Simulated long-term vegetation response to grazing heterogeneity in semi-arid rangelands

1, The long-term effects of small-scale spatial heterogeneity of livestock grazing on vegetation dynamics were studied with a grid-based model of majo r life forms of savanna vegetation. Based on southern Kalahari ecology, the model includes stochastic life-history variables, resource competition for soil water, and biomass production for annuals, perennial grasses and shru bs. 2, Grazed at individual severities, the model's 25 m(2) grid cells defined the spatial scale of heterogeneity. Different scenarios of grazing heteroge neity were generated by modifying distributional and behavioural features o f the grazing model. Simulations were run over 50 years under moderate to h igh constant stocking rates. 3, Results confirmed a previously reported threshold response of shrub cove r increase: under moderate grazing pressure, little change in shrub cover o ccurred; when grazing pressure exceeded a threshold, shrub cover increased drastically. 4. Under moderate or high stocking rates, grazing heterogeneity did not mod ify grazing effects. However, within an intermediate range of stocking rate s, small-scale heterogeneity determined the long-term impact of grazing. In particular, utilization intensity at the threshold of shrub cover increase was 60% less under high compared to low local grazing heterogeneity. 5, Sensitivity of vegetation dynamics to local grazing heterogeneity was al so exemplified under a landscape-scale grazing gradient as observed at wate ring points: at a given utilization intensity, a wide zone of increased shr ub cover occurred under large local grazing heterogeneity, while under the least heterogeneous grazing only a narrow zone of slightly increased shrub cover occurred. 6. Because of the slow progress of shrub cover increase, a mismatch of mana gement and ecological time scales was diagnosed and its implications for ma nagement are discussed. 7. We conclude that knowledge of local grazing heterogeneity is crucial for correct assessment of livestock impact on vegetation dynamics, Consequentl y, management aiming at sustainable land use should account for spatial gra zing aspects. These poorly understood aspects form a gap to be filled by bo th empirical and theoretical studies.