In many arid shrublands, soil resources are patchily distributed as shrub canopies allow for accumulation of organic material. The concept of these “islands of fertility” in association with soil mounds under desert shrubs is well studied, as is the effect of heterogeneous resource availability on native and invasive annual plants. While these soil nutrient patterns have previously been quantitatively characterized using randomly positioned samples and geostatistics, our research develops a predictive model that explicitly considers the location and canopy size of neighboring shrubs.
At 1-hectare study sites in both the Mojave and Sonoran deserts, we placed Plant Root Simulator TM probe ion-exchange membranes spaced 20cm along transects extending north and south from 18 creosote bush (Larrea tridentata) soil mounds to measure nutrient availability. We modeled the decline of nutrients with distance from the focal shrub using an exponential decay function within the context of nonlinear, hierarchical mixed models that included the effects of transect direction and shrub canopy size. After model equations were developed for a nutrient at a site, we estimated nutrient levels throughout the site at a resolution of 20cm, calculating nutrient values for locations given distance and direction from nearby Larrea.
Results/Conclusions
Of the nutrients sampled, nitrogen and potassium responded strongly to distance from focal shrubs in both the Mojave and Sonoran sites. At the Sonoran site, nitrogen and potassium levels associated with a fertility island were dependent on the shrub’s canopy size and had different patterns on the north and south sides of the shrub. At the Mojave site, nitrogen levels depended on the canopy size, but were consistent on both sides of the shrub. Potassium, as in the Sonoran, depended on both canopy size and on direction.
Using information on the location and canopy size of all Larrea at each site (303 in the Mojave and 716 in the Sonoran), we applied the predictive models to the entire hectare, which produced a map showing nutrient “hotspots” centered on clusters of Larrea. These models were able to predict up to 60% of the variation in nutrient levels the following growing season. The results of our predictive nutrient models will be used to determine patterns of annual plant growth and fuel accumulation that will be used to study the potential of fire risk.