Detecting landscape-level changes in tree biomass and biodiversity: methodological constraints and challenges of plot-based approaches

Abstract

Understanding how human-impacted landscapes are changing is crucial for effective adaptive management and payment for ecosystem services programs. Landscape-level shifts in land use pose challenges not seen in typical ecological studies of well-protected forests. In human-modified landscapes, forests are often monitored using unique sets of randomized plots at each visit rather than re-censusing in the same permanent plots. We contrast field-based forest change monitoring using these two techniques and investigate whether sampling more plots or bigger plots better detects forest changes. Our empirical analysis employs long-term data sets from old-growth, second-growth, and managed tropical forests. We find that resampling in permanent plots reduces variation among subsequent censuses, but more importantly, it enables more powerful statistical tests. Increasing the number of plots improves detection of forest biomass changes more effectively than enlarging existing plot sizes, cost considerations being equal. This effect arises from more extensive capture of spatial heterogeneity by sampling in a greater number of locations. We further show that typical sampling techniques poorly assess the biodiversity of tropical forests and struggle to identify big changes in populations of common species. We conclude with practical suggestions for forest sampling in human-impacted tropical landscapes, including defining monitoring goals and delineating forests vs. entire landscapes as study areas.