The Baringo Paleontological Research Project, directed by Andrew Hill since 1985, is an ongoing geological and anthropological study in the Tugen Hills/Lake Baringo region of Kenya. I am interested in paleobiogeography and how the range of variation among environments of the Pliocene may determine the presence or absence of hominins. Time averaging is an inevitable part of paleontological work and it is partly due to this that subtleties in ecological variation are overlooked but perhaps more detailed models can help to tease apart past complexities. Eventually a wide geographic range in Africa will be used in the analysis but for the moment the focus is on the Baringo region. Faunal material has traditionally been utilized to study paleoecology but many ecological tools can be used to then extrapolate about the past. The main tool I plan to use in ecological analysis with application to paleoecology is remote sensing to analyze differences in vegetation in relation to lake level/area. Remote sensing can be utilized to determine what types of changes occur with variations in water, such as area of a wetland, and can also be used to determine the paleolake margin (Le Blanc et al, 2006; LeTurdu, et al 1999; Schuster et al, 2005). Satellite images and their manipulation can be important tools particularly when coupled with lake core data and faunal material (Eriksson et al, 1999; Alin and Cohen, 2003; Cohen et al, 2005; Vincens et al, 2005).
Background - Theoretical
A number of paleoecological studies detail the geology of a lake bed or river at the hominin site of interest (Hay, 1981; Sikes et al, 1999; Walker, 2002). This information is important for taphonomic reasons but otherwise lakes are relegated to the simple role of water source for paleoecological purposes. This does not have to be the case and understanding the role of the lake in an ecosystem can provide a means for asking and answering questions about hominin evolution. Lake level is controlled by a number of factors including rainfall, evaporation, inlets, outlets, and the location of rainfall maximum (Nicholson, 1999). Nicholson’s (1999) study of Lakes Tanganyika and Rukwa in Tanzania indicate that lake level is correlated with rainfall on a yearly basis and strongly correlated when the previous five years of rainfall are taken into account. This is particularly pertinent to paleontological studies where time averaging is a factor. A relatively higher amount of rainfall during a period would be reflected in a relatively higher lake level than in another period with a lower amount of rainfall. Nicholson found that variation in the levels of both lakes is explained by rainfall, but there are also differences between the two lakes’ responses to that variability because Lake Rukwa is a closed basin. During exceptionally dry periods large portions of Lake Rukwa completely dried up taking on the appearance of two distinct lakes each with different chemical compositions and a wetland in between. This local variation indicates that studies of paleobiogeography are necessary to discover the range of habitats hominins could live in rather than trying to fit their evolution into a particular type of habitat.
Understanding the unique relationship of each individual lake to rainfall is necessary in establishing a water balance model for every lake before interpreting paleo-fluctuations. Lake Baringo has a number of inlets (Molo, Ol Arabel, Mukutan and Ndau Rivers) and no known outlet, meaning that the water of Lake Baringo is the sum of rainfall, inlets and evapotransporation (Thom and Martin, 1983; Barton et al, 1987). The water level of Lake Baringo serves as a good proxy for rainfall. Both methods, using lake level to infer rainfall or rainfall to infer lake level have been established (Nicholson 1999, 2000; Yin and Nicholson, 2002). The amount of rainfall in any given time period be it year, century or more will never be available to paleontologists so lake level will be used as a proxy for precipitation.
Changes in rainfall are reflected in the Normalized Difference Vegetation Index (NDVI) which is a measurement of the photosynthetic capability, and arguably leaf area, of vegetation (Nicholson et al, 1990). In a comparison of the West African Sahel to East Africa (Kenya and Tanzania) Nicholson found a log-linear relationship between annual rainfall and annual NDVI using monthly measurements. Rainfall is directly linked to changes in vegetation so if lake level is known and can be used as a proxy for rainfall then changes in vegetation can be predicted. As modern features of African rainfall variability are the same as the historical past, it stands to reason that they are similar to the paleontological past (Nicholson, 2000) and their impact on vegetation would also be similar.
Large scale patterns of modern and historical variation in rainfall, climate and vegetation have been examined (Nicholson et al, 1990; Nicholson 1999, 2000, 2001) but this has not yet been extended to paleontological sites. The majority of studies on lakes have examined the condition of a single lake during one time period or compared two lakes during modern and historical times (Ali and Cohen, 2005; Eriksson et al, 1999; Jury and Gwazantini, 2002; Kebede et al, 2006; Legesse et al, 2002). Lake Baringo has yet to be drilled for core analysis and no study exists on the modern and historical fluctuations or their relationship to rainfall and vegetation. Modern vegetation analyses in relation to climate are feasible because human impact in the area, even with goat farming, has a low to moderate impact on the area and is unlikely to significantly affect the tree canopy in those areas where economic reliance on goats is greatest (Burnett and Rowntree, 1990; Little 1996). The Lake Baringo area has a number of sites of paleontological importance (Hill, 1985; Jacobs and Kabuye, 1989; Hill et al, 1992; Hill, 2002; Hill et al, 2002). As lake level is a good proxy for rainfall, particularly in the case of Lake Baringo, changes in the level can be linked to rainfall and then to changes in vegetation. Changes in vegetation can shed light on biodiversity and the behavior of animals. Similar principles are expected to have been operating in the past (Nicholson, 2000) particularly at the local level. The work of the Baringo Paleontological Research Project is ongoing and research into past lake levels using historical records, core sampling and remote sensing techniques are being explored. This work coupled with work on the modern lake, rainfall and vegetation patterns can provide a different means to reconstructing the past environments surrounding the lake.
Related Research
Environmental change, and therefore paleoecological reconstruction, long thought to be an important part of understanding shifts in human evolution has often been analyzed through neoecological or geological methods (Dart, 1925; Hay, 1973; Hill, 1981; Reed, 1997; Vrba, 1985). A number of studies have focused on ecological techniques concerned with counting relative numbers of species but this method frequently ignores unique problems of taphonomy in fossil assemblages (Andrews and Evans, 1979; Pickford, 1987; Reed, 1997; Vrba, 1975). Although fossil plant material has also been utilize to reconstruct and categorize paleovegetation (Bonnefille 1987, 1994; Jacobs, 1987, 1989, 1992, 1996, 1999, 2002) the interplay between vegetation and other environmental factors is largely ignored. In particular there are no analyses of the differences between vegetation in lake margin or riverine gallery forests settings (i.e. the extent of the area or differences in plant and animal species occupying those environments) and no clear idea of how shifts in lake margins impact the extent of vegetation. The application of remote sensing to ecological questions pertaining to African environments may provide patterns useful to modeling past environments. This method of reconstruction, based solely on ecology, could serve as a useful check to models riddled with taphonomic problems.
The paleontological deposits of interest are in the area of the Lake Baringo and the nearby Tugen Hills. Lake Baringo is located at 0° 36’ N and 36° 04’ E, 60 km north of the equator and 975 meters above sea level (Odour et al, 2003). The Tugen Hills are west of the lake extending approximately 100 km NS along the Kenya Rift Valley (Hill, 2002). The first record of the geology of the area occurred in 1884 (Thomson) and since then the area has been of interest to geologists and paleontologists. The East African Geological Research Unit was formed in the mid-1960s and their survey of the area prompted a separate but related group headed by Bill Bishop to concentrate on the fossils of the area (Hill, 2002). In 1980 the Baringo Paleontological Research Project was formed by David Pilbeam and after five years direction of the group was given to Andrew Hill. The continued exploration and study of the area has contributed to our knowledge of the stratigraphy, radiometric and paleomagnetic age, fossil fauna, paleobotanical and isotopic work of the Baringo Basin and Africa as a whole throughout the last 16 million years. Although satellite data has been used in the area for land management, geologic and hydrologic purposes (Conant, 1982; Deatsch et al, 1985; Grimaud et al, 1994; Hautot et al, 2000; Le Turdu et al, 1995) it has yet to be used for ecological purposes and applied to paleoecology.
Elsewhere in Africa remote sensing has frequently been used for ecological purposes. In the Lake Elementeita Basin of Kenya three Landsat images from 1973, 1976 and 1984 viewed with one band in the infra-red combined with field surveys were used to calculate changes in the surrounding forested and woodland areas (Mwaura and Moore, 1991). Mwaura and Moore (1991) also comment on the impact these vegetation changes have on lake levels and their connection to rainfall and soil erosion. Changes of land cover during a twenty year period at a wildlife reserve in Kenya have been detected with the use of two satellite systems of differing spatial resolution, NOAA/AVHRR data and Landsat images (Serneels et al, 2001). It was discovered that the Landsat data, with higher resolution, better detect more subtle land cover changes although the temporal resolution is not as good (Serneels et al, 2001). Variability in vegetation and its link to rainfall has been examined in Kitui, Kenya by comparing 20 years of Landsat and NOAA/AVHRR images to meteorological statistics from the area and topographical maps. This has proven to be a very accurate method of detecting change over a large area (Tanaka et al, 2000). Satellite images have been used to ascertain levels of biodiversity through the integration of image based classification and field surveys of both flora and fauna (Fuller et al, 1998; Oindo, 2002; Oindo and Skidmore, 2002).
I plan to do a multi-step project correlating lake level to rainfall, rainfall to vegetation, and vegetation to animal distribution; then apply the modern model to the past. This will require time in the field ground truthing as well as time in the CEO Lab. My plans for this project are at the moment in the beginning stages and the first steps in moving forward include identifying the best system of classification for the Baringo area.