, Volume 125, Issue 2, pp 169–183

Vegetation change in semiarid communities

Simulating probabilities and time scales
  • T. Wiegand
  • S. J. Milton

DOI: 10.1007/BF00044649

Cite this article as:
Wiegand, T. & Milton, S.J. Vegetatio (1996) 125: 169. doi:10.1007/BF00044649


In arid regions, the effects of grazing or sparing management on natural communities of long-lived plants generally take decades to become evident. Event-driven dynamic behavior, unpredictable and low rainfall and complicated interactions between species make it difficult to assess probabilities and time scales of vegetation change.

To gain a better understanding of the main processes and mechanisms involved in vegetation change, we have developed a spatially explicit individual based model that simulates changes in plant communities over long time spans. The model, based on life-history attributes of the five dominant component plant species of a typical Karoo shrub community, follows the fate of each individual plant within the community, the sum of which is community dynamics. The model explores the differential effects of a realistic range of rainfall pattern on the abilities of these species to compete, survive, grow and reproduce.

The specific aim of the model is to identify key processes of vegetation change and to calculate probabilities and timespans for transitions between different vegetation states. Such knowledge is needed for species conservation and sustained animal production.

We show that the time-scale for changes of the dynamic state of the system are long compared with human lifespans. Employing the full range of possible rainfall scenarios showed that short-term community dynamics (years to decades) and species composition depend strongly on the short-term (years) sequence of rainfall events. In all simulation experiments the final vegetation state varied by more than 37% after a 60 year simulation period. Simulating resting of an overgrazed part of the shrub community indicated that little improvement in rangland condition was likely during a period of 60 years. Even such active management, as (simulated) clearing of unpalatable shrubs, resulted in only a 66% probability that degraded shrubland would be in good condition after 60 years resting. Simulated overgrazing of a rangeland in good initial condition only became obvious 40 or 50 years after the initiation of heavy grazing, and after 70 years the mean vegetation state eventually reached that of an overgrazed rangeland.

Key words

Event-driven dynamicsGrid based modelGrazing, Individual-based simulation modelKaroo

Copyright information

© Kluwer Academic Publishers 1996

Authors and Affiliations

  • T. Wiegand
    • 1
  • S. J. Milton
    • 2
  1. 1.Department of Ecological ModellingUFZ-Centre of Environmental ResearchLeipzigGermany
  2. 2.FitzPatrick InstituteUniversity of Cape TownRondeboschSouth Africa