resolution aerial photographs were characterized as exploratory analyses to extract alpha and beta diversity features using remotely sensed data. A zonation map design using multispectral habitat classification and other remote sensing data performed well in its spatial arrangement when potential indigenous land use was integrated. Based on the results of analyses for conservation biogeography, this dissertation concludes that local and regional tree diversity may be influenced by dispersal limitation and seasonality, and that the application of remote sensing for biodiversity conservation is feasible in very species-rich forests.