Stratigraphic and hydrologic responses to tropical climate variability: Scientific drilling in Lake Malawi, East Africa

Lake Malawi contains a long continuous sedimentary record of climate change in the southern hemisphere African tropics. A stratigraphic framework of this basin is developed over the last ∼150 ka by integrating several vintages of seismic-reflection data with recently acquired drill cores. In the seismic-reflection data set, we document three lake-level cycles where progradational delta seismic facies and erosional-truncation surfaces mark the basal boundary of each sequence. The clinoform packages and their down-dip, time-equivalent surfaces can be mapped throughout each basin, where each major lowstand surface was followed by a transgression and highstand. On several occasions, lake level dropped as much as 500 m below present lake level (BPLL) in the North Basin and 550 m BPLL in the Central Basin. Evidence for these lake-level fluctuations in the drill cores include major changes in saturated bulk density, natural gamma ray values, and total organic carbon. During lowstands, density values doubled, while total organic carbon values dropped from ∼5% to 0.2%. Coarse-grained sediment and organic matter flux into the basin were higher during transgressions, when precipitation, runoff, sediment supply, and nutrient input were high. This sedimentation pattern is also observed in seismic-reflection profiles, where coarse-grained seismic facies occur at the bases of sequences, and in the drill-core data where the highest total organic carbon values are observed immediately above lowstand surfaces.;An energy-balanced hydrologic model is used to quantitatively assess atmosphere-water budget relationships across the Lake Malawi catchment. The model first simulates the historical lake-level record over the last 100 years using climate station and vegetation data as inputs. Atmospheric conditions required to sustain equilibrium water balance are then estimated at known critical lake levels: modern (700 m maximum water depth), basin closure (696 m maximum water depth), 500 m, 350 m, 200 m, and 150 m maximum water depth. The critical low lake stages were determined from analysis of seismic-reflection and deep lake drill-core data. The model predicts modern precipitation rate to be 955 mm/yr, which is consistent with observed climate station precipitation records. The minimum lowstand observed in geological and geophysical records is 150 m water depth (550 m below present lake level), and occurred about 95,000 years before present. The precipitation rate required to sustain equilibrium conditions at this low lake stage is 557 mm/yr, assuming modern Lake Malawi temperature and vegetation, and 374 mm/yr using modern temperature and vegetation data from the Little Karoo Basin, an analogue for the Malawi paleo-environment during severe arid intervals that resulted in major low lake stages. The latter result is consistent with the range of precipitation measured from the Little Karoo Basin (100 to 500 mm/yr), and from interpretations of drill-core data sets (Cohen et al., 2007). The time required to drop lake level from its modern maximum to the most severe low lake stage determined from paleoclimate data sets (from 700 m to 150 m maximum water depth) is less than 2500 years, even when accounting for additional water volume loss stored as groundwater. A lake-level fall of this magnitude reduces the lake surface area by 94% and reduces the total lake volume by 99%.;Geochemical and geophysical analyses from the 380-m long Drill Site 1 show consistent relationships and are incorporated into a principal component analysis, where the first principal component (PC(1)) is an enhanced signal of lake-level variability, representing 63% of the variance in all incorporated data sets. Application of age models on PC(1) show a correlation with rainy season insolation at 10°S, suggesting an orbital precession control on lake-level variability that is modulated by eccentricity. When eccentricity is high, precession is the dominant control on lake-level variability. When eccentricity is low, lake levels stabilize at highstand conditions. Evolutionary wavelet spectra of PC(1) and cross-wavelet transforms of PC(1) with rainy season insolation shows a similar result, where intervals of high eccentricity show a strong, in-phase relationship between the two data sets. PC(1) is also positively correlated to annual maximum insolation at 5°S, suggesting an additional half-precession cyclicity is present in the data set. PC(1) is also correlated to primary productivity records in the Indian Ocean, suggesting an out-of-phase correlation to Indian Ocean Dipole strength at the precession band (∼23 kyr). This may be due to a subtropical moisture source for the Malawi catchment that is opposite in polarity to the IOD. It also may be due to meridional gradients observed in climate modeling, where there is a shift in polarity of the control of precession latitudinally across the equator. The frequent, high-magnitude lake-level variability observed here would have had a major impact on the evolution of cichlid fish due to a reduction in the total habitat area, especially the amount of rocky coastline available, and isolation of Lake Malawi into separate basins. Isolation occurred at least 23 times over the past 540--580 ka, which could have easily induced allopatric speciation of the lake's pelagic cichlids.