Spatiotemporal variability in landscape function in a semi-arid rangeland in Kenya

Background/Question/Methods

Drylands cover about 40% of the earth’s surface.  Loss of ecosystem function in these water-limited landscapes, as a result of anthropogenic pressures and climate change, poses a profound challenge for pastoralist societies, particularly in less developed regions where few livelihood alternatives exist.  Understanding and managing the trajectories of change requires knowledge of, and empirical assessment of, the water-soil-vegetation feedbacks that govern ecosystem function and productivity.  In order to evaluate the spatial and temporal coupling among changes in different ecosystem functions, and to derive vegetation transition probabilities and appropriate length scales for a spatially explicit model of landscape dynamics, we used two protocols to monitor changes in spatial structure of vegetation and soil surface condition on 26 permanent 50m line transects in a degraded Kenyan rangeland over 12 rainfall pulse-driven growing seasons (in 4 years).  The final monitoring was conducted after the end of a severe drought.

Results/Conclusions

The protocol for ecosystem function analysis based on soil surface condition generated indices for soil stability, infiltration capacity, nutrient cycling, and indices of spatial patterning of vegetation cover, for each transect at each monitoring interval.  The means of all indices varied significantly from season to season. Interestingly, each index showed distinctly different patterns of change through time. The severe drought had differential impacts on indices; soil infiltration index was significantly lower than the 4-year average, while vegetation indices were not.  Fine-scale vegetation monitoring revealed that both annual and perennial herbaceous species abundance were highly labile through time, and the degree of seasonal variability was not predicted by 4-year mean vegetation cover.  Herbaceous vegetation associated with trees and shrubs, however, was less labile than open herbaceous vegetation.  The differences in herbaceous change probabilities associated with woody vegetation will be used to refine the parameterization of stochastic transition probabilities in a dynamic model developed to evaluate the impacts of different climate, local rainfall, and land use scenarios on landscape function and productivity.