A heterocyte glycolipid-based calibration to reconstruct past continental climate change

Accurately predicting future climate change hinges on understanding Earth's past climate responses to forcing. While marine sediment records provide valuable insights using proxies like UK′37, TEX86, and LDI to reconstruct sea surface temperatures (SSTs), reliable proxies for continental climates are limited. Existing lacustrine proxies such as TEX86 have limited applicability in certain lakes, while long-chain alkenones are generally absent from low-latitude lake sediments, and MBT′14 is complicated by mixed aquatic and terrestrial sources. The lack of universally applicable temperature proxies in lacustrine sediments hinders accurate reconstruction of past continental climate change and testing of climate model hindcasts. Heterocyte glycolipids (HGs), found exclusively in the heterocyte cell envelopes of nitrogen-fixing cyanobacteria, have emerged as potential temperature proxies in freshwater environments. Their relative abundance varies systematically with growth temperature, with HG26 diols increasing and HG26 keto-ols decreasing with increasing temperature. Previous studies have shown that the heterocyte diol index (HDI26) tracks seasonal changes in SWT, but a comprehensive investigation across various temperature regimes is lacking. This research aims to address this gap by studying the spatial variability of HGs, particularly HG26 diols and keto-ols, in a diverse range of lakes, establishing the relationship between HDI26 and environmental factors, and applying this proxy to a long-term sediment record from Lake Tanganyika to reconstruct past temperature changes in tropical East Africa.

Several organic temperature proxies have proven invaluable in reconstructing past sea surface temperatures (SSTs) from marine sediments. These include the alkenone unsaturation index (UK′37), the tetraether index of 86 carbon atoms (TEX86), and the long-chain diol index (LDI). However, applying these proxies to lacustrine environments presents challenges. TEX86 is only suitable for some large lakes, while long-chain alkenones are often absent from low-latitude lakes. The branched glycerol dialkyl glycerol tetraether (brGDGT) paleothermometer (MBT′14) offers an alternative, but its interpretation in lakes is complicated by the mix of aquatic and terrestrial GDGTs. The need for additional and more reliable temperature proxies for lacustrine sediments is evident. Previous work has suggested the use of heterocyte glycolipids (HGs) as potential temperature proxies in freshwater environments. HGs are found exclusively in nitrogen-fixing cyanobacteria and their relative abundance has been shown to vary with temperature in laboratory cultures. While the HDI26 has been shown to correlate with seasonal changes in SWT in a single lake, the extent of its applicability across diverse climates remains unknown. This study builds upon previous work by examining a wider range of lakes and extending the analysis to a long-term sediment record.

This study analyzed the distribution of heterocyte glycolipids (HGs) in surface sediments from 46 tropical East African lakes spanning a wide altitudinal range (615–4504 m asl), along with eight additional lakes and ponds globally. Environmental parameters such as surface water temperature (SWT), pH, oxygen concentration, lake depth, size, surface area, productivity, and cyanobacterial community composition were considered. A modified Bligh and Dyer procedure was used to extract HGs from the sediments. High-performance liquid chromatography coupled to electrospray ionization tandem mass spectrometry (HPLC/ESI-MS) was employed to separate and quantify fourteen different HGs, ranging from 26 to 32 carbon atoms. Fractional abundances (FA) of HGs were calculated by dividing the abundance of individual HGs by the summed abundances of all HGs detected in a given sample. The HDI26 was calculated using the most abundant structural isomer of HG26 diol and HG26 keto-ol in each sample: HDI26 = HG26diol/(HG26keto-ol + HG26diol). Statistical analyses, including bivariate correlation, partial correlation, and analysis of covariance (ANCOVA), were performed to determine correlations between HDI26 and environmental parameters. Hierarchical clustering was used to group lakes based on HG distribution patterns. For the long-term climate reconstruction, a sediment core from Lake Tanganyika, covering the last ~37,000 years, was analyzed using the same methods. The accuracy of the HDI26 was assessed through replicate analysis of selected lake surface sediments.

The study found HGs to be present in all freshwater environments examined, consistent with previous findings. In East African lakes, three biozones were identified based on distinct HG distribution patterns and cyanobacterial communities, indicating significant altitude-driven changes in cyanobacterial community composition. HG26 diols increased in abundance with decreasing altitude and increasing SWT, while HG26 keto-ols showed the opposite trend. A strong positive correlation (r = 0.975; p < 0.0001; n = 42) was found between HDI26 and SWT in the East African lakes. Similar correlations were observed in the globally distributed lakes and ponds, with a strong linear global correlation: HDI26 = 0.0167 × SWT + 0.5041 (r² = 0.99, n = 8). Although a strong relationship between HDI26 and SWT was observed, some deviation from the general trend existed. This scatter might be explained by shifts in the cyanobacterial community, species-specific effects, or multiple species with different blooming periods. The time-integrating nature of sediment archives can, to some extent counteract these effects. The analytical accuracy of the HDI26 was determined to be ±0.002, equaling ±0.2 °C. Application of the HDI26 to the Lake Tanganyika sediment record revealed a ~4.1 °C warming from the Last Glacial Maximum (LGM) to the present, consistent with other regional climate records. The record also shows evidence of abrupt climate change events, including a cooling event around 12,800 years BP, coinciding with the Younger Dryas. Although generally consistent with TEX86-based reconstructions, discrepancies in the mid-to-late Holocene were noted, possibly due to the primary residence depth of GDGT-producing Thaumarchaeota and varying lake stratification.

This study demonstrates that HDI26, based on the relative abundances of HG26 diols and keto-ols, is a robust proxy for reconstructing past surface water temperatures (SWTs) in diverse freshwater environments. The strong correlation between HDI26 and SWT, observed across a wide range of latitudes and altitudes, supports its global applicability and potential for reconstructing past continental climate change. The consistent findings from both the modern lake survey and the Lake Tanganyika sediment record highlight the reliability of HDI26 as a novel temperature proxy. While some scatter exists in the relationship between HDI26 and SWT, likely due to factors such as cyanobacterial community composition and species-specific responses, the time-averaging nature of sediment archives mitigates the impact of interannual variability. The agreement between HDI26-derived temperatures and other regional records further validates this proxy. The discrepancies between HDI26 and TEX86 in Lake Tanganyika during the Holocene emphasize the importance of considering the biological sources and residence depths of different proxies. This suggests that careful calibration and consideration of potential biases are crucial for interpreting paleoclimatic records based on organic proxies.

This study introduces the heterocyte diol index (HDI26) as a novel, globally applicable proxy for reconstructing past surface water temperatures in freshwater environments. The strong correlation between HDI26 and SWT across diverse lake settings, coupled with its successful application to a long-term sediment record from Lake Tanganyika, demonstrates its potential for advancing our understanding of past continental climate change. Future research should focus on refining regional calibrations to account for species-specific effects and improve the accuracy of temperature reconstructions. Expanding the analysis to additional lake systems and temporal scales will further enhance the robustness of the HDI26 and provide more detailed insights into past climate dynamics.

While HDI26 shows a strong correlation with SWT, some scatter remains, potentially influenced by variations in cyanobacterial community composition and species-specific responses to temperature. Regional calibrations might be needed to account for such variations. The accuracy of the HDI26-based reconstructions is also limited by the inherent uncertainty associated with the calibration function and potential diagenetic alterations of HGs in the sediments. The Lake Tanganyika record, while providing valuable insights into long-term climate trends, has relatively low temporal resolution, limiting the detailed study of shorter-term climate events. The study's reliance on surface water temperature measurements for calibration may not fully capture the complexity of the relationship between temperature and HDI26 in situations with significant thermal stratification.