Environmental controls on the distribution of bacterial tetraether membrane lipids: Constraints on the MBT-CBT paleothermometer

Branched glycerol dialkyl glycerol tetraethers (GDGTs) are membrane lipids of soil bacteria that occur ubiquitously in soils, peats, and marine sediments. The structures of the branched GDGTs vary in the number of methyl groups (4 to 6) attached to the alkyl chains and can contain up to two cyclopentane moieties. An empirical study showed that their distribution in over 130 globally distributed soils is mainly controlled by mean annual air temperature (MAT) and soil pH, and can be expressed using the Methylation of Branched Tetraethers (MBT) index and the Cyclisation of Branched Tetraethers (CBT) index. The CBT index relates with soil pH, whereas the MBT index is related to both MAT and, to a lesser extent, soil pH. The combination of the MBT and CBT indices can then be used as a proxy for paleotemperature and past soil pH; the so-called MBT-CBT proxy. Branched GDGTs are transported with soil organic matter by rivers and deposited in marine coastal regions, where down-core analysis of branched GDGTs may yield an integrated MAT record for the river basin. However, before this proxy can be confidently applied, the exact influence of environmental controls on the distribution of branched GDGTs needs to be further investigated and possible constraints on its applicability need to be examined.The influence of MAT and soil pH on the distribution of branched GDGTs, and thus the MBT-CBT proxy, was assessed in different natural settings. The direct effect of temperature changes was determined by analyzing branched GDGTs in soil transects away from two hot springs. The distribution of branched GDGTs in these geothermally heated soils changed substantially; MBT index values decrease with increasing distance from the hot spring and decreasing soil temperature. Although the slope of this relation is similar to that based on the global soil calibration set, the intercept is different. This difference can be explained by the use of in situ soil temperature in the relation for the geothermally heated soils rather than MAT, which was used in the initial study.The influence of soil pH on the degree of cyclisation and the abundance of branched GDGTs was tested in soils from long term pH manipulated field plots with a pH range of 4.5-7.5. The distribution of core lipid (CL) and intact polar lipid (i.e. with headgroup still attached; IPL)-derived branched GDGTs was analyzed for each plot. The CBT index values of both CLs and IPL-derived branched GDGTs were significantly related with the actual soil pH, confirming the direct influence of soil pH on the distribution of branched GDGTs. Also, CBT-based pH estimates were similar to actual soil pH. Furthermore, soil pH affected the abundance of branched GDGTs, which showed a decrease with increasing pH. The abundance of Acidobacteria, suggested as a potential source of branched GDGTs, is also known to increase with decreasing pH, providing further support for this suggestion.Analysis of branched GDGTs in soils along an altitude transect on Mount Gongga in China showed that among the many varying environmental factors, MAT and soil pH explain most of the variation in the distribution of branched GDGTs. Despite the relatively large scatter, the temperature lapse rate of MBT-CBT-based MAT estimates resembles the lapse rate caused by adiabatic cooling of air with increasing altitude. Based on this (indirect) relation between branched GDGTs and altitude, the MBT-CBT proxy was suggested as a proxy for paleoelevation, in addition to soil n-alkane dD values. To assess their suitability as proxies for paleoelevation, the relation of altitude of both branched GDGTs and the dD of n-alkanes in soils along the slopes of Mt. Gongga and Mt. Kilimanjaro (Tanzania) was studied. The data show that both proxies can be subject to relatively large uncertainties, but that a combination of both proxies likely results in a more reliable paleoelevation reconstruction.Although the MBT-CBT proxy was found to correlate best with annual MAT, several studies suggest a potential seasonal bias in MBT-CBT-derived air temperature estimates, for example caused by a season of optimum growth of the bacteria that produce branched GDGTs. Therefore, the distributions and abundance of CL and IPL-derived branched GDGTs in mid-latitude soils from eight different locations were monitored over an annual cycle. MBT-CBT-reconstructed temperatures remained constant throughout the year and also branched GDGT abundances (both CL and IPL-derived) showed no bias towards a specific season. These results indicate that seasonality does not influence MBT-CBT-derived temperature estimates, and that the turnover time of the branched GDGT pool in soils is in the order of one year or more.Branched GDGTs in Svalbard soils and nearby fjord sediment were analyzed to verify if the MBT-CBT proxy is applicable in high latitude environments, characterized by a low MAT, and to what extent branched GDGT distributions in marine sediments reflect those on land. Although branched GDGT concentrations in the soils were relatively low due to the cold climate and little soil development, MBT-CBT based MAT estimates were similar to the measured MAT (ca. -4°C and -6°C, respectively). The distribution of branched GDGTs in the fjord sediments was strikingly different from that in the soils, and was dominated by branched GDGTs with one cyclopentane moiety. This resulted in unrealistic MAT estimates, and suggests that at least part of the branched GDGTs in coastal marine environments receiving little soil organic matter may be predominantly produced in situ. The MBT-CBT proxy should therefore only be used for marine coastal sediments at sites receiving a substantial input of soil organic matter relative to the amount of marine organic matter, i.e. close to a river mouth.More information on the possible origin of the branched GDGTs was gained by the analysis of IPL-branched GDGTs. Branched GDGTs with a glucose, glycuronic acid, hexose-glycuronic acid, phospho-hexose, or hexose-phosphoglycerol head group were identified in a Swedish peat, based on mass spectrometry. Changes in distribution of these IPL-branched GDGTs with depth in the Swedish peat bog were monitored using a newly developed Selected Reaction Monitoring (SRM) method. The response, as a measure for concentration, of all monitored IPLs increased below the water table, suggesting that they are primarily produced in this part of the peat. This is supported by the increase of absolute concentrations of CL and IPL-derived branched GDGTs below the water table. The trends of the individual IPLs with depth are different, but phospholipids show a larger increase in response with depth compared to glycolipids. The relatively large abundance of the less stable phospholipids implies that branched GDGTs are presumably produced by anaerobic bacteria in the anoxic part of the peat. Comparison of the IPL trends with 16S rRNA data hints toward Acidobacteria as possible producers of the branched GDGTs. Further application of the SRM method on two different soil types showed that most IPL-branched GDGTs were also present in these soils, albeit in different distributions.The soil set on which the MBT-CBT proxy was initially calibrated, was extended to 278 globally distributed surface soils to recalibrate the proxy. As only 26% of these surface soils contained all nine different branched GDGTs, soils containing the seven most common branched GDGTs were selected for calibration purposes. Regression analysis resulted in new transfer functions for pH and MAT, based on the CBT index and adjusted MBT index (MBT’): pH= 7.90 – 1.97×CBT (r2=0.70, RMSE=0.8) and MAT = 0.81 – 5.67×CBT + 31.0×MBT’ (r2=0.59, RMSE=5.0 °C, n=176). The correlation coefficient of the new function is substantially lower than that of the original equation. Other statistically derived functions only marginally improved the correlation coefficient, while these functions could not be explained straightforwardly by physiological mechanisms. Local factors or seasonality cannot fully explain the reSummarymaining scatter in the calibration, but reconstructed MATs for soils from arid regions tend to substantially underestimate the actual MAT, which indicates that absolute branched GDGT-derived MAT records should be interpreted with caution. Application of the new transfer functions on previously published MBT-CBT derived paleotemperature records shows that trends remain similar, but that absolute temperature estimates and the amplitude of temperature changes are lower for most records, although in agreement with independent proxy data.Application of the MBT-CBT proxy on a loess-paleosol sequence in China resulted in a paleotemperature record covering the last 34 kyr. The record shows that air temperatures varied in phase with northern hemisphere summer insolation, and that the onset of deglacial atmospheric warming started around 19 kyr BP, parallel in timing with other continental records. Comparison of the branched GDGT-based MAT record with oxygen isotope records of stalagmites from Chinese caves, representing East Asian summer monsoon precipitation, shows that the intensification of the summer monsoon lagged that of deglacial atmospheric warming. The results indicate that the onset of deglacial atmospheric warming and the intensification of the summer monsoon may have been controlled by different factors.In conclusion, the data presented in this thesis show that temperature and soil pH are indeed the main controls the distribution of branched GDGTs in soils, and that the MBT and CBT indices are suitable as paleothermometer and proxy for past soil pH in terrestrial settings as well as coastal marine sites receiving a substantial input of soil organic matter. However, care must be taken in interpreting absolute MAT values generated with this proxy