Technical Abstract: When choosing which environmental attributes to monitor in assessing disturbance, it is important to consider not only which metric will provide the most sensitive indicator of disturbance, but also the spatial considerations of where in the landscape that metric will be most responsive to change. Degradation in landscapes is often unevenly expressed because: (i) disturbance is spatially localised, (ii) landscape elements differ in their sensitivity to disturbance, and (iii) degradation following localised disturbance is spatially contagious. The spatial heterogeneity of degradation has proven to be a key obstacle to rangeland monitoring (e.g. where the initiating processes of broad-scale degradation are concentrated inlandscape locations that are not detected by surveys) but can also provide opportunities to focus monitoring efforts. We propose that the effectiveness of monitoring could be enhanced by identifying and selectively monitoring ‘indicator patches’, i.e. specific landscape locations that provide the most management-relevant and timely information about the consequences of a monitored disturbance. We tested and demonstrated the utility of the ‘indicator patch’ concept in the rangelands of the Succulent Karoo in southern Africa. We contrasted the grazing response of dominant ‘representative’ vegetation, with responses of interspersed patches of distinct vegetation associated with zoogenic mounds. Since mound vegetation is more palatable and preferentially grazed by sheep, we tested whether mounds could serve as ‘indicator patches’ in providing a sensitive measure of grazing disturbance. Percentage canopy cover measurements in dominant off-mound vegetation provided a poor indicator of grazing disturbance (although more intensive plant size measurements did reveal grazing impacts on plant population dynamics). In contrast, vegetation on mounds displayed patterns of changes in speciesabundances that were easier to detect and useful for interpreting and quantifying the effects of grazing. Mound vegetation could, therefore, be used as ‘indicator patches’ and targeted for exclusive sampling as a sensitive method for monitoring rangeland condition and detecting early warnings of vegetation change. This approach could be widely employed to better harness the extensive knowledge base regarding the patchy, spatially localised nature of degradation-initiating processes in numerous other landscapes. Routinely incorporating this understanding into the design of monitoring programs could improve the effectiveness of sampling effort, allow detection of more subtle trends (changes), and provide earlier warning of impending degradation so remedial action can be taken before degradation becomes severe and widespread.