This review focuses on bacteria and their role in mineral formation. As a consequence of their small size and diverse metabolic capabilities bacteria, more than any other type of living organism, are able to interact intimately with metal ions present in their environment. Some metals are required for metabolism and are taken into the cell through various mechanisms, then incorporated into the necessary physiological pathways and biosynthetic structures. This physiological aspect of metal-bacterial interaction will not be discussed but, rather, the ability of bacteria to accumulate metal ions and incorporate them into mineral phases will be described. This activity has widespread importance for the shaping of our planet and the recycling of mineral elements. Since bacteria are most frequently found as part of microbial communities it is within this context that their mineral-forming ability will be discussed.
In almost any environmental sample, examination of the cells by electron microscopy reveals mineral precipitates closely associated with bacterial cells (Fig. 1). There are several characteristics which bacteria exhibit that make them ideal nucleating agents for mineral precipitation. Bacterial cells are very small, especially in low-nutrient environments. Typically, a single rod-shaped bacterium will have a diameter of approximately 0.5 μm and a length of 1 μm . Due to their small size, bacteria as a group have the highest surface area-to-volume ratio of any group of living organisms and this, together with the presence of charged chemical groups on their cell surface, is responsible for the potent mineral-nucleating ability of these cells. Eubacterial cell walls come in two main formats, Gram-positive or Gram-negative . Either type may be overlain by a number of other surface structures. These may be proteinaceous in nature (e.g., S-layers) or composed primarily of carbohydrate polymers (e.g., capsules) and may occur singly or in combination; a more detailed description of these structures can be found in  and . Whatever types of cell surface structure the cell may have, the main charged chemical constituents found in these structures at neutral pH are carboxyl, phosphoryl, and amino groups. In general, negatively charged groups dominate over positively charged ones, giving the cell surface an overall anionic charge .
Initiation of mineral formation on bacterial surfaces has been proposed to follow a generalised pattern which can be thought of as occurring in two steps . In the first step, metal ions present in the aqueous surroundings of the cell interact with charged groups in the surface structures. The interaction is stoichiometric such that there is an electrostatic charge complementation between the charged groups in cellular polymers and the metal ions. Subsequently, the presence of bound metal ions in the wall fabric lowers the total free energy of the system, thereby initiating further metal deposition. In this case, precipitates form at the nucleation site between metal ions and excess counter ions from the fluid phase; the wall binds more metal than would have been expected based solely upon charge interactions with wall polymers. Thus, metal aggregates are formed within the wall matrix, their size constrained by the physical presence of the polymer meshwork itself. Depending on microenvironmental geochemistry, negatively charged counterions (e.g., sulfate, phosphate, carbonate, sulfide, or silicate ions) determine specific mineral phases . Mineral formation on the bacteria is generally not controlled by the organism; it happens because of the physicochemistry of the bacterial surface and the chemistry of the cell's environment. Actively metabolising bacteria with highly energised plasma membranes can inhibit mineral formation since the cell wall is flooded with protons which compete with metal cations .
The following sections will describe several different types of environments in which bacterially mediated mineral formation has been suggested to play a major role in formative geological processes. These by no means represent an exhaustive survey of the many descriptions of in situ bacterial mineral formation but are meant to serve as representative examples of the types of minerals that have most often been found in association with bacterial cells.
2Microbial ecology of sulfidic mine tailings
An example of an environment in which the links between geochemistry, microbial physiology, and mineralogy can be seen is in sulfidic mine tailings. These interactions have recently been documented in detailed reports of geomicrobiological studies of abandoned and active mine tailings dumps in northern Ontario [6–9]. The exploitation of base metal (Cu, Zn, Fe, Ni, etc.) ore deposits has led to extensive environmental contamination in the areas surrounding and including the tailings dumps from such mines. In general, pyrites (metal sulfide minerals) are the most common form mined for metal extraction. After extraction the finely ground waste rock is mixed with water and deposited as a slurry (tailings) in large impoundments, often nearby lakes. The tailings still contain a significant amount of sulfidic minerals which, together with Fe (and to a lesser extent other metals), represent an oxidisable energy source for a large group of bacteria collectively termed acidophilic lithotrophs . This group is exemplified by the thiobacilli, a group of Gram-negative rod-shaped bacteria which are autotrophic and can oxidise both ferrous Fe and sulfide to produce ferric iron and sulfuric acid, often with pHs as low as 1–2 . Abiotic acidification can also occur but is a much slower process than the bacterially mediated one . This ‘acid mine drainage’ can enter groundwater, nearby rivers, and lakes. Many abandoned mine tailings dumps exist and represent an enormous environmental and economic problem. An understanding of the microbiology of these environments is the key to developing long-term, ideally self-sustaining bioremediation tactics.
Detailed studies [7, 8] have unravelled some of the relationships among the major groups of microorganisms present in tailings from sulfidic base metal mines. Samples were taken at 5-cm intervals with depth (down a vertical tailings profile) to learn about the geomicrobiological inter-relationships that existed in this environment. The samples consisted of sediment with little free water and were analysed for bacterial numbers, mineralogy (both bulk and cell-associated), dissolved metals and sulfate, and pH. Two groups of bacteria, the thiobacilli and the sulfate-reducing bacteria (SRB), were enumerated by the most probable number method (MPN) and examined for their ability to affect porewater geochemistry and precipitate mineral phases. Even though most tailings are pyritic, no previous studies had included investigations of SRB activity since these bacteria require anaerobic, neutral to alkaline environments and were not thought to be significantly represented in oxic, metal-rich, acidic tailings.
A dynamic relationship exists between the thiobacilli and the SRB [7, 8]. In the tailings, a colour change from orange to grey with depth roughly corresponded to a change from oxic to anoxic conditions. Even in the orange oxic zone, where thiobacilli fix inorganic carbon and oxidise Fe and sulfide from the waste rock, significant numbers of SRB were found. The peak numbers of thiobacilli correlated well with maximum sulfate and soluble iron in the tailings, as well as with the region of lowest pH . As the leachate percolates through the tailings some buffering occurs through interaction with carbonate minerals, and free metal concentrations in the pore water may be reduced by precipitation into minerals. This process is enhanced by the presence of bacterial cells, which provide nucleation sites for their deposition. As the tailings become anoxic with depth, generally due to saturation by groundwater, SRB numbers begin to increase . These organisms, in an energy-yielding anaerobic respiratory process, reduce sulfate to sulfide which is released from the cell . Sulfide reacts strongly with metal ions, resulting in the formation of sulfide precipitates both on the SRB cell surface and throughout the immediate surroundings .
Although it is probable that autotrophic SRB exist, these bacteria are generally considered to require organic molecules (usually simple organic acids) as carbon sources and electron donors for sulfate reduction . The main source of organic nutrients in the tailings is that contained in rainfall and derived from the degradation of microbial cells. Thus, thiobacilli and other bacteria in the oxic zone represent an important source of organic carbon in this environment [7, 8]. SRB can be found in lower numbers in the oxic zone they must survive within anaerobic, circumneutral pH microenvironments, likely maintained through bicarbonate production.
Several types of secondary minerals have been found associated with bacterial cells in tailings and it is believed that these would not have formed without the presence of the bacteria . As in other environments, the cells provide nucleation sites for mineral development and also affect microenvironmental chemistry such that particular (and otherwise unexpected) mineral types form. In the oxic zone, the predominant mineral type associated with thiobacilli was Fe-oxides , while on and near the SRB cells in the oxygen-depleted zone Fe monosulfides were formed . The large amount of Fe and sulfate released by the microbial activity in the oxic zone also led to the precipitation of melanterite (FeSO4·7H2O), a mineral form observed throughout the tailings profile and not associated with cell surfaces . In other tailings environments  the metal content of re-precipitated minerals reflected the original heavy metal content of the tailings. However, the mineral form was unlike that of unaltered tailings; it included minerals containing phosphorus presumably derived from the accompanying bacterial cells. These studies serve to highlight the profound effect bacteria can have on the geochemical processes occurring in a tailings environment.
3Silicate formation by bacterial cells
In many diverse environments, including river sediments [12, 13], mine tailings , hot springs [14, 15], and surface sulfide springs (Douglas, unpublished observation), silicate minerals, often of a clay-like composition and structure, are found in intimate association with bacteria (see Fig. 1). It is quite likely that the bacteria were instrumental in their formation, rather than simply binding pre-formed detrital minerals, since the composition of silicates clearly not associated with bacterial cells often differs markedly from those found on the cell. In addition, these precipitates are more crystalline and larger in structure than those found on bacterial cell surfaces. Even in environments with low dissolved silica concentrations, such as freshwater lakes and rivers, clay-like precipitates are found on bacterial cells. The metal content of the clays is usually reflective of the metals present in the surrounding water .
In some cases, exemplified by microbial communities in silica hot springs, ‘pure’ silica (i.e., with no associated metal ions) is found to form around bacterial cells, eventually encasing them in an amorphous silica matrix; a process that may be analogous to that which led to the preservation of bacteria as microfossils in the geological record [16, 17]. In a study of hot spring microbial mats in Iceland, it was found that, with depth in the mat, bacterial cells became progressively more mineralised so that eventually all that remained was the bacterial surface layers encased in amorphous silica . Even though hot spring waters are supersaturated with respect to silica, silicate minerals will not precipitate unless a nucleation surface, such as that presented by a bacterial cell, is present .
Laboratory simulations of microfossil formation have indicated that cells exposed to heavy metal ions such as Fe before being subjected to high silica concentrations with or without raised temperatures (60°C) for extended periods of time were better preserved and remained as recognisable cells . It was concluded that pre-treatment of cells with Fe inhibited the activity of autolytic enzymes, thus preventing degradation of cellular structures prior to their preservation [16, 18]. In the light of these results, it is remarkable that recognisable sheath structures remained in the Iceland silica matrix. No Fe or other metal ions were detected in these silicates, indicating that, if present, the metal concentration was below the detection limits of the energy-dispersive X-ray spectrometer used in the analysis of the precipitates (i.e., <1% by weight). At another site in Iceland, Fe silicates were found as the cell-enclosing matrix and it was suggested that, due to the presence of Fe, it is plausible that these cells may eventually be preserved in the geological record as microfossils .
Recently, it has been shown that bacterial surfaces are also good sorption interfaces for silicate ions directly. Although the cell surface has a net electronegative charge, some positive amine groups are present within the wall matrix. At near-neutral pH, these represent sites for interaction with silicate anions yet, there are not enough available amine groups to account for the large amount of silicate which is eventually bound. Additional silicate binding occurs as a result of crossbridging involving metal ions which link Si anions to carboxyl or phosphoryl groups within the wall matrix . Thus, negatively charged polymeric groups can also be involved in silicate deposition. Binding of silicate anions leads to the deposition of poorly ordered silicate mineral phases in association with the bacterial cell surface. These eventually become crystalline and specific clay phases can be formed. As these silicate minerals develop they also act as sorption interfaces for metal ions, increasing the overall metal binding ability of the bacteria [20, 21]. However, the bacterial cell wall appears to have a greater affinity for metal ions than the clay phases and, on a per weight basis, has a higher metal binding capacity [22–24].
4Mineral formation in microbial mats or biofilms
Carbonaceous organosedimentary structures known collectively as microbialites provide the best studied examples of mineralised microbial communities. Such structures have been preserved in the geological record, providing information of past geomicrobiological activity and environmental conditions. Extant microbialites are presently still forming in certain environments, usually warm shallow marine waters which provide adequate light and shelter from disruptive physical forces and protection from grazing invertebrates (e.g., molluscs).
The formative processes responsible for the presence of microbialites have been under debate for some time but a consensus appears to be forming that microorganisms (particularly bacteria and cyanobacteria) are necessary for their deposition [24, 25]. These structures basically consist of a microbial community, made up of a dominant type of oxygenic phototroph and associated microorganisms which have fine-grained carbonate precipitates (usually calcite or aragonite) associated with them. The formation of these minerals may be promoted by physiologically induced alkalisation of the microenvironment around the cyanobacterial cells or may be due to abiotic geochemical factors such as evaporation from the calcite-supersaturated aqueous environment within which these structures are usually found [24, 26]. It is probably a combination of both, with each type predominating at particular times, depending on seasonal environmental variations. Whatever the mechanism of CaCO3 formation may be, it is clear that the presence of bacterial cells is necessary for mineralisation. Again, the bacteria provide nucleation sites for mineral deposition. It has also been suggested that the often extensive gelatinous sheaths of many unicellular and filamentous cyanobacteria are important for trapping of abiotically formed precipitates [27, 28].
Microbialites are generally classified as one of two major structural types as described by Kennard and James . Stromatolites have a laminated structure in cross section, with light-coloured carbonate-rich bands alternating with darker organic-dominated layers on scales of millimetres. Over time, cementation processes result in a hardening of the layers which are no longer actively forming . In general, the structure of stromatolitic microbialites arises from the presence of filamentous cyanobacteria (as opposed to unicellular forms) as the main structure-determining factor of the mat . The second structural type is commonly referred to as a thrombolitic microbialite and has an open clotted texture which appears to be made up of many small globular structures cemented together. Thrombolites appear to be formed due to the presence of unicellular cyanobacteria as the main structural entity .
Different forms of carbonaceous microbialites have been found in widespread and often fascinating environments. The largest known calcareous microbialites were formed in the late Precambrian. However, extremely large, presently forming microbialites were recently discovered in Lake Van, Turkey . This is a large, deep alkaline (pH>9.7) lake with microbialitic tower-like structures growing up to 40 m high to depths of 100 m in a dense population of structures. The columns are covered in a dark green to black microbial mat comprised of coccoid cyanobacteria as the major phototroph with the formation of aragonite occurring within their sheaths. The columns are porous, with Ca-rich groundwater percolating up through them so that, at the ends of pinnacles where the ground water exits the column structure, abiotic calcite is deposited. It is not known why the difference in mineral phases between biogenic and abiotic calcium carbonate exists. However, the authors speculated  that somehow the organic matrix provided by the cyanobacteria directs the formation of aragonite rather than calcite.
Stromatolite-like structures have also been found in rather unlikely environments such as caves, where laminated structures (‘speleotherms’) are being deposited by a cyanobacterially dominated microbial community . Stromatolite-like structures have even been found in deserts where coccoid cyanobacteria belonging to the pleurocapsa group are responsible for formation of ‘crusts’ consisting of calcified laminated layers over desert soils in California (Borrego Desert) and the Sinai desert in Israel .
5Possible mechanisms for carbonate formation
A general mechanism for carbonate mineral deposition by bacteria involves the ability of the cells to produce an alkaline microenvironment as a result of their physiological activity. Organisms capable of doing this include SRB, which release bicarbonate , nitrate reducers (releasing ammonium ions), and urea-degrading bacteria (also release ammonium ions ). By far the most prevalent reaction is that of oxygenic photosynthetic microorganisms, principally cyanobacteria, which release hydroxyl ions as a result of using bicarbonate ions as a carbon source in the aqueous neutral to alkaline environments they inhabit .
5.1Fayetteville Green Lake
The detailed mechanism of carbonate deposition on a cyanobacterial cell was outlined for Synechococcus GL24 (Fig. 2), a unicellular cyanobacterium isolated from Fayetteville Green Lake, New York . In the lake, Synechococcus cells represent the dominant phytoplankton due to the highly oligotrophic nature of this small, but deep (55 m), meromictic lake . The lake water has high levels of Ca (10 mM), SO2−4 (11 mM) and carbonate (∼10 mM) ions in its hypolimnion . Synechococcus lives as a planktonic population, the numbers of which follow a seasonal progression with a maximum of 107 cells ml−1 in summer to 104 cells ml−1 under ice cover in winter . They also exist as a benthic population where they are the main formative agent of a thrombolitic bioherm on the steep sides of the lake basin and on almost every solid surface (sticks, rocks, etc.) within the photic zone. The bioherm is formed by similar processes as described for mineralisation in the water column but, since these cells are growing as biofilms on the lake shore, their calcification and eventual entombment leads to an outbuilding of the bioherm structure .
Each Synechococcus cell is surrounded by a paracrystalline surface array as its outermost, defining layer (Fig. 3). These are structures made of many identical protein molecules which self-assemble on the bacterial surface to form a structure that completely encloses the bacterium and that, when viewed by electron microscopy, has a regularly symmetrical pattern to it. The S-layer of Synechococcus GL24 is hexagonally symmetrical with diamond-shaped pores 11×22 nm. The S-layer plays an important role in the ability of this cyanobacterium to precipitate gypsum (CaSO4·nH2O) and calcite (CaCO3) from the lake water (Fig. 4).
The S-layer is more hydrophobic than most bacterial surfaces but negatively charged sites which can bind Ca2+ are present in the pores . Once Ca is bound, it complexes SO2−4 ions when microenvironmental pH is not much higher than that of bulk lake water (pH 7.9), or CO2−3, when bicarbonate metabolism pushes the microenvironmental pH above 8.3. For the latter, HCO−3 is taken into the cell and converted by carbonic anhydrase to CO2 and OH−. The CO2 is incorporated into cell biomass while OH− ions are released into the cell's microenvironment  and concentrated around the cell. As a consequence, the microenvironmental pH around a Synechococcus cell can become highly alkaline when the cells are active. When they are grown to high densities (106 cells ml−1) in test tubes containing natural lake water (total volume 10 ml) the pH can increase to 10.5 within 48 h . In Fayetteville Green Lake, as the lake water warms in mid-summer, it takes on a milky appearance due to the presence of large numbers of suspended Synechococcus cells, all surrounded by calcite minerals and existing as cell-mineral aggregates. These eventually sink to the lake bottom, forming its extensive marl sediment.
In the laboratory Synechococcus cells suspended in Fayetteville Green Lake water can become totally encrusted in gypsum or calcite within 8 h, yet these bacteria divide only once within 72 h . To prevent their total encasement in mineral, and death, they continuously shed off patches of mineralised S-layer which is rapidly replaced by new material. Eventually, as cells enter senescence, this shedding ceases, and the bacteria become surrounded by a crust of calcite, die, and sink to the bottom as a fine-grained calcitic sediment particle.
6Precipitation of minerals other than carbonates in natural microbial communities
Stromatolitic structures involving non-carbonaceous mineral types have also been observed and postulated to have a microbially catalysed deposition mechanism. Phosphorites are P-rich sedimentary rocks, usually composed of carbonate hydroxyl fluorapatite (Ca10(PO4CO3)6F2–3) which occurs as nodules and crusts originally formed in oceanic environments . The nodular and stromatolitic forms of these mineral deposits could be of microbial origin. Southgate  examined columnar phoscretes in Australia in an attempt to discern their formation mechanism. In these laminated structures, a ubiquitous association between the phosphate and organic matter was found. It was concluded that phosphates may preferentially nucleate upon the ‘organic matter’, which would provide nucleation sites for its deposition. Soudry and Champtier  interpreted P-coated structures seen by scanning electron microscopy to be cyanobacterial sheaths due to their size and morphology. A similar study  suggested these structures may represent fungal hyphae coated with phosphoritic minerals.
In addition to the minerals described above, organosedimentary structures of microbial origin have also been associated with dolomite , gypsum , and deposits of elemental metals such as Cu  and Au . It is interesting that bacteria can form placer gold during laboratory simulations using aqueous solutions of gold chloride . In all cases, bacteria were implicated as the ‘concentrating factor’ for the metals and it is likely that, as more investigations are undertaken, their role will become clearer and the phenomenon more widely seen.
The preceding discussion reveals that bacteria have played and are continuing to play a determinative role in the mineralogical characteristics of most soil and sediment environments. As the number and variety of geomicrobiological studies continue to grow it is hoped that the complex geochemical role of bacteria will become clearer. Many essential elements (C, S, N, P) have a large lithospheric reservoir within which they are largely unavailable to many living organisms. Bacteria can liberate such elements by processes such as weathering as well as reformulate them into a wide range of minerals.
The ability of bacteria to accumulate metal ions has led to speculation that these organisms represent an important cleansing mechanism in natural environments. The immobilisation of metal ions and their subsequent transport into aquatic sediments or deep soil layers represents an important means of lowering the ‘free’ metal concentration in environments threatened by the presence of toxic metals (e.g., Cd, Ni, Cu). Our understanding of the mechanism and extent of bacterial metal binding activities can help us to learn how much of a heavy metal load a natural environment can take without breaking down. In addition, bacteria may represent a potentially significant tool in bioremediation strategies.
T.J.B.'s research has been funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) through operating and infrastructure grants. The latter provided partial support for the NSERC Regional STEM Facility where all the authors’ electron microscopical work was performed.