Report

Photosynthesis-Induced Biofilm Calcification and Calcium Concentrations in Phanerozoic Oceans

See allHide authors and affiliations

Science  01 Jun 2001:
Vol. 292, Issue 5522, pp. 1701-1704
DOI: 10.1126/science.1057204

Abstract

Photosynthetic carbon assimilation is commonly invoked as the cause of calcium carbonate precipitation in cyanobacterial biofilms that results in the formation of calcareous stromatolites. However, biofilm calcification patterns in recent lakes and simulation of photosynthetically induced rise in calcium carbonate supersaturation demonstrate that this mechanism applies only in settings low in dissolved inorganic carbon and high in calcium. Taking into account paleo–partial pressure curves for carbon dioxide, we show that Phanerozoic oceans sustaining calcified cyanobacteria must have had considerably higher calcium concentrations than oceans of today. In turn, the enigmatic lack of calcified cyanobacteria in stromatolite-bearing Precambrian sequences can now be explained as a result of high dissolved inorganic carbon concentrations.

There is increasing evidence that the ocean's chemical composition fluctuated considerably during the Precambrian as well as during the Phanerozoic (1–3). Conditions favoring carbonate precipitation can be inferred from cyanobacteria, because their calcification is dependent on the supersaturation of the ambient water with respect to CaCO3 minerals (4,5). Recurrent calcification of microbial mats and biofilms, commonly dominated by cyanobacteria and/or nonphototrophic bacteria, leads to the formation of stromatolites, laminated reeflike structures that once were common in the marine realm. Starting with a limited number of Archean occurrences, stromatolites were most widespread and abundant in the Proterozoic, followed by a severe decline in the latest Proterozoic (6, 7). Subsequently, clotted microbial deposits, so-called thrombolites, were common since the beginning of the Cambrian, but finally vanished almost completely from normal-marine environments during the Late Cretaceous, together with the remaining stromatolites (8).

Carbonate stromatolites of the Precambrian almost never exhibit tubular or vesicular carbonate microfossils of cyanobacteria, which should have been the dominant group of microorganisms in most of the shallow-marine stromatolite occurrences (8, 9). This appears enigmatic (9) because cyanobacterial filament and cell remains have been preserved when early silification occurred (10). On the other hand, such calcified cyanobacteria occur nearly continuously in fluctuating abundances from the Cambrian to the Cretaceous, either forming skeletal stromatolites, oncolites, or distinct microfossils such as Girvanella (10).

Three crucial steps in cyanobacterial biofilm calcification can be distinguished: (i) the initial supersaturation of the macroenvironment with regard to CaCO3 minerals, (ii) the shift of the carbonate equilibrium to surpass the critical supersaturation for CaCO3 mineral formation, and (iii) the process of seed crystal formation. In the following, saturation is given by the saturation indexEmbedded Image(1)where IAP denotes the ion activity product andK SO is the solubility product of the corresponding mineral (solid phase) (11). Case studies of nonmarine settings exhibiting present-day calcifying cyanobacterial biofilms indicate a 9.5- to 15-fold supersaturation of the ambient water with respect to calcite (12–14). Therefore, an ∼10-fold supersaturation (SICc = 1.0) is here taken as a prerequisite for cyanobacterial calcification.

CaCO3 mineral nucleation in microbial biofilms always starts from the common gel-like exopolymer matrix. Apart from their biological functions (15), extracellular polymeric substances (EPS), arranged in a highly hydrated network, affect mineralization in two ways. (i) They permit the establishment of physicochemical and chemical microgradients by reducing diffusion rates, and (ii) they bind divalent cations such as Ca2+ and Mg2+ from the liquid phase. The latter reflects the abundance of acidic groups within the EPS, which is not merely a sticky mass trapping and binding particles, but is composed largely of highly reactive, carboxylated polysaccharides (15,16). In accordance with the concept of organic matrix–mediated biomineralization (17, 18), the stereochemical arrangement of these acidic groups is of crucial importance in mineral nucleation. Highly ordered acidic groups at defined distances that correspond to the crystal lattice promote nucleation, whereas irregularly arranged groups inhibit precipitation. Because EPS are, in contrast to organic matrices in biomineralization, highly diverse and disordered, this latter effect prevails in biofilms. Nucleation of CaCO3 only occurs at acidic groups, which are suitably arranged mainly by accident, after a sufficient diffusive Ca2+ supply surpasses the complexation capacity of the EPS. As a result, if no effective physicochemical microgradients are established by the microorganisms, randomly distributed crystals form in the EPS (19).

Stimulated by the hypothesis (20) that ocean chemistry before 1000 Ma (million years ago) might have been highly alkaline, with pH values up to 10, and that soda lake microbialites may have Proterozoic counterparts, we investigated biofilm mineralization in soda lakes. Typically, physicochemically driven CaCO3precipitation of randomly distributed crystals occurs in the biofilm EPS at the contact to the supersaturated liquid phase so that no defined sheath impregnation results (12–14, 19).

By contrast, for the formation of calcified cyanobacteria, it is critical that photosynthetic carbon withdrawl creates a pH-microgradient within the trichome-surrounding sheath to induce a spatially defined CaCO3 impregnation (19). A causal relation of photosynthetic carbon assimilation and carbonate precipitation in cyanobacteria has been shown convincingly for freshwater settings, both by isotopic and experimental studies (21, 22). So, how strong is the relative effect of photosynthesis in shifting carbonate equilibrium in the different hydrochemical settings?

The basic concept is expressed by the solubility productEmbedded Image(2)A rise in the concentration of CO3 2−([CO3 2−]) will cause the ion activity product to exceed the solubility product, resulting in higher supersaturation the lower the initial concentration in dissolved inorganic carbon (DIC) is. In the case of photosynthesis, the rise in [CO3 2−] results from the disproportionation of bicarbonate to carbonate and CO2, which is fixed by the organisms (23):Embedded Image(3)To simulate the susceptibility of calcite supersaturation of high-Ca2+/low-DIC and low-Ca2+/high-DIC waters against a given photosynthetic carbon removal, we calculated model curves (24). The same procedure was applied to water-chemistry data of 29 settings sustaining calcifying biofilms (24). The resulting difference between the calcite supersaturation index before and after carbon subtraction, termed ΔSICc, is plotted against DIC at equilibrium conditions in Fig. 1.

Figure 1

Computed microenvironmental rise in the calcite supersaturation index (ΔSICc) of natural waters (24) upon a carbon decrease by 200 μmol liter−1 as a function of DIC for present-day atmospheric pCO2 = 10−3.5atm. ΔSICc = 0 corresponds to an initial 10-fold calcite supersaturation necessary for any cyanobacterial biofilm calcification.

As a result, high-DIC waters—i.e., soda lakes and adjacent mixing zones at freshwater inputs—show a negligible to minor rise in calcite supersaturation upon carbon removal (Fig. 1). This is because of the large carbon pool and the strong pH buffering of soda lake waters. All of these high-DIC settings lack CaCO3 precipitates spatially linked to cyanobacterial sheaths (25). Low-DIC marine settings and Na-Cl–dominated lakes show minor to moderately high ΔSICc values. Lake Tanganyika (Tanzania), Lake Satonda (Indonesia), Lake Thetis, and Shark Bay (Western Australia) cluster between 0.1 and 0.2 ΔSICc. These settings, which presently sustain microbial reef growth, show clotted, microcrystalline precipitates within the cyanobacterial biofilm EPS (25). Nonetheless, precipitates immediately linked to cyanobacteria occur regularly in Lake Clifton (Western Australia) at a ΔSICcof 0.20, although a clotted fabric still dominates in these thrombolites (25). Thus, a ΔSICc of at least 0.20 is here considered necessary for the formation of calcified cyanobacteria as defined by CaCO3-impregnated cyanobacterial sheaths. Saline pools of the Aldabra atoll (Seychelles) (ΔSICc = 0.18 − 0.30) exhibit stromatolites that locally contain cyanobacterial filament moulds encased in microcrystalline carbonate (25). Riding (26) described calcareous sheath impregnation inPlectonema from another pool on Aldabra high in Ca2+ as a possible analog of the calcified cyanobacteriumGirvanella.

Low-DIC nonmarine settings—i.e., hardwater lakes and creeks—show a relatively strong rise in calcite supersaturation upon DIC removal (ΔSICc = 0.3 − 0.5). The ΔSICcis generally higher compared with marine water of the same initial DIC (Fig. 1) where Mg2+ and SO4 2− tend to lower free Ca2+ and CO3 2−concentrations as a result of ion pairing and complexation. Commonly, physicochemical CO2 degassing dominates in hardwater creeks, and cyanobacterial filaments mineralize externally (4, 6). Only when in equilibrium with atmospheric pCO2 does sheath impregnation by microcrystalline carbonate occur. Notably, the few known case studies of proven photosynthesis-induced CaCO3 precipitation show the most substantial calculated rise in supersaturation upon DIC removal (Fig. 1): Fayetteville Green Lake (21) and a pond from the Everglades (22).

Figure 2 demonstrates the photosynthetic effect on supersaturation at pCO2 levels expected for certain periods in Earth's history. Whereas high ΔSICc values (>0.20) can be achieved for apCO2 between 10−3 and 10−2.5 atm, if sufficient Ca2+ is added, at higher pCO2 (10−2, 10−1 atm) ΔSICc generally remains below this threshold, and calcified cyanobacteria are unlikely to occur. The dotted line in Fig. 2 illustrates the susceptibility of standard seawater to carbon removal by rising pCO2 at a constantly held 10-fold initial supersaturation. After reaching a maximum ΔSICc at pCO2 = 10−3 atm, standard marine waters show an increasingly low reaction to carbon removal at higher pCO2conditions. In any case, only an additional Ca2+ input can cause the ΔSICc of seawater to exceed the threshold for cyanobacterial sheath calcification of 0.20.

Figure 2

Computed rise in the calcite supersaturation index (ΔSICc) of marine waters upon a carbon removal of 200 μmol liter-1 as a function of DIC at various values of pCO2. Calculations are based on a supposed 10-fold initial supersaturation (ΔSICc= 0). The rise in supersaturation refers only to the microenvironment surrounding cyanobacterial cells, not the whole ocean.

As a result of this effect, a lower threshold of Ca2+concentrations can be calculated for Phanerozoic oceans that sustained calcified cyanobacteria (Fig. 3). Accepting palaeo-pCO2 curves of Berner (27) and Ekart et al. (28) and a ΔSICc of 0.20 to enable CaCO3 impregnation of cyanobacterial sheaths by photosynthesis, at least 13 mmol liter−1 Ca2+ were necessary to sustain calcified cyanobacteria in Carboniferous to recent oceans. Earlier in the Paleozoic, Ca2+ concentrations up to at least 23 mmol liter−1 must be assumed.

Figure 3

(A) Calculated Ca2+ minimum curves for photosynthetically induced cyanobacterial sheath impregnation in comparison with the calculated Ca2+ curve of the modeling by Stanley and Hardie (2). Model I is based on the pCO2curve of Berner (27), and model II on thepCO2 curve of Ekart et al.(28). (B and C) Predicted relative abundances of calcified marine cyanobacteria throughout the Phanerozoic based on models I and II, respectively. (D) Observed occurrences of calcified marine cyanobacteria on the basis of reports in 15 sedimentological-paleontological journals (29). The number of reported occurrences in each chronostratigraphic stage has been normalized to 10-Ma intervals. The three arrows indicate gaps in Phanerozoic occurrences.

The resulting Ca2+ minimum concentrations are compared with the Ca2+ concentrations derived from model calculations by Stanley and Hardie (2) (Fig. 3A), which are based on steady-state mixing of river water and mid-ocean ridge hydrothermal brines coupled with precipitation of solid CaCO3 and SiO2 phases. The more the absolute Ca2+concentrations exceed these Ca2+ minimum estimates, the higher the likelihood that calcified cyanobacteria formed in the corresponding ocean (Fig. 3, B and C). The fossil record of calcified cyanobacteria from open-marine settings should reflect this relation (29) (Fig. 3D). Although the predicted and observed abundances of occurrences do not exactly coincide at the 10-Ma intervals, their rough tendencies do (Fig. 3). Above all, the predicted gaps in the Lower Permian, Lower Triassic, and Lower Jurassic apparently correspond to the Asselian, Scythian, and Toarcian, which lack marine calcified cyanobacteria, although stromatolites occur.

The most notable discrepancy exists with regard to the Cretaceous and Paleogene. Very high Ca2+ concentrations are suggested by the Ca2+ model curve of Stanley and Hardie (2), but fewer and fewer occurrences of calcified cyanobacteria are known from the Berriasian to Santonian (Fig. 3). However, biogenic Ca2+ removal is not included in the Hardie-model calculation, and the successive vanishing of calcified cyanobacteria may point to a substantial drawdown of Ca2+ by the calcareous nanoplankton and planktonic foraminifera since the Late Jurassic (30). This is consistent with the volume of Upper Cretaceous–Early Paleocene chalk deposits of coccolithophorids. In the first instance, these sediments reflect high Ca2+ flux rates rather than high absolute Ca2+ concentrations.

Precambrian marine carbonate stromatolites almost completely lack calcified cyanobacteria (8, 9), although cyanobacterial remains comparable to those of recent taxa were preserved upon early silification (10). This lack appears astonishing because most stromatolites, at least those of the Proterozoic, are considered as mineralized cyanobacteria-dominated biofilms (9). However, this lack can be explained by high DIC concentrations (Fig. 2). Regardless of whether high-pCO2 conditions (>10−2 atm) with a slightly acidic to near-neutral ocean (31) or low-pCO2 conditions (10−3.5 atm) with an alkaline “soda ocean” of 5 to 500 meq liter−1carbonate alkalinity (20) are assumed, cyanobacteria could not have induced sheath calcification by photosynthetic carbon assimilation (Figs. 1 and 2). Thus, episodic Precambrian stromatolite formation should reflect exopolymer-mediated precipitation driven by physicochemical fluctuations of the whole-ocean basins. This is consistent with correlation of Proterozoic stromatolite morphotypes between different continents (32) and with the fractal geometry shown by some Archean stromatolites (33). Consequently, discoveries of 700- to 750-million-year-old calcified cyanobacteria (34) that unequivocally resulted from micritic sheath impregnation would set an upper limit of thepCO2 at 10−2 atm and would be consistent with a decrease in atmospheric pCO2during the Late Precambrian glaciations.

Several elements of uncertainty remain. The Ca2+ curve of the Hardie model shifts considerably in response to changes in the mid-ocean ridge hydrothermal brine/river water ratio, although the overall shape of the curve does not change (1,2). Nevertheless, correlation with the occurrences of calcified cyanobacteria indicate that the order of magnitude of Ca2+calculated by Stanley and Hardie (2) is, in principle, correct. Fluctuating levels of seawater CaCO3supersaturation during the Phanerozoic could be expected (20). However, the effect on the Ca2+ minimum curve is minor, because lowering of the initial SICc of 1.0 results in higher ΔSICc values for the settings that first show calcified cyanobacteria. Only the effect of varying temperature, Mg2+, and SO4 2−concentrations (35) on the ΔSICc will lead to a further modification of the Ca2+ minimum curve. Abundances of calcified cyanobacteria may also be influenced by the distribution of carbonate platforms or alkalinity plumes from anoxic basins (20). Nonetheless, the presence of calcified cyanobacteria sets a lower limit of Ca2+ concentration for ancient oceans, provided that hydrospheric-atmosphericpCO2 is known. Furthermore, the effect of carbon fixation by chemolithotrophic bacteria on carbonate precipitation should have been more significant than today during times of high seawater Ca2+. This may explain the occurrence of calcareous microbial fossils such as Renalcis in reef cavities, and the great abundance of microbialites during large parts of the Phanerozoic. The recently discovered calcitic microbialites containing Epiphyton- and Girvanella-like structures in a hardwater lake (36) may therefore indeed serve as an analog for microbial reefs formed during high-Ca2+ times of the Paleozoic.

  • * To whom correspondence should be addressed. E-mail: garp{at}gwdg.de

REFERENCES AND NOTES

View Abstract

Navigate This Article