Reconstructing Past Environments: Paleoecological Transfer Functions Deborah Balch University of Arizona Humans have long recognized the relationship between fossil organisms and past environments. For example, in Ancient Greece, Xenophanes and Herodotus found marine fossils far inland and deduced that the area must have been an ocean in previous times (Imbrie and Newell, 1964). Paleoecologists use this same concept as the basis for their scientific research. They use the fossil record to reconstruct past environments and climates. One way this is accomplished is by studying modern assemblages and their environmental preferences and correlating this with their fossil assemblages. As an example, imagine taxa assemblage “X” is always found in environmental conditions termed “Y”. If we find this same assemblage in the fossil record, we can deduce that the environmental conditions at that time in the past were similar to “Y”. This concept has been quantified in a method called paleoecological transfer functions. The first step to building a transfer function is to develop a training set. Training sets are composed of two sets of data: measurements of the environmental parameter (i.e. temperature, pH, salinity, etc.) and abundance of its associated taxa assemblage. These data are then developed into a set of equations (called transfer functions) that correlate the biological and environmental data. These equations often take the form: Tm=XFm and Tp=XFp T is the estimated environmental parameter (m=modern, p=paleo), F is a matrix of modern and fossil data and X is a transfer coefficient or set of coefficients (Imbrie and Kipp, 1971; Hutson, 1977). Calibration is accomplished by solving this equation for X with a set of biological data (F) and observed values of the environmental parameter (T). Once calibrated, the transfer function is applied to the fossil assemblage and the paleoenvironmental parameters can be estimated. There are several different computational techniques available to reconstruct past environments through the use of transfer functions (refer to Sachs et.al, 1977 for a comprehensive list). This summer, my research focused on collecting data for a training set which will be used to reconstruct the past environments of Lake Malawi using modern ostracode assemblages. From July 8 to August 3rd 2001, I collected data from various small lakes and ponds in areas around Kigoma and Sibwesa, Tanzania (Table 1). In the field, I measured water temperature, dissolved oxygen, pH, conductivity, turbidity and depth. Water and surface sediment samples were taken back to the lab for further analysis. Water samples were analyzed for total phosphorus, PO4-P, NH4, NO3-N, NO2, Alkalinity and SiO2. The surface sediments were analyzed for % organic carbon and ostracode abundances. Please refer to Table 2 for all results (except ostracode counts, which remain to be analyzed). It is very important that a training set is entirely representative of the fossil assemblage and that no-analog conditions are avoided (Huston, 1977). Therefore, these data are just the beginning component to a large training set that will incorporate samples from many locations in Kenya, Tanzania, and Malawi. The transfer function, once computed and calibrated, will be applied to fossil assemblages found in long cores taken from Lake Malawi (drilling is projected for February 2003). After all of the analyses are complete, it is our goal to obtain a profound understanding of the Quaternary environments of Lake Malawi and to use this information to predict how the lake may be changing due to recent anthropogenic activities. Table 1: Sites sampled in Summer 2001 Site Site 1 Name Ujiji Beach Latitude o 4 55.189s o Longitdue Date Time Weather o 8-Jul-01 4:30pm clear sky, slight breeze o 8-Jul-01 20-Jul01 20-Jul01 20-Jul01 20-Jul01 27-Jul01 27-Jul01 27-Jul01 27-Jul01 3-Aug01 3-Aug01 3-Aug01 3-Aug01 3-Aug01 4:55pm clear sky, slight breeze 10:08am hazy, slight breeze 10:29am hazy, slight breeze 1:10pm sunny, hot with breeze 1:52pm sunny, hot with breeze 10:37am cloudy, slight breeze 12:13pm cloudy 4:05pm partially cloudy 5:13pm partially cloudy 10:08am sunny, very windy 10:40am sunny, wind was blocked from vegetation 12:00pm partially cloudy, wind 12:41pm partially cloudy, wind 3:10pm sunny, slight wind 29 40.393e Site 2 Ujiji Beach, 100m south of site 1 4 55.189s 29 40.420e Site 3 Katobelo Lake, Sibwesa 6o29.251s 29o57.519e Site 4 Katobelo Lake, 200m SW of site 3 6o29.307s 29o57.741e Site 5 Kabogo Lake, Sibwesa 6o28.968s 29o58.830e Site 6 Kabogo Lake,300m south of site 5 6o29.033s 29o59.005e Site 7 Nyakubeleka Lake, near Burega 4o54.525s 29o38.938e Site 8 Katosho Lake, near Butanga 4o50.930s 29o39.349e Site 9 Site 10 Site 11 Site 12 Site 13 Site 14 Site 15 o o Kikamba Lake 4 54.473s 29 37.109e small empheral pond east of airport 4o53.273s 29o41.879e Nyakubeleka Lake, deeper sample 4o54.567s 29o38.971e Nyakubeleka Lake, offshore sample 4o54.536s 29o38.934e o o Katosho Lake, deeper sample 4 50.901s 29 39.357e Katosho Lake, other side of lake 4o50.997s 29o39.526e Kikamba Lake, deeper sample N/A N/A References Hutson, W.H., 1977. Transfer functions under no-analog conditions: experiments with Indian Ocean planktonic foraminifera. Quaternary Research 8:355-367. Imbrie, J. and N.G. Kipp. 1971. A new micropalaeontological method for quantitative Paleoclimatology: application to late Pleistocene Carribean core V28-238. In: The Late Cenozoic Glacial Ages (K.K. Turkian ed.) Yale University Press, New Haven:77-181. Imbrie, J. and N. Newell. 1964. Approaches to Paleoecology. John Wiley and Sons Inc., New York. 432p. Sachs, H.M., T. Webb III and D.R. Clark., 1977. Paleoecological transfer functions. Annual Review of Earth and Planetary Sciences 5:159-178.