Grazing Gradients in Central Australian Rangelands: Ground Verification of Remote Sensing-Based Approaches.

Abstract

Remotely-sensed data collected by satellites have been proposed for investigating grazing effects across the large paddocks of arid Australia. These data are used to compute indices of vegetation cover which are then analysed with reference to patterns of grazing behaviour around watering points. Grazing pressure typically increases as water is approached, resulting in a decrease in herbage cover. This pattern of cover change is called a grazing gradient. The change in these gradients from a dry to wet period forms the basis for assessing land degradation as described in an accompanying paper. This study demonstrates that grazing gradients do exist, that they can be detected with field-based methods of data collection, and that there is close correspondence between ground data and indices of vegetation cover obtained from contemporary Landsat Multispectral Scanner satellite data. Field data representing aerial cover of the herbage and woody species layers were collected along transects radiating away from water at two sites grazed by cattle in central Australia. Graphical representation of the litter and herbage components demonstrate that gradients of decreasing cover attributable to increasing grazing pressure occur along all, or sections, of each transect. Highly significant correlations exist between the field data and satellite indices of vegetation cover. Localised shrub increase and patches of recent erosion obscured trends of increasing cover with distance from water on parts of some transects. Soil surface state (describing past erosion) was a significant covariate of cover change at one site. Our ability to characterise gradients of increasing vegetation cover with distance from water using both field and satellite data should mean that the grazing gradient method, when used with satellite data, is a suitable technique for assessing the extent of landscape recovery following good rainfall.