James M. Russell, PhD, Brown University
Branched glycerol dialkyl glycerol tetraethers (brGDGTs) are bacterial membrane rigidifying compounds found in soils, peats, lake, and marine sediments. We have a limited understanding of the sources of these compounds in different environments, appropriate environmental calibrations in different environments, and the applicability of those calibrations to sedimentary sequences. The goals of this award were to investigate 1) The sources and environmental controls on the relative abundances of nine brGDGTs in lake sediments; 2) New models to quantitatively relate GDGT abundances in lake sediments to environmental parameters, particularly temperature, and 3) Test the applicability of these models to reconstructing past temperature by applying the new GDGT proxy to lake sediment cores. Our work has been successful in all three areas.
1) Origin of brGDGTs in tropical lake systems
BrGDGTs are ubiquitous in soils and lake sediments; however, there has been much debate as to whether lacustrine brGDGTs are derived from eroded soils or if they are produced in situ. To better understand sources of these compounds to lake sediments, we compared brGDGT distributions and concentrations in lake sediments and catchment soils within a 3600 m altitudinal transect in western Uganda. We observed significant offsets between observed and reconstructed MAAT in soils from wet, high elevation soils but not in most dry, low elevation soils. BrGDGT distributions differ significantly between lake sediments and soils at all elevations, with greater differences at low elevations than at high elevations. These data support previous hypotheses that brGDGTs are produced in situ in lakes and suggest that the abundance of water in soil environments may play a role in controlling the distribution of branched GDGTs.
2) Development of a brGDGT temperature calibration
In 2011, we published the first temperature calibration for lacustrine brGDGTs to temperature based upon a set of 46 lake surface sediment samples in tropical Africa. These data indicated strong correlations between brGDGT distributions and temperature in these sediments. Since that time, several calibrations have been put forth relating brGDGT distributions to MAAT using a variety of linear regressions. To test these calibrations, and to improve upon the calibrations of Tierney et al. (2011), we measured the brGDGTs in 111 East African lakes surface sediments and examined three methods of calibrating brGDGTs to MAAT. These methods include traditional brGDGT calibration methods, as well as a new stepwise forward selection (SFS) calibration that uses the four combined brGDGTs that explain the most variance in temperature in our calibration set. We applied these new calibrations as well as five previously published lacustrine brGDGT calibrations to the brGDGT distributions of our surface sediment dataset and a 48 kyr sediment core from Sacred Lake, Mt. Kenya, producing the first brGDGT temperature reconstruction available from a small tropical lake. We found that the SFS calibration has a consistently lower root mean squared error of prediction (RMSEP) over the entire range of MAAT, while the other calibrations have relatively large RMSEPs, particularly between lakes with similar temperatures but variable pH. We further find that only the SFS calibration produces a credible reconstructed temperature history from Sacred Lake when compared to other last glacial maximum paleotemperature estimates from East Africa.
3) Reproducibility of temperature reconstructions in tropical East Africa
To test the reproducibility of our results from Sacred Lake, and to expand our understanding of the thermal history of tropical Africa, we have expanded our sediment core analyses to include additional lakes from Mt. Kenya, and comparison of different temperature proxies within African lakes. We used our new brGDGT SFS calibration to reconstruct temperatures from Sacred Lake (2350 m above sea level, asl) and Lake Rutundu (3081 m asl) on Mt. Kenya. Comparison of these two datasets suggest virtually identical orbital-scale trends, with a Last Glacial Maximum that is cooler than present, an early Holocene warm interval, and ~1 oC cooling from the mid-Holocene to the present. Amplitudes of deglacial temperature change at Lake Rutundu exceed those at Sacred Lake by ~2 oC, suggesting that temperatures since the LGM have increased more at higher elevations than low elevations due to changes in the temperature lapse rates.
We have also developed a new temperature record based upon brGDGTs from Lake Malawi, tropical Africa to investigate whether temperature reconstructed from brGDGTs correlate favorably with the those reconstructed from the better-known TEX86 proxy, and to assess whether brGDGTs can provide temperature information from large, tropical rift basins. TEX86 and brGDGT reconstructions show nearly identical temperature changes between the LGM and present, suggesting that brGDGTs can reconstruct temperature change from a variety of East African lake sediment cores.
4) Monitoring seasonal changes in brGDGTs in a temperate lake
Our studies of brGDGT behavior in tropical lakes take advantage of the fact that tropical lake temperature have negligible temperature variations over the seasonal cycle. To investigate possible seasonal biases in temperate systems, we collected sediment trap and water column samples on a biweekly to monthly basis for over two years at South King Pond, a small kettle lake near Montpellier, Vermont. Sediment trap samples were analyzed for brGDGT distributions, absolute abundances, and compound fluxes, and for indicators of lacustrine primary productivity. brGDGT fluxes were also compared to lake surface sediments, water column filter samples, and soil samples from the catchment. While brGDGT distributions vary over the seasonal cycle within the water column, brGDGT distributions in the sediment trap samples are constant over the seasonal cycle and are very similar to those of surface sediments across the lake. brGDGT fluxes to the sediment trap increase dramatically during the spring and fall mixing periods, in concert with increases in biogenic opal and organic carbon. Temperature reconstructions from surface sediments from brGDGTs match the flux-weighted reconstructed temperatures from sediment trap samples, and are approximately equal to the temperature of the water column during the fall mixing period. These data suggest that brGDGTs produced during fall and, to a lesser extent, spring mixing dominate the mean annual temperature in temperate lakes.