|Title||Spatial autocorrelation of shrub cover in an arid rangleand ecosystem from 1937-2008|
|Publication Type||Conference Proceedings|
|Year of Publication||2010|
|Authors||Browning D.M., Laliberte, Andrea S., Rango A., Moreno A.|
|Conference Name||IEEE Transactions on Geoscience and Remote Sensing|
|ARIS Log Number||250340|
|Keywords||remote sensing, shrub cover|
The proliferation of trees and shrubs in grassland and savanna ecosystems has been widely observed, although future trajectories remain difficult to ascertain. Spatial manifestations of shrub proliferation bear relevance on surface hydrology and the spread of disturbance. We quantify changes in spatial autocorrelation in shrub cover in an arid grassland ecosystem to test the validity of a conceptual framework highlighting the role of shrub patch dynamics in the shrub encroachment process. We build upon the framework to suggest how the encroachment and stabilization phases manifest spatially. We suggest that indicators of spatial autocorrelation illustrate where local interactions are more prevalent, thereby functioning as an indicator of biotic regulation and stabilization in shrub cover. We analyzed binary maps of shrub cover from 1937 to 2008 derived from time series aerial photography. Spatial analyses entailed resampling 1-m depictions of shrub patches to percent shrub cover within 20-m X 20-m quadrats to examine the spatial autocorrelation with Moran’s I using GeoDa software. We decomposed the global index of spatial association into contributions from local neighborhoods using Local Indicator of Spatial Association, which quantifies the extent to which cells are similar to or different from their neighbors. Shrub cover in 1937 was highly heterogenous and became more dispersed and less patchy over the 71 years. As shrub cover increased, it became increasingly positively autocorrelated with Moran’s I increasing from 0.383 in 1937 to 0.742 in 2008. We found no clear evidence of self-organization, although we did detect clear evidence of local effects that resulted in broad scale patterns with hot-spots of low and high cover. Assessment of shrub patch dynamics (i.e., coalescence, fragmentation of patches) is on-going to identify mechanisms associated with changes in autocorrelation.