Introduction

Proxies for lake productivity are crucial to the understanding of past ecosystem development. Information about palaeo-productivity is often inferred from stable carbon isotope analyses of bulk organic samples assuming that the organic matter in lake sediments is predominantly of aquatic origin (Lücke et al. 2003; Parplies et al. 2008; Wu et al. 2008). In contrast to terrestrial plants, aquatic plants such as algae and submerged macrophytes are often capable of utilising HCO3 − for photosynthesis at times of CO2(aq) shortage (Raven 1970) resulting in the enrichment of 13C in their organic matter. The theory for numerically linking the δ13C signal of lake organic matter (δ13Corganic) to the dissolved inorganic carbon (DIC) supply was first suggested several years ago (Oana and Deevey 1960; Hollander and McKenzie 1991). Several experiments and field studies have been undertaken to establish a proxy that will enable the derivation of information about former carbon availability from marine and lacustrine sediments (Freeman and Hayes 1992; Burkhardt et al. 1999; Bade et al. 2006). Whilst these studies enabled us to form an understanding of the basic information about the mechanisms of carbon fractionation in phytoplankton, no reliable transfer function has yet been developed. In addition to other factors that hinder the establishment of δ13Corganic as a reliable proxy, one primary limitation is that sedimentary organic matter is derived from a mixture of organisms that differ in their capabilities to utilise bicarbonate for photosynthesis. Brenner et al. (1999) and Bade et al. (2006) advise palaeolimnologists to refrain from using a transfer function that directly relates changes in bulk organic sediment 13C to changes in productivity or palaeo-CO2 levels.

To overcome this problem, isotopic analyses can be performed on separate identified fossil leaves and seeds of submerged macrophytes that are preserved in coastal and lacustrine sediments. In comparison to land plants, the interpretation of stable carbon isotope values of submerged aquatic plants (δ13Cmacrophyte) is complex due to the influence of a variety of factors. The relationship between environmental conditions and δ13Cmacrophyte has scarcely been investigated to date, and no study has yet focused on the application of stable carbon isotope compositions of organic macrophyte fossils for palaeoecological reconstructions (but see Street-Perrott et al. 2004).

Three main mechanisms might considerably influence the δ13C values of submerged aquatic plants: (1) the switch from CO2(aq) to HCO3 − uptake for photosynthesis during periods of CO2(aq) limitation; (2) variation in 13C in the DIC source within the lake; and (3) the occurrence and magnitude of the plant’s isotopic fractionation as a function of lake water inorganic carbon supply.

Many submerged plant species such as Elodea canadensis, several Potamogeton species, Zannichellia palustris, Ceratophyllum demersum, Ranunculus aquatilis and Myriophyllum spicatum have the ability to utilise HCO3 − as an additional carbon source for photosynthesis (Sand-Jensen 1983). The physiological mechanism of HCO3 − uptake of submerged macrophytes has been investigated in many studies (Steemann Nielsen 1947; Lucas 1983; Maberly and Spence 1983; Elzenga and Prins 1988; Prins and Elezenga 1989; Maberly and Madsen, 1998). Across the range of temperatures commonly encountered by aquatic plants, the δ13CHCO3 − will be 7–12‰ enriched compared to \( \delta^{ 1 3} {\text{C}}_{{{\text{CO}}_{ 2} ({\text{aq}})}} \) (Mook et al. 1974; Romanek et al. 1992; Zhang et al. 1995). Therefore, the proportion of carbon assimilation arising from the active uptake of HCO3 − will affect δ13Cmacrophyte. Accordingly, Keeley and Sandquist (1992) reported that aquatic plants exhibit a wide range of δ13C values from −33 to −11‰. The proportion of the two carbon species that is assimilated is dependent upon species-specific differences the in the capacity for active transport of the HCO3 − ion and on the proportion of CO2 to HCO3 − in the boundary layer of the leaf (Keeley and Sandquist 1992). The ratio of CO2 to HCO3 − in the water is largely a function of ambient pH (Stumm and Morgan 1970). For comparison: at pH 5.5, 80% of the inorganic carbon occurs as CO2 (aq), whereas at pH 8.5, CO2(aq) accounts for <1% and at pH 10, HCO3 − accounts for <50%, while most of the inorganic carbon occurs in form of CO3 2−, which is not available for uptake by plants.

The species-specific relationship of photosynthesis rates and dissolved inorganic carbon source is traditionally tested in pH-drift experiments in laboratory cultures. The photosynthesis-pH curve of Potamogeton pectinatus for instance indicates that this species can use HCO3 − very effectively, but that it still has a higher affinity for CO2(aq) (Sand-Jensen 1983). Stable carbon isotope ratios of −25.0 and −25.7‰ for Potamogeton pectinatus collected from flowing water of pH 7.0 and 7.5 in Finland (Osmond et al. 1981) confirm that the plants mostly utilised CO2 for photosynthesis at neutral pH and high CO2(aq) levels. However in shallow productive lakes, pH usually rises above 8 (Pentecost et al. 2006) and CO2(aq) concentration approaches zero (Stumm and Morgan 1970). Under such conditions, Potamogeton pectinatus relies solely on HCO3 − uptake as inferred from pH-drift experiments under comparable conditions (Sand-Jensen 1983).

The stable carbon isotope signatures of submerged plants are, to a certain extent, influenced by the isotopic composition of the carbon source (Papadimitriou et al. 2005). Accordingly, the δ13C signals of Potamogeton pectinatus, when growing in neutral to alkaline lakes, are a function of lake-water δ13CDIC. In contrast to modern studies, direct measurements of the stable carbon isotope composition of DIC are not possible for palaeolimnological studies of lake sediments, and analyses of various carbonate materials such as authigenic carbonates, ostracods, or molluscs can be taken as proxies for former δ13CDIC. Regardless of the carbonate component which is used in palaeolimnological studies, the detailed interpretation of stable carbon isotope records from lakes is still complicated due to the variety of factors that affect the carbon isotopic composition of lake waters (Leng 2004). Furthermore, a detailed carbonate-isotope study by Pentecost et al. (2006) revealed that pH in dense charophyte swards can increase to ~11, allowing atmospheric CO2 to directly combine with the OH− ions of water which results in strong isotopic disequilibrium between charophyte carbonate and HCO3 −. Whether or not, under such conditions, other biogenic carbonates such as mollusc and ostracod shells are also affected by isotopic disequilibrium was not investigated in this study, although it would be reasonable to assume so. In this instance, the reliability of δ13Costracods/molluscs as a proxy of former δ13CDIC is reduced and in palaeolimnological studies, such disequilibrium phases between δ13CDIC and δ13Costracods/molluscs, typically occurring in dense calcified charophyte mats, have to be identified and excluded prior to relating δ13Costracods/molluscs to δ13Cmacrophyte. These complications could be one reason why stable carbon isotopic records of organic matter and carbonate have rarely been related to each other. However, it can be concluded that when δ13C determined from aquatic macrophytes fossils from shallow lake sediments co-varies with δ13Ccarbonate, it probably reflects former changes in δ13CDIC.

In previous palaeolimnological studies, little attention has been paid to the constraints of δ13C values of organic matter to former HCO3 − availability. This is presumably due to the fact that inorganic carbon has been considered an inexhaustible resource because of continuous supply from the atmosphere (Vadstrup and Madsen 1995). However, the short-term exchange of CO2 across the air–water interface is slow, and may be inadequate to keep pace with the photosynthetic removal of CO2(aq) and HCO3 − by aquatic photoautotrophic organisms (Vadstrup and Madsen 1995). Thus, in very productive lakes and in lakes of high pH, HCO3 − concentrations drop below the level needed to saturate photosynthesis (Maberly, 1996). According to conceptual models and empirical studies, the isotopic fractionation between inorganic carbon and organic carbon in biological tissue is smaller when carbon demand exceeds supply, and the isotopic composition of the biomass may approach the value of DIC at very low HCO3 − concentrations (Fogel and Cifuentes 1993).

The occurrence and magnitude of carbon isotope fractionation related to biological carbon supply can be explored by calculating the apparent isotope enrichment factor of the submerged plant relative to external DIC (εmacrophyte–DIC). Details and equations for the underlying conceptual model are given in Burkhardt et al. (1999) and Papadimitriou et al. (2005, 2006).

When solely utilizing HCO3 −, values for εmacrophyte–DIC range between \( \varepsilon_{{{\text{CO}}_{ 2} ({\text{aq}})--{\text{HCO}}_{{ 3^{ - } }} }} \) (=temperature-dependent equilibrium isotopic difference between CO2(aq) and HCO3 −: −12 to −9‰ in the temperature range of 0 and 25°C) and zero (=maximum 13C in enrichment in new biomass occurring during periods of fast growth rate, when high carbon demand relative to its supply can hinder isotopic fractionation during assimilation). Papadimitriou et al. (2005, 2006) showed that the variability of εZostera–DIC reflects the imbalance between inorganic carbon supply and plant demand caused by spatially and seasonally varying productivity. Their studies represent, to a certain extent, the ground-truthing for the usage of palaeo-ε as a proxy of former carbon supply influenced by productivity changes.

This paper has three objectives: Firstly we determine whether modern εPotamogeton–DIC are significantly related to DIC supply. Secondly, we apply the inferred relationship to palaeo-ePotamogeton–ostracod/molluscs from a small Tibetan Lake in order to reconstruct former DIC concentration quantitatively and thirdly we verify this information with macrophyte community reconstruction as inferred from the plant macrofossil record. Potamogeton pectinatus (Stuckenia pectinata [L.] Börner) was selected as the reference species for stable carbon isotope analysis as it is a common submerged aquatic plant native to all continents except Antarctica (Van Wijk 1988). It often forms dense mats in shallow neutral to alkaline water bodies. Compared with other aquatic plant species, seeds of Potamogeton pectinatus are often preserved in the plant macrofossil records of lakes.

The Tibetan Plateau and Central Yakutia have been selected as study areas for two reasons. Firstly, they are rich in shallow lakes of variable chemical and stable carbon isotopic composition and have only been minimally influenced by anthropogenic activity. Secondly, the submerged macrophyte diversity is low or often monospecific in most lakes, which makes it easier to identify the causes of variations in δ13C Potamogeton .

Luanhaizi Lake is located on the NE Tibetan Plateau (37.59°N; 101.35°E; 3200 m a.s.l.; Fig. 1). This freshwater lake (area ~1.5 km2) has a present-day maximum depth of ~0.4 m; it is fed by a number of small streams (catchment area ~33 km2). The lake is situated in the eastern–central part of the Qilian Mountains (NE margin of the Tibetan Plateau) which are characterised by a semi-humid climate (mean annual precipitation ~500 mm, mean annual temperature −1°C). The majority of the results from a multi-proxy study of the sediment core have already been published elsewhere (Herzschuh et al. 2005, 2006; Mischke et al. 2005).

Fig. 1
figure 1

Location of Luanhaizi Lake on the north-eastern Tibetan Plateau (China)

Full size image

Materials and methods

Recent plant material was collected from shallow lakes (median lake depth: 0.5 m) on the central and northeastern Tibetan Plateau and in Central Yakutia, Russia (ESM 1). The majority of lakes are closed systems fed by surface run-off. Individual specimens of Potamogeton pectinatus, preferentially those bearing fruits, were collected at monospecific sites. Plant material was dried using tissue paper. Modern submerged vegetation coverage was qualitatively investigated at all plant sample points and the Potamogeton coverage was estimated on a three part scale (loose <50%, dense 50–90, very dense >90%) at 8 sites. Water samples were collected for chemical analysis at 0.3 m below water surface immediately before the plant samples were obtained. Electrical conductivity (EC), pH, dissolved oxygen content and water temperatures were measured using a portable field instrument (WTW Multi 340i) at 0.3 m water depth. Alkalinity was measured in the field using the Alkalinity AL 7 titration test kit of Macherey–Nagel. HCO3 − concentration was calculated using pH and alkalinity values (HCO3 −) = (2 × alkalinity-10−14+pH)/(1 + 2 × K 10 × pH2 ) and was corrected for ionic strength. Further information on the methods used for water analysis is provided in Mischke et al. (2007).

A core of 13.94 m in length was retrieved from the central part of Luanhaizi Lake The sediments were dated by means of 14C AMS and U/Th dating. Some general information on the recovered sediments is provided in Fig. 2. Abundances of carbonate-, organic- and plant fossil-rich sediments are indicative of lake phases. Core sections, which are dominated by gravelly sands and silts, indicate a rapid accumulation of flash-flood or river sediments lacking fossils. As the data presented here will not be evaluated for its chronostratigraphic significance, a generic age scale is shown in Fig. 2. Details concerning the age-depth model are described in Mischke et al. (2005). For plant macrofossil analysis, approximately 50 g of dry sediment was prepared and investigated according to Birks (2001). Detailed information on macrofossil sample preparation and a complete plant macrofossil diagram of the core are presented in Herzschuh et al. (2005).

Fig. 2
figure 2

Results of stable carbon isotope measurements on Potamogeton pectinatus (δ13C Potamogeton ) and mollusc and ostracod shells (δ13Costracods/molluscs) from Luanhaizi Lake profile. Information on age, sediment characteristics, organic and carbonate content and δ13C of bulk organic are also given (from Herzschuh et al. 2005; age-depth-profile from Herzschuh et al. 2006). Phases of probably disequilibrium between δ13CDIC and δ13Costracods/δ13Cmolluscs, indicated by abundant Chara gyrogonites (for further identification criteria see text), were shaded in grey

Full size image

To avoid the problem of carbonate contamination for isotopic measurement, recent and fossil plant material was treated with HCl (2%) for 3 h. The modern leave material was already very clean before treatment and HCl yielded no reactions in most cases. The visible reaction of HCl with carbonate at Potamogeton seeds (bubbling) ended mostly after few seconds. Sample material was then rinsed several times with H2O. Analyses of δ13C were performed using a Finnigan Mat Delta-S isotope ratio mass spectrometer. The sample is combusted at 950°C under O2 supply and the organic carbon quantitatively transferred to CO2, which enters the mass spectrometer through an elemental analyser and CONFLO III gas mixing system.

We will use stable carbon isotope values of ostracods and mollusc as a proxy on former DIC. Molluscs foreseen for carbon stable isotope determination were picked from the plant macrofossil samples. Information regarding the preparation of ostracod samples and the complete ostracod record are reported in Mischke et al. (2005). DIC was precipitated using BaCl2 directly after sampling in the field. Its stable carbon isotope composition was analysed using a Finnigan Delta Plus XL at GeoForschungsZentrum Potsdam. Mollusc shells (genus Gyraulus) and ostracod valves were reacted with 100% phosphoric acid at +75°C in a Kiel III (Kiel IV at Geoforschungszentrum) online carbonate preparation line connected to a ThermoFinnigan 252 mass spectrometer at the Institute of Geology and Mineralogy, University of Erlangen (molluscs) and Geoforschungszentrum Potsdam (ostracods). All carbon isotope ratios were determined relative to laboratory standards of known isotopic composition. All values are reported in ‰ relative to V-PDB by assigning a δ13C value of +1.95‰ to NBS19. Reproducibility was checked by replicate analysis of a laboratory standard (IAEA NBS19) and is better than 0.05‰ (1σ) for δ13C.

Results

Stable carbon isotope analyses of modern Potamogeton material, DIC and ostracods

Stable carbon isotope measurements on modern Potamogeton pectinatus leaves (δ13C Potamogeton ) from shallow Tibetan and Yakutian lakes yielded values ranging between −23.3 and +0.4% (n = 68; medianall: −12.1%; medianTibet; −12.67% medianYakutian: −9.76‰ ESM 1). The difference between the stable carbon isotope values of leaves and seeds, determined from 11 single modern Potamogeton pectinatus plants was not significant (median difference between leaves and seeds: 0.3‰; r 2: 0.97).

The host waters of sampled Potamogeton pectinatus plants cover a wide range of the key chemical parameters such as pH, alkalinity, and conductivity (ESM 1). Host water δ13CDIC measurements, available from 23 sites, range between −14.0‰ and +6.5‰ (n = 23), while δ13Costracods values were available only from 7 sites. Compared to the large total range, the isotopic differences between both carbonates were relatively small. For samples from two lakes (CTP-13; CTP-16) with the most depleted δ13CDIC, high pH (~10), and some calcified Chara in the submerged vegetation the difference between δ13CDIC and δ13Costracods is as large as −6.0 and −13.0%, respectively, indicating disequilibrium during shell calcification. Fractionation processes in these two lakes are considered to be atypical and were therefore excluded from further analysis. δ13C Potamogeton of all other lakes shows moderate correlation with δ13CDIC (r = 0.59; r 2 = 0.35).

The apparent isotope enrichment factor εPotamogeton–DIC, the difference between δ13C Potamogeton and δ13CDIC, ranges between −15.4 and +1.1%. εPotamogeton–DIC is negatively correlated to the HCO3 − concentration of the host water (range 78–2200 mg/l; Fig. 3). Regression analysis yielded an r 2 value of 0.74 (P < 0.00001) after the identification (2σ) and exclusion of one outlier (Yak-18). However the reliability of the transfer function is reduced at the upper and lower ends of the εDIC range. Potamogeton pectinatus leaf tissue from Tibetan Lakes, which are largely very shallow and alkaline, was more enriched in 13C relative to DIC (median εDIC: −4.3%) than that from deeper and less alkaline Yakutian Lakes (median εDIC: −11.0%).

Fig. 3
figure 3

Plot between εPotamogeton–DIC (as inferred from analysis of modern leaves of Potamogeton pectinatus and host-water DIC) and host-water lnHCO3 − from shallow lakes on the eastern Tibetan Plateau and Central Yakutia. Correlation coefficient (r), coefficient of determination (r 2) and inferred transfer function were indicated as well

Full size image

As information regarding plant coverage is available for only eight sites, we cannot statistically determine whether HCO3 − concentration, Potamogeton pectinatus coverage and εPotamogeton–DIC are significantly related to each other. A slight trend of decreasing HCO3 − concentrations with increasing Potamogeton coverage may be inferred Fig. 4.

Fig. 4
figure 4

Plot showing the relationship between modern εPotamogeton–DIC and Potamogeton coverage from 6 monospecific lakes on the eastern Tibetan Plateau

Full size image

Stable carbon isotope analyses of fossil material from Luanhaizi Lake

Stable carbon isotope measurements of fossil Potamogeton pectinatus seeds from the Luanhaizi Lake profile range between −24.0 and +2.8% (n = 158, median: −8.9%). The values vary strongly between the different sections of the core but are quite similar within one single section.

δ13Costracods/molluscs was taken as a proxy of former δ13CDIC to infer palaeo-εPotamogeton–ostracods/molluscs. Stable carbon isotope values from ostracod valves and fossil mollusc shells (Gyraulus sp.) from the same core range between −7.8 and +7.5% and between −20.2 and +8.4%, respectively, indicating co-variation throughout the core (Fig. 2).

Prior to the calculation of palaeo-εPotamogeton–ostracods/molluscs, we assessed the Luanhaizi Lake record for phases of disequilibrium (Pentecost et al. 2006) between former δ13CDIC and δ13Costracods/molluscs by applying the following criteria: (1) abundant charophytic remains (gyrogonites, stem encrustations) in the macrofossil record (Herzschuh et al. 2005; Fig. 2); (2) increased carbonate content compared with the organic content (Mischke et al. 2005, Fig. 2), and (3) strongly depleted δ13C Potamogeton (Fig. 2). The core sections 12.10–11.50 m, 4.60–4.30 m and 3.20–1.95 m show these characteristics and were therefore excluded from the calculation of palaeo-εPotamogeton–ostracods/molluscs.

Outside the sections of probable disequilibrium, values of δ13C Potamogeton from the Luanhaizi Lake core are weakly correlated with δ13Costracods (r = 0.33). Palaeo-εPotamogeton–ostracods outside these sections ranges between −14.7 and 1.6% and palaeo-εPotamogeton–molluscs ranges between −10.9 and +6.2%. Both curves show similar trends (Fig. 2). The εPotamogeton–DIC-lnHCO3 − transfer function inferred from the modern data set was applied to palaeo-εPotamogeton–ostracods values in order to quantify palaeo-HCO3 − concentrations (Fig. 5). Reconstructed lnHCO3 − concentration shows considerable variation throughout the core (lnHCO3 − range 4.45–7.36 mg/l, median: 5.68 mg/l). Sections with low (10.0–9.1 m), intermediate (12.6–12.1; 8.9–8.3; 7.8–7.4 m) and high HCO3 − concentrations (11.5–10.9; 7.1–6.8; 9.1–8.9; 8.3–7.8; 4.3–3.9; 0.9–0 m) alternate throughout the core.

Fig. 5
figure 5

Reconstructed lnHCO3 − for Lake Luanhaizi (NE Tibetan Plateau) as inferred from the application of the εPotamogeton–DIC-lnHCO3 − transfer function to palaeo-ε Potamogeton –ostracods. Phases of low, intermediate, and high HCO3 − concentration are shaded in dark grey, grey, and light gray, respectively. For comparison, information on lake development concerning macrophyte composition and productivity (as indicated by the fossil macrophyte record after Herzschuh et al. 2005) lake depth and stability (as indicated by the ostracod Fabaeformiscandona danielopoli and by the carbonate content after from Mischke et al. 2005), and algae productivity (indicated by Pediastrum concentration after Herzschuh et al. 2005)

Full size image

Discussion

Modern δ13CPotamogeton and inferred εPotamogeton–DIC

Measurements of δ13C Potamogeton yielded an extremely wide range for modern plant leaves (−23.3 to +0.4%). A comparable δ13C range has never been reported for any other plant species. Morrill et al. (2006) used Potamogeton seeds (probably Potamogeton pectinatus) from a shallow lake on the Tibetan Plateau for radiocarbon dating and provided δ13C values of Potamogeton but without any discussion of the δ13C Potamogeton results. Although the δ13C range of fossil Potamogeton in this lake is narrower (from −18.0 to −5.0%; n = 50), these data generally confirm the wide range of our observed δ13C Potamogeton values.

We initially checked whether a consistent offset existed between the δ13C values of leaves and seeds from modern plants. We expected seeds to be more depleted than leaves as more synthesis steps are involved in leaf formation, which could increase the total fractionation. However when compared with the wide data range for δ13C Potamogeton , the observed differences between the δ13C Potamogeton values of leaves and seeds are negligible. Modern δ13C Potamogeton measurements of leaves can therefore be directly compared with values obtained from fossil seeds.

The studied lakes on the Tibetan Plateau and in Central Yakutia had pH values between 8 and 11 (medianTibet: 9.8 medianYakutia: 8.6; ESM 1). In such lakes, HCO3 − is the only carbon which is available for uptake by plants (Zhang et al. 1995). When compared with the values of Potamogeton pectinatus growing in neutral flowing water in Finland (Osmond et al. 1981), our modern δ13C Potamogeton are enriched in 13C.

In alkaline lakes, δ13Cmacrophyte is assumed to partly reflect the δ13CDIC and partly reflect the kinetics of assimilation of carbon isotopes (Hemminga and Mateo 1996; Papadimitriou et al. 2005). Approximately 35% of the site-to-site variability in δ13C Potamogeton from Tibetan and Yakutian Lakes can be explained by its modest but significant correlation with δ13CDIC. A similar relationship was observed for intra- and inter-annual variations of δ13C Zostera and δ13CDIC in Z. marina and Z. noltii meadows at the Danish and Welsh coasts (Papadimitriou et al. 2005, 2006). Our data set, composed of information from various small closed lake systems has a much wider range in δ13CDIC and δ13Cmacrophyte than the samples from different coastal habitats. This is due to the nature of the macrophyte’s host waters being more homogeneous in coastal areas but highly variable in each closed lake system due to variation in hydrological conditions (Leng and Marshall 2004).

The specific kinetics of the assimilation of carbon by aquatic plants yields that fractionation during phases of strong growth rates and limited carbon supply is suppressed, leading to isotopically enriched biomass compared with the biomass produced during phases of carbon excess (Burkhardt et al. 1999). The calculated apparent isotope enrichment factors εPotamogeton–DIC of Tibetan and Yakutian plants indicates the occurrence and magnitude of the growth related biological carbon isotope effect. The observed range of εPotamogeton–DIC covers the entire possible range of −12 < εPotamogeton–DIC < 0. It is larger than the ε Zostera–DIC range from Zostera marina meadows on the Danish coast (Papadimitriou et al. 2005). This indicates that we sampled both habitats with optimal growing conditions for Potamogeton pectinatus in terms of salinity, depth and temperature and also sites at the tolerance boundaries where productivity is reduced. The studies of Papadimitriou et al. (2005, 2006) have shown that seasonal variability of isotopic fractionation of macrophytes is a function of productivity-driven carbon supply. Our inferred εPotamogeton–DIC are also significantly related to lnHCO3 − concentration (r − 0.86 r 2 0.74; P < 0.00001). However, several factors may reduce the reliability of the proposed transfer function:

  1. 1.

    The residuals between the observed values and those estimated by the transfer function increase considerably at ε Potamogeton–DIC values below −10% (=lnHCO3 − 6.51 mg/l). This is reasonable as such a strong depletion in 13C of the biomass relative to the DIC probably indicates that plants at these sites have not solely relied on HCO3 − as a carbon source but that some isotopically light CO2 was assimilated as well.

  2. 2.

    In our study, stable carbon isotope measurements performed on Potamogeton pectinatus leaves were related to the δ13CDIC of the host water. The time windows for lake water sampling (to infer HCO3 − concentration) and leaf formation differ strongly: DIC uptake during leaf formation is an integration of values over time during the growing period whilst HCO3 − concentration in sampled host waters is just a snapshot of the conditions at that particular moment.

  3. 3.

    Information concerning Potamogeton pectinatus coverage for our modern sample set was incomplete and lacked data on Potamogeton pectinatus productivity. Hence, the relationship between productivity and HCO3 − concentration is not established in our data set. This would represent an important piece of evidence when using εPotamogeton–DIC as a proxy of former productivity. However, it has been proven by many studies that HCO3 − concentration decreases with increased macrophyte productivity (Van den Berg et al. 2002).

  4. 4.

    The number of lakes sampled for DIC and hence the number of data points available for setting up the transfer function was comparatively low (n = 20).

  5. 5.

    Modern ε Potamogeton–DIC values originate from two different regions, the Tibetan Plateau and Central Yakutia. These regions differ markedly in heights a.s.l. by about 4000 m (Table S1. Electronic Supplementary Material) meaning large differences in atmospheric CO2 partial pressures (pCO2) which could have an effect on isotopic fractionation during plant uptake. Benthien et al. (2007) found in a laboratory study that even the isotopic signals of marine algae, which were assumed to rely on CO2(aq) rather than on HCO3 − such as Potamogeton pectinatus, were negligibly effected by artificial variations of atmospheric pCO2.

  6. 6.

    The plasticity of Potamogeton pectinatus has to be taken into account when applying the proposed transfer function to fossil plant material from further regions (Van Wijk et al. 1988; Pilon et al. 2003) although we found no obvious regional difference in the ability of plants to assimilate inorganic carbon.

Nevertheless, the significance of the relationship between lnHCO3 − and εPotamogeton–DIC and the consistency between the results and the conceptual model justifies the application of the inferred transfer function to palaeo-εPotamogeton-ostracods/molluscs.

Changes in palaeo-ε, HCO3 − and productivity at Luanhaizi Lake

The reliability of the εPotamogeton–DIC-lnHCO3 − transfer function was applied to the stable carbon isotope record from Luanhaizi Lake. As with modern leaves, measurements of δ13C Potamogeton for fossil seeds yielded an extremely wide range (−24.0 to +2.8%). The curves for δ13C Potamogeton and δ13Corganic (from Herzschuh et al. 2005) show no co-variation and their absolute ranges show only slight overlap (Fig. 2). Bulk organic matter at Luanhaizi Lake has various origins (terrestrial plants, algae, submerged plants) characterised by different stable carbon isotopic fingerprints (Herzschuh et al. 2005).

Calculation of palaeo-εostracods/molluscs-Potamogeton and the application of the εPotamogeton-DIC-lnHCO3 − transfer function yielded reasonable results. The information obtained can be compared to a variety of environmental variables providing a comprehensive environmental history. These variables include aquatic plant macrofossils, non-pollen palynomorphs, δ13Corganic, C/N ratios, n-alkane distribution (Herzschuh et al. 2005), ostracods, element concentrations, (Mischke et al. 2005) and terrestrial pollen (Herzschuh et al. 2006).

Phases of strongly enriched δ13C Potamogeton relative to δ13Costracods/molluscs (high palaeo-ε, Fig. 2) and reconstructed low lnHCO3 − concentration (Fig. 5) are largely related to sections of high macrophyte fossil abundance in the sediment (taken from Herzschuh et al. 2005). Submerged plant macrofossils are considered to represent a reliable proxy of local macrophyte colonization and productivity (Birks 1973; Zhao et al. 2006).

Intermediate lnHCO3 − concentrations were recorded between 12.6 and 12.1 m. Beside Potamogeton pectinatus seeds, the most frequent macrofossils in that part of the core are fruits of Zannichellia palustris and oospores of Chara cf. aspera (Fig. 5). Hence, dense macrophyte meadows probably led to the limitation of HCO3 − in the water. A modern analogue for the former situation in Luanhaizi Lake may be the recent vegetation conditions in Veluwemeer (Netherlands), which is a shallow alkaline lake dominated by Potamogeton pectinatus and Chara aspera (Van den Berg et al. 2002). Detailed investigations of the relationship between HCO3 − availability and vegetation composition in Veluwemeer showed that very low HCO3 − concentrations were observed in areas dominated by Chara, while the areas dominated by Potamogeton had higher HCO3 − concentrations. In mixed stands, the growth rates of both species suffer from low HCO3 − concentrations. Chara aspera seems to exploit the HCO3 − reservoir more efficiently under conditions of low HCO3 − supply than Potamogeton pectinatus. It is probable that the considerable variations of reconstructed lnHCO3 − in the core section 12.6–12.1 m from Luanhaizi Lake can be attributed to local changes in the relative abundances of the single macrophyte species. In contrast to the observations from Veluwemeer, the lowest lnHCO3 − concentrations of the whole Luanhaizi Lake core were recorded during a phase of almost monospecific dense Potamogeton pectinatus mats as inferred from the macrofossil record between 10.0 and 9.1 m.

Phases of low palaeo-ε/high lnHCO3 − occur mostly during phases of deep stable lake conditions as indicated by the ostracod species Fabaeformiscandona danielopoli (Fig. 5; the complete ostracod record is given in Mischke et al. 2005) and/or high planktonic productivity as indicated by Pediastrum and Daphnia ephippia abundances (Herzschuh et al. 2005).

To conclude, absolute values of reconstructed HCO3 − concentrations are confirmed by modern analogues. Variations can reasonably be related to changing conditions within the lake as suggested by other proxies. However, several factors may reduce the reliability of palaeo-ε reconstruction.

Conclusions

The modern and fossil data support our hypothesis that the combined interpretation of stable carbon isotope data from submerged macrophyte fossils and from lake carbonates could in general provide information about former lake DIC concentrations and about former lake productivity. Our investigations of modern Potamogeton pectinatus from Tibetan Plateau in relation to lakewater environmental variables demonstrate that δ13C Potamogeton is mainly influenced by the ability of the species to utilise HCO3 − uptake for photosynthesis, by the stable carbon isotope composition of DIC in the lake and by a growth-rate dependent enrichment. The growth-rate dependent apparent isotope enrichment factor can be approximated by the difference between δ13C Potamogeton and δ13CDIC. Calculated εPotamogeton–DIC from modern Tibetan and Yakutian Lakes is significantly related to HCO3 −concentration and is probably driven by submerged plant productivity.

Palaeo-HCO3 − supply was reconstructed for Luanhaizi Lake by applying the εPotamogeton–DIC-lnHCO3 − transfer function to palaeo εPotamogeton–ostracods data. Our results can be reasonably well explained in terms of palaeo-productivity changes known from other lines of evidence (e.g. plant macrofossils, ostradods). From a palaeolimnological perspective, the calculation of palaeo-εmacrophyte is a potential new proxy for inferring former hydrochemical conditions in neutral and alkaline lakes. Future results from similar studies can be directly compared to our investigation as Potamogeton pectinatus has a world-wide distribution. The former availability of carbon, a main nutrient element for plants, can be reconstructed using δ13C Potamogeton . The proxy should be capable of tracing ecosystem state changes and main changes in the aquatic submerged plant community in cases where no comprehensive macrofossil record is available. Our investigation shows how complex the interpretation for single species δ13C is and that it is almost impossible to identify former productivity/carbon availability changes from bulk organic analysis. The stable carbon analysis of submerged plant biomarkers (e.g. mid-chain n-alkanes) could possibly be used when macrofossils are not preserved in lake sediment records.