Regime shifts from grass- to shrub-dominated states are widespread in arid and semiarid regions globally. These patterns of grass production and shifts to shrub dominance are spatially variable and correlate weakly with precipitation, suggesting that processes at different spatial and temporal scales interact to create spatially heterogeneous patterns of grass recruitment and growth. Focusing on a 5-year wet period at the Jornada Basin LTER site in southern New Mexico, we tested the hypothesis that a dynamic template of soil characteristics and spatial context mediate the response of grasses to periods of elevated precipitation. Data consisted of Normalized Difference Vegetation Index (NDVI) measurements from repeat satellite images before and after semiannual growing seasons, soil measurements from 75 locations across a grassland-shrubland ecotone, and monthly precipitation totals from 14 nearby rain gauges. Spatial arrangement of grasses, shrubs, and bare gaps were quantified from an object-based classification of a satellite image from immediately before the extended wet period.
Results/Conclusions
During this 5-year period of elevated precipitation, the greatest grass establishment and growth occurred at locations already dominated by grasses. Production also increased in some areas of mesquite (Prosopis glandulosa) dunes where fine-scale processes facilitating recruitment are normally overwhelmed by broader-scale drivers such as wind erosion. On soils with greater water-holding capacity, these fine-scale processes propagated spatially to influence patterns of recruitment and growth. On soils with lower water-holding capacity, grasses were unable to exploit an increase in precipitation and exhibited little increase in primary production. These results suggest that focusing on only fine-scale processes, such as competition, is of limited utility when cross-scale interactions govern observed patterns. A more explicit focus on multiple interacting spatial and temporal scales is often necessary to more fully understand variability in regime shifts.