A comparison of the palaeoclimate signals from diatom oxygen isotope ratios and carbonate oxygen isotope ratios from a low latitude crater lake

Abstract

The analysis of oxygen isotope ratios (δ¹⁸O) from authigenic lake carbonates has become a well-established palaeoclimate technique. Less common is the use of δ¹⁸O in biogenic silica (e.g. diatoms, sponge spicules, phytoliths), but the technique is being increasingly utilised for non-carbonate lakes. This paper aims to compare δ¹⁸O_diatom with δ¹⁸O_carbonate in a low latitude, closed basin lake. Due to the influence of pH, rarely are both carbonate and diatoms preserved in lake sediments in concentrations high enough to measure oxygen isotopes. The δ¹⁸O composition of both carbonate and silica should reflect the isotope composition and temperature of the lake water at the time of precipitation and hence theoretically they should be equivalent. Only one comparative study of δ¹⁸O_diatom and δ¹⁸O_carbonate currently exists and that found differences between their δ¹⁸O compositions, probably because they precipitated in different seasons. Unless there is evidence to suggest that silica and carbonate precipitate at the same time, their δ¹⁸O records are likely to be different but complementary. In this study, we show that δ¹⁸O_diatom and δ¹⁸Ocalcite from Lake Tilo, Ethiopia show some similar broad climate trends. However, the δ¹⁸O_diatom curve is generally more variable and does not record two regional arid events, picked up by the δ¹⁸O_calcite data. Several possible reasons for this are discussed: the precision of the δ¹⁸O_diatom extraction technique, contamination of the diatom samples from tephra (volcanic glass), vital effects in the diatom samples, differences in their respective equilibrium isotope fractionation rates and the possibility that the dominant diatom, Aulacoseira granulata, and calcite precipitated in different seasons during this time. Currently, the problems associated with the cleaning and extraction of δ¹⁸O_diatom suggests that details found in δ¹⁸O_calcite records may be lost in δ¹⁸O_diatom records where contaminants such as tephra and clay minerals are difficult to remove.

Introduction

The isotope compositions of authigenic (e.g. calcite) and biogenic (e.g. foraminifera, ostracods, molluscs) materials are commonly used as palaeoclimate proxies. In lacustrine studies, oxygen isotope ratios record the temperature and isotope composition of the lake water at the time of formation (see Leng and Marshall, 2004 for review). Carbonates are the most commonly used material and the technique is now well established, but in non-alkaline, dilute, open lakes, carbonates may be rare or absent. In these types of lakes, diatom silica can be used as an alternative isotopic host to carbonates (e.g. Hu and Shemesh, 2003). Diatoms are photosynthetic algae that secrete a shell composed of opaline silica (SiO₂·nH₂O) and are present in most lake sediments. In very alkaline lakes (> pH 9, e.g. Lake Bosumtwi; Talbot et al., 1984), or lakes where silica is limited, diatoms are often preserved only in low concentration (Barker et al., 1994).

Whereas the temperature effects on δ¹⁸O_calcite are well understood and defined by various temperature dependant equations (e.g. Kim and O'Neil, 1997; originating from the original calculations of Epstein et al., 1953), the temperature dependence of δ¹⁸O fractionation between diatom silica and water is less certain. Schmidt et al. (1997) found no consistent value for the oxygen isotope temperature fractionation between diatoms and water when they analysed live diatoms collected from the oceans and it has been suggested that this is because the fractionation may be established during diagenesis and not primary diatom growth. Other studies, made predominantly in the marine setting, have found more consistent results and temperature fractionation values of between − 0.2‰ and − 0.5‰ °C^− 1 (Juillet-Leclerc and Labeyrie, 1987, Shemesh et al., 1992, Brandriss et al., 1998). So far, there are no recorded δ¹⁸O_diatom species differences in the marine (Shemesh et al., 1995) or freshwater environment (Rosqvist et al., 1999), hence δ¹⁸O_diatom and δ¹⁸O_calcite should only depend on δ¹⁸O_water and ambient temperature.

Shemesh et al. (2001) measured δ¹⁸O_diatom in a lake in northern Scandinavia and suggested that the record was controlled by changes in the isotope composition of the lake water during the summer months, when the majority of the diatoms formed. As evaporation was minimal, this was controlled by changes in the δ¹⁸O composition of precipitation (δp). Many high latitude δ¹⁸O_diatom studies have contributed to an understanding of variations in temperature or the δp from which a summer specific signal can be gained, although early summer thaw of winter precipitation may also be recorded as it enters the catchment (Shemesh et al., 2001, Jones et al., 2004, Rosqvist et al., 2004). Barker et al. (2001) analysed δ¹⁸O_diatom from two alpine lakes on Mount Kenya covering the last 14,000 years. There were several large isotopic shifts in δ¹⁸O of up to −18‰ in the record that were too large to be solely explained by temperature or source effects. As the negative shifts correlated to periods of higher sea-surface temperatures in the Indian Ocean, Barker et al. (2001) argued that the resulting heavier rainfall over Kenya would result in lower δ¹⁸O values of precipitation falling into the lake due to the amount effect. No studies have so far analysed δ¹⁸O_diatom from a low-latitude lake, where evaporation often dominates the hydrology and hence precipitation/evaporation balances control δ¹⁸O with changes in temperature assumed to have a negligible effect (e.g. Lamb et al., 2000).

It is rare that both carbonate and diatom silica exist together in lake sediments in significant concentrations as diatomaceous sediments tend to preserve well in lakes with a pH < 8 and carbonate in lakes with a pH > 7. Lake pH is a central controlling factor in diatom dissolution rates as it controls the dissociation of silicic acid (e.g. Barker et al., 1994). Lakes in carbonate catchments where silica is not limited and that have a pH of 7 or 8, can sometimes preserve both carbonates and diatoms in high concentrations (e.g. Sidi Ali, Morocco; Lamb et al., 1999) but this is quite unusual and as a result, very few studies have measured δ¹⁸O in both carbonate and silica (Leng et al., 2001). A comparison of δ¹⁸O in calcite and diatom silica from a small spring fed lake in southern Turkey suggested that the δ¹⁸O_diatom variations were not directly comparable to δ¹⁸O_calcite, because the silica and calcite precipitated in different seasons (Leng et al., 2001). In temperate climates, diatom growth tends to relate to seasonal changes that affect nutrients, temperature and light availability. Often diatoms will bloom in the spring when nutrients are well distributed through the lake and may peak again in the autumn (Reynolds, 1984). In the case of the Turkish lake, diatoms are thought to grow predominantly in spring, and calcite forms during the summer months (Leng et al., 2001), resulting in two contrasting, but complementary, seasonal records.

In this study we compare δ¹⁸O_diatom and δ¹⁸O_calcite from a small crater lake in the tropics. In low-latitude lakes, higher temperatures can encourage diatoms to grow throughout the year, although other mechanisms may control algal blooms such as lake-water mixing and nutrient availability (Kifle and Belay, 1990). Previous studies suggest that in the early Holocene, the lake had a high solute input due to hydrothermal groundwater inflow and the lake-level was close to the crater rim, preventing significant clastic input. A slightly alkaline pH (estimated mean 7.6) (Telford, 1998) and a constant solute supply meant that both diatom and calcite production was high. We measured δ¹⁸O_diatom and δ¹⁸O_calcite from the early Holocene period to assess the differences between the palaeoenvironmental signals produced.