A remote sensing spatio-temporal framework for interpreting sparse indicators in highly variable arid landscapes

•Ecological indicators in highly variable landscapes need context for interpretation.•High temporal frequency satellite data provided such context for arid Australia.•Vegetation-dynamics based classes differed from commonly used bioregionalisation.•A case study presents long-term vegetation response of the new classes.•This defined the framework for interpretation of indicator data. The world's extensive and often remote arid landscapes are receiving increasing attention to maintain their ecological and productive values. Monitoring and management of these lands requires indicators and evidence of ecosystem condition and trend, generally derived from widely distributed and infrequently repeated site-based records. However adequate geographic representation and frequent site revisits are difficult to achieve because of the remoteness and vast extent of these landscapes. Interpreting such sparse ecological indicators is difficult, particularly within landscapes that are highly variable in space and time. To interpret ecological indicator data collected in such environments long-term patterns of natural landscape variability need to be understood. This paper presents a framework of landscape spatio-temporal variability within which to interpret ecological indicator data. This framework is based on long-term patterns of vegetation growth across the Australian arid zone, derived from twenty-five years of high temporal resolution National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) satellite imagery. We present a case study of the extensive Alinytjara Wilurara (AW) Natural Resource Management (NRM) region in far western South Australia to illustrate new insights about landscape function gained from this approach, and their implications for collection and interpretation of ecological indicator data. We illustrate how variability in vegetation response is expressed across the region, and how stratification based on active vegetation response differs from more commonly used biogeographic stratifications in this region. Lastly we demonstrate the unique patterns of long-term vegetation response for the major vegetation response classes. Average amount, seasonality, magnitude, timing and variability of vegetation response over time are used to characterise the natural “envelope” of variability of the new landscape classes. The study region showed low vegetation response in summer and higher response in winter. Onset o