Satellite Remote Sensing: Remote Sensing Data for Biodiversity
Management in Sahelian, Africa
By Erik Prins

Introduction
By the use of low as well as high resolution satellite data it is possible to map and model natural resources, resource utilization, wildlife habitats and land degradation processes in two semi-arid West African fauna reserves. The satellite data serve as an essential background for understanding the resource utilization and wildlife habitats in Sahelian Senegal. Only by using satellite images, could this information be collected in an economic feasible way. Managing biodiversity in the African Sahelian is essentially related to management of natural resources in order to improve livelihood for both human beings and wildlife in a sustainable manner. The land use system in these areas is characterized by semi-nomads who seasonally drag their livestock up to the edge of the Sahara (Sahel) to profit from the rich green vegetation that follows the seasonal rain.
     Over the past decades, a growing number of humans and livestock have utilized these areas supported by increased water availability from boreholes and wells. The consequence has been a "spotwise" degradation of the vegetation which not only has led to land degradation and desertification, but also loss of habitats for larger wildlife animals in the area. The result is that nearly all wildlife, except birds, has vanished or is extinct. The satellite data have provided basis for mapping distribution of landcovers, human activities and land degradation processes. The data have been used for various GIS analyzes of resource utilization and biodiversity modeling which together are used for decision support for biodiversity management strategies in two fauna reserves.

The Project Context
The project on Integrated Management of Biological Diversity in Sahelian Senegal, situated in Ferlo in Northern Senegal, covers two fauna reserves of an area of approximately 11.000 km2. The area is flat with vegetation dominated by woody savanna in south, and grassland in the north, transected by rich wadis often located in fossil river valleys. Population density is very low and the utilization of the area is primarily pastoral.
     The aim is to develop general biodiversity management measures with respect to the dependence of local communities on their utilization of natural resources. Furthermore, the project will establish special conservation areas and/or improve the management of specific areas already targeted for natural resources management.
     The project is funded by the Danish ODA, DANIDA, and the recipient in Senegal is the National Department for Natural Parks. Technical assistance is being provided by Ornis Consult ltd, Denmark. Other partners in Senegal include: IUCN (The World Conservation Union); CSE (Centre de Suivi Ecologique) and Direction des Eaux, For�ts, Chasses et de la Conservation des Sols (DEFCCS).
     The project started in late 1995 with a two year preparatory phase concentrating on collection of background information to point out core-areas and to start demonstration projects. As a decision tool for natural resource management, a GIS has been established. Apart from socio-economic data (which have been collected by the use of Participatory Rural Appraisals and questionaries), a large part of the background information has been collected through the use of satellite data which were integrated into the GIS. The combined use of Landsat TM and NOAA AVHRR has made it possible to collect information on natural resources, including land cover and use, land degradation and biodiversity assessment.

Landsat TM to Map Land Cover and Use
The Landsat TM data have been used to classify the study area into 14 land cover classes of which 12 are vegetation classes. In addition to the supervised land cover mapping, wadies, visible tracks, including wheel, animal tracks and firebreaks, have been digitized from the Landsat TM image.
     The land cover/use map is used for is to analyze the impact of settlement and boreholes on the surrounding environment. By using distance analysis of settlements and boreholes, the effects of human land use on specific land covers have been determined. The analysis revealed some interesting observations: the most dense vegetation class, woody savanna, is 250 percent higher within a 1 km radius than the average reserve distribution. This high presence of larger trees has by other satellite studies in Sahel been linked with some kind of a management practice of this vegetation type close to the settlements. In the Ferlo area, the connection between settlements and the high presence of trees seems more to be linked to the vicinity to woody wadies. Bush and dense bush savanna are below half the normal distribution, probably due to pastoral utilization. Furthermore, degraded bushland, bush steppe on slopes and sandy bush steppe are more than double the normal density - indicating the consequence of the heavy pastoral resource utilization for the bush savanna. The general picture indicates a general resource pressure and land degradation close to the villages (at least in a 3 km radius).
     During the dry season, livestock access to water is highly dependent on boreholes. This is especially clear in the dryer, northern part of the reserve area. A distance analysis of land covers shows that degradation is still evident 15 km away; particularly dense bush land seems affected. The analysis indicates that soil degradation, where long term effects are traumatic to the ecosystem, is present at a distance of 3-5 km from the boreholes. Sandy bush steppe is 10 times more current and bare soil occupies approximately an area five times larger than normal.

Time Series NOAA AVHRR to Study Land Degradation and Bush Fires
Former, recent and ongoing land degradation has been detected by a combination of low and high resolution satellite data. NOAA AVHRR LAC (1 km resolution) data have been used to determine land degradation in the area. Based upon daily data between 1987-93, deviation from normal patterns have been calculated leaving a general picture of a decrease in biomass production over the seven years studied. In order to suppress effects of climatic fluctuations such as El Ni–o/Southern Oscillation, only areas with more than 10 percent of average decrease in plant production have been taken into account. The multi-temporal NOAA AVHRR data, indicating areas with more than 10 percent plant production, combined with land covers that indicate land degradation from the Landsat TM data, indicate which areas have been degraded and are under further degradation, and which are 'healthy' areas under degradation. When adding villages, tracks and boreholes on top of the degradation map, reasons for degradation, including an impression of where the livestock are degrading the landscape, are visible. Figure 2 is a combination of areas with 10-25 percent decrease (blue) in average plant productivity (NDVI). The map shows co-occurrence between the two forms of degradation (yellow); characteristics of these areas are the vicinity to certain wells.

Landsat TM to Map Landscape and Bird Diversity
Due to the very low numbers of larger mammals, which primarily have been crowded out by the browsing of livestock, biodiversity assessment has been based upon bird data. Compared to other biological data, there are certain advantages in using avian data for biodiversity assessment in Ferlo: birds occur in all habitats and are numerous.
     The biodiversity assessment method was previously developed by Ornis Consult ltd. during a three-year research project mapping biodiversity in Sahel with the use of two avian data sets: birds and satellites. The first step was to produce a landscape diversity map by applying a Shannon diversity index (using diversity index in IDRISI with a 5 x 5 template) to the Landsat TM classified land cover map. Best correlations were achieved with 200-400 m pixel size. Secondly, landscape diversity has been observed on the basis of average biomass production measured from integrated NOAA NDVI data. This method has been tested by correlating the digital scores with bird counting from the transect walking (transect walking where georegistered by GPS). The observation revealed high correlation with avian diversity and/or number of bird species. It was determined that areas around wadis have the highest score and pediplanes; corresponding, dryer areas and occurrences of soils with iron panes have the lowest.
     More data will be added to the GIS during the preparation phase to complete the final data collection for decision support.

The Future
Fires are another degrading factor in the Sahelian eco-system. The Ferlo Reserves have several means to combat the spread of these fires. Bushfire data will be used to evaluate the effects of firebreaks and to detect if exposed areas are connected to some of the degraded areas. Information on historical bush fires are for the moment extracted from NOAA AVHRR by various research institutions in Europe. At the present stage, one annual recording of fire occurrence has been added to the system.
     Furthermore, modeling of biodiversity will take place for some large endemic animals. An example is the ostrich habitat modeling. Ostriches are occurring in low numbers in the northern reserve. It is an option to manage the prime habitat area or set-out raised individuals. Due to the fairly good habitat description of this type of ostrich, it has been possible to model their core habitat areas. The validity of the model is currently under evaluation by plotting observations of the ostrich in the reserve into the GIS. So far, the only known nesting area has occurred in one very suitable area.

About the Author:
Erik Prins is natural resource manager who specializes in remote sensing and GIS applications. He has more than 10 years experience in digital remote sensing in tropical environments, and can be reached through Ornis Consult.

Back