Get access

Pasture degradation in the central Amazon: linking changes in carbon and nutrient cycling with remote sensing


G. P. Asner, Carnegie Institution, Stanford University 260 Panama Street Stanford, CA 94305, USA, tel. +1 650 325 1521, fax +1 650 325 6857, e-mail:


The majority of deforested land in the Amazon Basin has become cattle pasture, making forest-to-pasture conversion an important contributor to the carbon (C) and climate dynamics of the region. However, our understanding of biogeochemical dynamics in pasturelands remains poor, especially when attempting to scale up predictions of C cycle changes. A wide range of pasture ages, soil types, management strategies, and climates make remote sensing the only realistic means to regionalize our understanding of pasture biogeochemistry and C cycling over such an enormous geographic area. However, the use of remote sensing has been impeded by a lack of effective links between variables that can be observed from satellites (e.g. live and senescent biomass) and variables that cannot be observed, but which may drive key changes in C storage and trace gas fluxes (e.g. soil nutrient status). We studied patterns in canopy biophysical–biochemical properties and soil biogeochemical processes along pasture age gradients on two important soil types in the central Amazon. Our goals were to (1) improve our understanding of the plot-scale biogeochemical dynamics of this land-use change, (2) evaluate the effects of pasture development on two contrasting soil types (clayey Oxisols and sandy Entisols), and (3) attempt to use remotely sensed variables to scale up the site-specific variability in biogeochemical conditions of pasturelands.

The biogeochemical analyses showed that (1) aboveground and soil C stocks decreased with pasture age on both clayey and sandy soils, (2) declines in plant biomass were well correlated with declines in soil C and with available phosphorus (P) and calcium (Ca), and (3) despite low initial values for total and available soil P, ecosystem P stocks declined further with pasture age, as did a number of other nutrients. Spectral mixture analysis of Landsat imagery provided estimates of photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV) that were highly correlated with field measurements of these variables and plant biomass. In turn, the remotely sensed sum PV+NPV was well correlated with the changes in soil organic carbon and nitrogen, and available P and Ca. These results suggest that remote sensing can be an excellent indicator of not only pasture area, but of pasture condition and C storage, thereby greatly improving regional estimates of the environmental consequences of such land-use change.