Regional patterns and controls of biomass in semiarid woodlands: lessons from the Northern Argentina Dry Chaco

Abstract

Land use change, particularly in forested ecosystems,has a direct impact on the global carbon cycle. Consequently, the regional assessment of biomass and the understanding of its current spatial controls are research priorities for regional ecology and land use. Field data and satellite imagery were combined here to map woodlands and estimate their above-ground biomass (AGB) in the Dry Chaco ecoregion of northern Argentina. Allometric equations were used to derive AGB from diameter at breast height data collected at 50 samples during 2007. In order to generate the AGB regional map, this information was later associated with MODIS-Terra spectral data (NDVI) using the Random Forest (RF) method. Finally, AGB spatial patterns were associated with potential biophysical and human controlling factors through correlation and regression analyses. Results indicate that the use of RF and NDVI of the dry season derived from MODIS-Terra was suitable to map regional AGB, what makes this methodology applicable to other dry woodlands. The RF model used to map AGB showed a mean deviation of 2.9 % and a precision of 15 % for one prediction. At this regional scale of analysis, biophysical rather than human factors controlled AGB spatial patterns, in part because the region includes a wide range of environmental situations. Warmer conditions showed a higher biomass, suggesting an energetic limitation for AGB accumulation. However, human controls (distance to towns, cultivation, and roads) also conditioned AGB patterns, suggesting lower AGB values near cultivated areas. The relation between AGB and water availability was surprisingly weak, but partially obscured by the land use history and degradation due to extensive cattle ranching. We propose that a combination of environmental factor and land use affects the AGB regional patterns and promotes unexpected relationships with environmental factors. This work represents the first spatially explicit AGB (patterns and controls) analysis for an extensive subtropical dry woodland area (113,000 km2) and shows how biophysical and human factors co-control regional patterns.