Savanna and GIS Simulation Modeling

Addressing Management Questions for Ngorongoro Conservation Area Using Savanna, an Integrated GIS-Simulation Modeling System Randall B. Boone¹, Michael B. Coughenour¹, James E. Ellis¹, and Kathleen A. Galvin^1,2 ¹Natural Resource Ecology Laboratory and ²Department of Anthropology Colorado State University, Fort Collins, Colorado, USA
Introduction Ngorongoro Conservation Area (NCA), in northeastern Tanzania, was created explicitly as a multiple-use area, with both wildlife conservation and human welfare to be considered. The NCA is home to more than 50,000 Maasai pastoralists and their livestock, as well as world-renowned wildlife populations and archeological sites. Large-scale changes in NCA have occurred in the last 25 years, with wildebeest numbers tripling to about 1.5 million at their peak, agriculture outlawed, then reinstated, and increasing Maasai populations becoming more sedentary. Land managers require improved tools to assess management options on NCA. We used Savanna, a loosely linked GIS and ecological modeling system to model the NCA ecosystem. We addressed six types of management questions using the system. Methods Development of the Savanna Modeling System began more than 15 years ago (Coughenour et al. 1985), and has now been applied at sites throughout the world. Savanna is a series of interconnected FORTRAN subprograms joined through a common set of GIS data layers. Savanna is spatially explicit, using a girded system of cells to represent the landscape, with each cell further divided horizontally into vegetation types and vertically into layers of vegetation and soil. Several GIS layers are used as input into Savanna, and the model outputs GIS layers as well. These GIS layers were generalized to 1 km x 1 km cells, and used in modeling. The model was parameterized to describe responses of seven plant groups (e.g., grasses, forbs, evergreen trees) and eight animal groups (e.g., cattle, wildebeest, goats and sheep, giraffe). Model output was then assessed to the degree possible. A model run was conducted based upon recent conditions for use in comparisons, for the period 1973 to 1988. Then a series of six types of management questions were addressed using Savanna. Three of these types of questions were addressed simply by modifying the GIS data used in Savanna, and we present a summary of these analyses here. They included: - Assessing the ecosystem effects of a two-year long drought, - Increasing access to grazing areas by livestock, and - Adding water sources. These analyses required modifying climate data, maps controlling access to grazing, and distance to water maps, respectively. Example results are shown.

Results Effects of Drought Drought was emulated by reducing rainfall by 50% in 1983 and 1984, relative to the control model. When modeled, annual net primary productivity, herbaceous biomass, livestock populations, and wildlife populations were reduced during the dry period. Reduced livestock populations are shown recovering, in a density dependent response.

Grazing in Ngorongoro Crater Maasai may not graze livestock in Ngorongoro Crater through legal restrictions. This constraint was incorporated into Savanna using a map limiting distributions, and modified to allow grazing in Ngorongoro Crater. Allowing livestock to graze in Ngorongoro Crater added a few thousand cattle to the population. During dry years, cattle were able to maintain body condition better than when excluded from the crater.

Grazing with Improved Security Modifications were made to the maps described to allow livestock to graze in the southwest, where they are currently excluded because of the likelihood of theft. Several thousand more cattle could be supported if security in the south was improved. However, the number of elephants dropped by more than 13%.

Adding Water Sources Distance to water maps were modified to incorporate new permanent water sources. More cattle could be supported in the system with improved water supply, but these additional cattle competed for forage with goats and sheep, reducing their populations.

Discussion In a series of meetings held in July 1999, these results (and others) were presented to almost 100 East Africans involved in natural resource management, including managers from Ngorongoro Conservation Area. Feedback was favorable, and managers look forward to having the tool available. Favorable comments notwithstanding, the results presented were understood to be exemplary. Assessments and comments we received spawned the following improvements: 1) increase plant biomass estimates to match recorded data, 2) further divide animal groups to improve realism, and 3) include density dependent effects from disease when modeling animal populations, rather than model at carrying capacity. These modifications are ongoing.
References: Human population numbers from 1954 to 1999 were from: Ngorongoro Conservation Area Authority. 1999. 1998 Aerial boma count, 1999 people and livestock census, and human population trend between 1954 and 1999 in the NCA. Research and Planning Unit, Ngorongoro Crater, Tanzania. Unpublished report. Coughenour, M.B., J.E. Ellis, D.M. Swift, D.L. Coppock, K. Galvin, J.T. McCabe, and T.C. Hart. 1985. Energy extraction and use in a nomadic pastoral population. Science 230:619-625.