Résumé
The role of mountain evolution on local climate is poorly understood and potentially underestimated in climate models. One prominent example is East Africa, which underwent major geodynamic changes with the onset of the East African Rift System (EARS) more than 250 Myr ago. This study explores, at the regional East African scale, a molecular approach for terrestrially-based paleo-climatic reconstructions that takes into account both changes in temperature and in altitude, potentially leading to an improved concept in paleo-climatic reconstructions. Using surface soils collected along pronounced altitudinal gradients in Mt. Rungwe (n = 40; Southwest Tanzania) and Mt. Kenya (n = 20; Central Kenya), we investigate the combination of 2 terrestrial proxies, leaf wax n-alkane delta H-2 (delta H-2(wax)) and branched glycerol dialkyl glycerol tetraether (br GDGT) membrane lipids, as (paleo) elevation and (paleo) temperature proxies, respectively. At the mountain scale, a weak link between delta H-2(wax) and altitude (R-2 = 0.33) is observed at Mt. Kenya, but no relationship is observed at Mt. Rungwe. It is likely that additional parameters, such as decreasing relative humidity (RH) or vegetation changes with altitude, are outcompeting the expected H-2-depletion trend along Mt. Rungwe. In contrast, br GDGT-derived absolute mean annual air temperature (MAAT) and temperature lapse rate (0.65 degrees C/100 m) for both mountains are in good agreement with direct field measurements, further supporting the robustness of this molecular proxy for (paleo) temperature reconstructions. At the regional scale, estimated and observed d 2 H data in precipitation along 3 mountains in East Africa (Mts. Rungwe, Kenya and Kilimanjaro) highlight a strong spatial heterogeneity, preventing the establishment of a regional based calibration of delta H-2(wax) for paeloaltitudinal reconstructions. Different from that, an improved regional soil calibration is developed between br GDGT distribution and MAAT by combining the data from this study (Mts. Rungwe and Kenya) with previous results from East African surface soils along Mts. Kilimanjaro (Tanzania) and Rwenzori (Uganda). This new regional calibration, based on 105 samples, improves both the R-2 (0.77) and RMSE (root mean square error; 2.4 degrees C) of br GDGT-derived MAAT over the global soil calibrations previously established (R-2 = 0.56; RMSE = 4.2 degrees C) and leads to more accurate (paleo) temperature reconstructions in the region. (C) 2016 Elsevier Ltd. All rights reserved.