Introduction

In recent years Africa has been plagued with wars, famine, deforestation, desertification, overpopulation, and general resource depletion. There is no question that Africa is in crisis (Mortimore, 1989). Just how severe is that crisis remains the question facing the world development community as represented by such diverse groups as the United Nations, World Bank, bilateral donors, international non-governmental organizations (NGO's), and of course the countries themselves.

This "crisis" has been particularly apparent in the Sahel. Droughts and famine have long been a factor of life in this region, but mechanisms were previously in place to withstand them. Now, as population pressures, international economies, and mechanized agriculture have become more prominent, this is less the case. These factors have forced sub-Saharan Africans into land use practices that are often inappropriate for the land (Toure, 1989; Franke and Chasin, 1980; Freudenberger and Freudenberger, 1993).

It has been suggested that the increasing rate of land use/land cover change is one the of the most important ecological issues in Africa (de Graf, 1993). Because the majority of African countries have rural economies and depend heavily upon their natural resources (e.g. for food, fuelwood, commodity exports), degradation of those resources can result in rapid declines in standards of living. Senegal, in West Africa, is an example of such a country. Buffeted by drought, currency devaluation, and rapid population growth, Senegal's ecosystems are coming under increasing pressure. There is evidence of rapid changes in the landscape, which in many cases is threatening local livelihoods (Tappan and Wood, 1995).

It is also clear that degradation of natural resources in Africa cannot be treated as solely a biophysical problem (Kates, 1990; Blaikie, 1985; Bryant and LeDrew, 1989; Jacobson,1990). For researchers in Africa, as globally, separating the human induced from the climatically induced change is a major challenge. Determining the socioeconomic factors responsible for the human induced landscape change is even more complicated and has been an evolving focus of governments and donors throughout the continent.

Objectives

The major goal of this research is to develop a methodology in the form of descriptive and predictive models, which will facilitate the determination of factors driving land use/land cover change in southern Senegal. In the process of accomplishing this, we hope to determine the most effective method(s) for modeling those changes. "Effectiveness" in this context consists of two key requirements : (1) that the models are technically straightforward enough to be applied by interested government agencies and (2) that they are accurate and flexible enough to be of use in providing realistic simulations for managers and policy makers.

The three objectives of this research are :

1. To determine the most accurate approach(es) for modeling land use/land cover change in southern Senegal. Several modeling approaches, built on the foundations of Markov processes and logistic regression and modified through implementation of rule based techniques, will be used to characterize and predict land use/land cover. Accuracy will be determined statistically based on each model's ability to characterize existing and simulate future land use/land cover. An iterative approach will be used in order develop a final model that is technically straightforward while still maintaining as high a degree of accuracy as is possible from the models being developed.

2. To determine which factors are driving land use/land cover change in southern Senegal. In order to accomplish this, land use /land cover classification and change techniques will be carried out using a historical time series of Landsat MSS and TM and ancillary data. Models will be developed to characterize land use change, as described above. From these models the significance of the contribution of each change factor will be determined.

3. To develop realistic simulations from predictive models for managers or policy makers. Upon development of the final predictive model as described above, a methodology for creating "what-if" simulations will be developed that will allow for hypothetical changes in key change factors in order to replicate the effects of a range of policy or management decisions, exogenous factors, or natural influences.

This modeling effort has the potential to assist natural resource managers, policy makers, and the scientific community in Senegal. To date there has been no extensive inventory of land use/land cover change in Senegal, let alone a systematic investigation of causal relationships or what changes are likely to occur in the future. By providing the decision making community with a technically feasible modeling approach, the opportunity exists to correct this situation. Techniques developed here may also contribute to the land use/land cover modeling knowledge base in general.

As a result of extensive field work in the study area and investigation of relevant literature and datasets, a number of research questions related to land use have surfaced. The modeling process will attempt to "answer" many of these. The following are examples of a few key questions with others becoming apparent as the research continues.

* Does the rate of population growth correspond to the rate of expansion of land into agriculture? How does the rate of population growth effect the fallow system?

* Is agricultural expansion driven more by proximity to the village or existing fields, or by appropriate soil type? To what extent is agriculture expanding into inappropriate soil types?

* Is the land use/land cover change profile in the two northern departments of the study area (Kaffrine and Tambacounda) being reflected in the two departments in the Casamance portion of the study area (Kolda and Velingara)? How is the land use/land cover change profile effected by the stage the department is in as a charcoal source, i.e. in the pre-extraction, extraction, or post-extraction stage?

* Is expansion along major roads occurring despite inappropriate soil types? Along minor roads in inappropriate soil types? At what rates?

* Is the Niokola Koba National Park withstanding edge intrusion in areas suitable for agriculture? Does the rate of change adjacent to the park vary from the norm, i.e. is there any self-generating buffer effect? Is the same true for reserves of a different status, i.e. foret classe, sylvo-pastoral or some other non-park status?

* Does the species richness and abundance (biodiversity components) of a non-reserve forest, including the presence of "noble species", effect the rate of land use/land cover change?

* Is the previous (at time t₀ ) land use/land cover state significant in determining the transition state (at t₁)?

* What are the spatial factors (e.g. contiguity, patch size, edge, etc.) most influencing change ? How do they influence change?

* Are the activities of the agricultural parastatals (e.g. Sodifitex, Sodagri) accelerating the rate of change?

* What influence have key policy reforms (e.g. agriculture, forestry, land laws) had on the rates and types of land use / land cover change?

Study site

The general study area consists of four departments located in south-central Senegal (Kaffrine, Tambacounda, Kolda and Velingara). There are 30 departments in all of Senegal, the four above being of particular interest in that they have abundant natural resources, fall within the zone of viable rainfed agriculture, and as a result are undergoing significant land use/land cover change. The initial investigation will be carried out in the department of Velingara.

This department, like the other three, faces a number of critical land use issues, including the effects of charcoal and fuelwood production, cultivation of cash crops (peanuts and cotton), excessive burning, and agricultural expansion.

Data

The primary focus for the socioeconomic data collection was to determine land use management practices and general resource management efforts, but has resulted in collection of a variety of other data. These consist of site observations, intermittent field contacts, short village visits to answer specific field questions, and more in-depth structured village studies. The latter have been carried out in five ecologically distinct villages with the use of rapid rural appraisal (RRA) techniques (e.g. Freudenberger, 1996) . Also, numerous half-day village interviews have been carried out using some of these same RRA techniques, most notably the semi-structured interview.

Other socioeconomic datasets include USAID's three Knowledge, Attitudes, and Practices (KAP) studies; the results of the RRAs carried out in USAID's seven village REMAP study ; village specific RRAs by other investigators (e.g. Freudenberger and Freudenberger, 1992); and numerous project reports by the Government of Senegal and various bilateral or multilateral donors .

Modeling

A series of models of increasing complexity will be developed in order to determine the range of accuracy that can be obtained. In each case the model will be calibrated and validated using the change series generated from the classification of the 1973, 1978, 1985 and 1990 Landsat MSS scenes for the department of Velingara. The models will then be used to predict the land use/land cover in 1995 from 1990 or previous data and subsequent accuracy determined. The best performing models will then be applied to another department (within the general study area - refer to Section 4.1) in order to determine their general robustness. The choice of 1995 is based on the approximate 5 year time interval between historical images

The four general categories of models to be investigated consist of :

Simple Markov
Markov model with spatial modifier
Logistic regression
Knowledge based system

Results

It is anticipated that the results of this research will include:

1. Spatially explicit biophysically and socioeconomically driven models developed to describe existing landscape change . Output will include maps, GIS coverages, and tabular datasets explicitly designating the spatial extent of both actual and modeled changes in land cover.

2. Comparative analysis of the above models in terms of descriptive and predictive accuracy.

3. A qualitative assessment of the data processing and development expenditures required for implementation of each model.

4. A design for implementation of the above models as a tools for applying hypothetical changes in key change factors in order to replicate the effects of a range of policy or management decisions, exogenous factors, or natural influences.

5. In depth discussion of the relationship between biophysical and socioeconomic factors and land use/land cover change; based both on statistical evidence as well as, on convergence of evidence (i.e. accommodates RRA, etc., that may not be able to be captured in a model of types developed).

Blaikie, P. (1985). The Political Economy of Soil Erosion in Developing Countries. New York: Longman Scientific & Technical.

Bryant, R.B., and E.F. LeDrew (1989). The Scope of Land Change Research. In R. B. Bryant,E.F. LeDrew,C. Marios,and F. Cavayas (Eds.), Remote Sensing Methodologies of Land Use Change Analysis. Waterloo, Ontario: University of Waterloo.

de Graff, J. (1993). Soil Conservation and Sustainable Land Use. Amsterdam: Royal Tropical Institute.

Franke, R.W., and B.H. Chasin (1980). Seeds of Famine: Ecological Destruction and the Development Dilemma in the West African Sahel. New York: Universe Books.

Freudenberger, K.S., and M.S. Freudenberger (1992). Fields Fallow, and Flexibility : Natural Resource Management in Ndam Mor Fademba, Senegal. (RRA Report ). World Vision.

Freudenberger, K.S. (1996). Kandia: Threatened Forest, Threatened Livelihood. Report, USAID/Dakar. (In press).

Jacobson, H.K., and M.F. Price (1990). A Framework for Research on the Human Dimensions of Global Environmental Change. ISSC Standing Committee on the Human Dimensions of Global Change.

Kates, R.W., W.C. Clark, V. Norberg-Bohm, and B.L. Turner II (1990). Human Sources of Global Change : Priority Research Initiatives for 1990-1995. Providence: Institute for International Studies.

Mortimore, M. (1989). Adapting to Drought : Farmers, Famines and Desertification in West Africa. Cambridge: Cambridge University Press.

Tappan, G.G. and E.C.Wood (1995).Structure pour le suivi a Long Terme des Ressources Naturelles du Senegal. Rapport Final.Colloque National sur La Gestion Des Ressources Naturelles Au Senegal.

Toure, O. (1991). Where Herders Don't Herd Anymore: Experience from the Ferlo, Northern Senegal. (Internal Report).Dakar, Senegal: Centre de Suivi Ecologique.