Paleotemperature reconstruction in tropical Africa using fossil Chironomidae (Insecta: Diptera)

Fossil assemblages of chironomid larvae (non-biting midges) preserved in lake sediments are well-established paleothermometers in north-temperate and boreal regions, but their potential for temperature reconstruction in tropical regions has never before been assessed. In this study, we surveyed subfossil chironomid assemblages in the surface sediments of 65 lakes and permanent pools in southwestern Uganda (including the Rwenzori Mountains) and central and southern Kenya (including Mount Kenya) to document the modern distribution of African chironomid communities along the regional temperature gradient covered by lakes situated between 489 and 4,575 m above sea level (a.s.l). We then combined these faunal data with linked Surface-Water Temperature (SWTemp: range 2.1–28.1 C) and Mean Annual Air Temperature (MATemp: range 1.1–24.9 C) data to develop inference models for quantitative paleotemperature reconstruction. Here we compare and discuss the performance of models based on different numerical techniques [weighted-averaging (WA), weighted-averaging partial-least-squares (WAPLS) and a weighted modern analogue technique (WMAT)], and on subsets of lakes with varying gradient lengths of temperature and other environmental variables. All inference models calibrated against MATemp have a high coefficient of determination (r jack = 0.81–0.97), low maximum bias (0.84– 2.59 C), and low root-mean-squared error of prediction (RMSEP = 0.61–1.50 C). The statistical power of SWTemp models is generally weaker (r jack = 0.77– 0.95; maximum bias 1.55–3.73 C; RMSEP = 1.39– 1.98 C), likely because the surface-water temperature data are spot measurements failing to catch significant daily and seasonal variation. Models based on calibration over the full temperature gradient suffer slightly from the limited number of study sites at intermediate elevation (2,000–3,000 m), and from the presence of morphologically indistinguishable but ecologically distinct taxa. Calibration confined to high-elevation sites ([3,000 m) has poorer error statistics, but is less susceptible to biogeographical Electronic supplementary material The online version of this article (doi:10.1007/s10933-009-9339-2) contains supplementary material, which is available to authorized users. H. Eggermont (&) L. Audenaert D. Verschuren Limnology Unit, Ghent University, K.L. Ledeganckstraat 35, 9000 Ghent, Belgium e-mail: [email protected] O. Heiri Palaeoecology, Laboratory of Palaeobotany and Palynology, Utrecht University, Budapestlaan 4, 3584 CD Utrecht, The Netherlands J. Russell Geological Sciences, Brown University, Box 1846, Providence, RI 02912, USA M. Vuille Department of Earth and Atmospheric Sciences, University at Albany, State University of New York, 1400 Washington Ave, Albany, NY 12222, USA 123 J Paleolimnol (2010) 43:413–435 DOI 10.1007/s10933-009-9339-2