Modeling and measurement of structural changes at a landscape scale in dryland areas

We present a technique to quantify and model the intensity of structural changes produced by management of dry grazing lands at a landscape scale. The technique is illustrated with the analysis of digitized black-white (b/w) imagery and an application to the study of changes induced by grazing gradients. Structural changes in patchy vegetation canopies were studied in the Patagonian Monte (Chubut, Argentina) at two resolution scales by means of linear transects in the field (50 m) and others drawn on aerial b/w photographs (2-5 km) of grazed paddocks. Spatial series of plant cover values along transects in the field and on photographs were analyzed with standard techniques of spectral analyses, including auto-correlation spectra and Fourier transforms. In order to test the internal consistency of the techniques used, synthetic plant canopies with patches of varying cover and size were generated by means of a stochastic model of plant growth under different stocking rates or after varying periods of recovery. The behavior of the simulation model is consistent with the observed dynamics of plant canopies in semiarid environments. There is a consistent relation between the number and geometric properties of plant patches (patch number, patch size, patch connectivity) and the signal/noise ratios of the Fourier decomposition describing plant density data. Signal/noise ratios corresponding to plant cover data in paddocks with different grazing treatment are consistent with the assumptions derived from modeled canopies, as well as those estimated from optical density of b/w aerial photographs of paddocks. We tested the hypotheses that patch arrangements as quantified by the signal/noise ratios vary in accordance with grazing gradients in paddocks with a permanent corner-located watering point. The use of digitized b/w images allows inspecting permanent changes over time periods when other types of images were not yet available.