If the term ‘coral’ is broadly used, paradoxically it has no valid taxonomic definition. ‘Coral’ initially referred to the Mediterranean coral, Corallium rubrum (Anthozoa: Octocorallia) with a calcitic skeleton. Afterwards, it was used to name other Octocorallians, like the blue coral (Heliopora), but also Hexacorallians, like the black coral with horny skeleton, Antipathes, or reef-building corals (Scleractinians or Madreporaria) with aragonitic skeleton, or even Hydrozoans like the fire coral (Millepora). To throw people into confusion, ‘coral’ also designates Octocorallia without hard skeleton, the Alcyonarians (soft corals!). Scleractinian corals are known for their skeleton and are, in this case, called ‘reef-building corals’ or hermatypic corals (herm = reef). Hermatypic corals host in their tissues unicellular Dinoflagellates symbionts, commonly called zooxanthellae (Symbiodinium sp.) [76]. But all Scleractinians are not reef builders (ahermatypic corals), even if they are characterized by their aragonitic skeleton.

The coral skeleton is extracellular, located at the base of coral tissue, like a finger covered by a glove. Hermatypic corals are modular animals, constituted of similar units, the polyps resulting from a clonal reproduction (asexual reproduction by budding). Each polyp looks like a bag whose walls are made by two layers of cells, an ectodermal layer, and an endodermal layer, separated by a network of collagen, called mesoglea. The endodermal cell layer delimits a cavity, the coelenteron, with only one opening, the mouth, surrounded by tentacles (of a multiple of 6 = Hexacorallia). This cavity acts simultaneously as a gastric and circulatory cavity between polyps, and is therefore also called gastro-vascular cavity.

Polyps are linked together by the coenosarc made by an oral (facing sea water) and an aboral (facing skeleton) epithelium, each of them being composed of an ectodermal and an endodermal cell layer (Fig. 1). Ectoderm and endoderm are differentiated between oral and aboral layers: most of zooxanthellae are located within the oral endoderm, whereas only a few are found in the aboral endoderm. The oral ectoderm is thick (10 to 20 μm) and rich in cnidocytes, whereas the aboral ectoderm, also called calicoblastic epithelium, is flat (0.5 to 3 μm) and does not possess cnidocytes. Calicoblastic epithelium is responsible for the formation of the aragonitic skeleton. Calicoblastic cells are long (10 to 100 μm), highly interdigitated and overlap each other. They contain numerous mitochondria, suggesting an active metabolic role. Skeletogenic cells are connected by desmosome junctions [48, Tambutté and Allemand, unpublished data]. They are attached to the exoskeleton by specialized cells, called desmocytes [62]. Cells are often separated by large intercellular spaces that may contain circular vesicles (50–70 nm in diameter) thought to function in exocytotic transport of organic matrix [48]. Consequently, the skeletogenic process occurs in a ‘biologically-controlled medium’ and is separated from the external seawater by four layers of cells, i.e. at least 40–50 μm thick.

Composition of biominerals includes two fractions, one mineral and one organic, called organic matrix, the amount of each varying among species. In Scleractinian corals, skeleton is made of calcium carbonate (CaCO₃) crystallized in aragonite (orthorhombic system). In order to build their skeleton, corals have to supply calcium and inorganic carbon from ambient seawater to the calcification site but also to eliminate protons resulting from the mineralising process: Epithelial transport of molecules can be achieved by two mechanisms: a paracellular pathway, between cells (in this case, transport of the molecule is only driven by its chemical gradient, transport is diffusional and passive), or a transcellular pathway through cells. In this last case, the molecule needs to enter the cell, cross the cell and then exit. Thus the transport is active, against a gradient with energy supply. Transport of charged species across the hydrophobic cell membrane requires specific carrier proteins.

Study of coral calcification was for a long time limited due to methodological problems linked with the use of radioisotopes, particularly ⁴⁵Ca [4]. Before 1990, most of coral scientists made their experiences with freshly-cut branches of corals (like Acropora or Stylophora), leading to a naked skeleton whose porosity facilitated the unspecific adsorption of ⁴⁵Ca. Furthermore, the presence of a large volume inside the coelenteric cavity led to errors, because the amount of radioactive calcium in this cavity was about 53% of total amount absorbed by the colony after 30 min of incubation with ⁴⁵Ca-labelled seawater [83]. Therefore, the lack of rinsing this cavity induced an important error. In order to cope with this problem, researchers from the ‘Centre scientifique de Monaco’ have set up two important technological improvements: (i) a specific preparation of a 1-cm-long coral colony (fragment of branched coral like Stylophora or Acropora, or isolated polyps like Galaxea), totally covered by coral tissue and called microcolony [3], [83] and [84], (ii) a protocol adapted to rinse the coelenteric cavity (half-life time of filling, 4 min; [83]).

Major calcifying organisms are also phototrophs, either directly (coccolithophorids) or indirectly through symbiosis (foraminifera, reef-building corals). All these organisms are responsible for the major part of the production of CaCO₃ of the earth. In these organisms, calcification is highly dependent on light, and values of light-enhanced calcification up to 127 have been reported, even if the mean stimulation value is around 3 (review by [36]). However, mechanisms underlying these interactions are not known and even the terminology is discussed between light-enhanced or dark-repressed calcification [54] (see discussion in [36]). Several mechanisms have been hypothesized to explain these interactions: •

modification of the DIC equilibrium within coral tissues caused by CO₂ uptake for photosynthesis [40] and [58];

An alternative mechanism may involve photosynthesis-induced OH^– secretion within the coelenteric cavity (Fig. 3C). This light-dependent OH^– secretion may neutralize H⁺ produced by CaCO₃ precipitation, thus allowing calcification to proceed. The net rate of OH^– efflux under saturating irradiance is 3.55 nequiv min^–1 mg^–1 _protein [33], while calcification in the scleractinian coral Stylophora pistillata occurs at a rate of 0.98 nequiv min^–1 mg^–1 _protein [85], i.e. at a rate 3.6-fold higher. Notwithstanding the fact that such a comparison between two phylogenetically distinct organisms may not be valid, it can be tentatively suggested that OH^– efflux may be sufficient to buffer calcification-induced H⁺ production. In this way, calcification and photosynthesis may proceed independently, which is in agreement with data showing that calcification may be inhibited without altering photosynthesis [1] and [96]. In the dark, acidification of the coelenteric cavity caused by increased CO₂ together with low O₂ tension [42] and reduced ATP, causes inhibition of calcification. In the light, high O₂ tension increases the respiratory rate and thus CO₂ production as a substrate for calcification and production of ATP for energy-dependent carriers, in conjunction with high pH within the coelenteric cavity, which helps to neutralize H⁺ produced by calcification.

Another hypothesis has been recently suggested by the group of J.-P. Cuif [20] and [38]. By analysing the composition in amino acids of OM from 13 symbiotic and 11 non-symbiotic corals, they showed that this composition is dependent upon the presence of symbionts. If aspartic acid remains the major amino acid in both groups of corals, the amount of glutamic acid is always higher in symbiotic vs. non-symbiotic corals. Similarly, threonine is high in symbiotic corals and low in non-symbiotic ones, while the opposite is observed for threonine. OM content of sugars (mannose, galactose, galactosamine, glucosamine) is also significantly different in the two groups of corals. All these results suggest that symbionts may affect OM precursor synthesis, or may provide these precursors. This hypothesis is not exclusive with that of DIC equilibrium.

One goal of coral-reef ecologists is to predict the effects of environmental conditions (natural or anthropogenic) on the integrity of scleractinian coral communities. It is now well established that a change in most of the environmental parameters such as light, temperature, CO₂ partial pressure (p _CO2 ) and nutrients directly affects coral growth rates and thus coral calcification. It has become patently clear that coral reefs around the world are coming under increasing pressures and their status is declining rapidly [92]. One prominent example of stress is the increased frequency of mass coral bleaching events (loss of chlorophyll, zooxanthellae and reduced calcification) since the early 1980s [43]. Corals are subject to stresses at the local, regional, and global scales. At the local and regional scale, reefs deteriorate in all areas where human activities are concentrated, due to an increase in seawater pollution (heavy metals, eutrophication, bacterial and viral diseases), and human pressures (diving, fishing, tourism...). In addition to these stresses, corals also experience the effect of the ‘global change’ [45], i.e. increased seawater temperature, ultra-violet radiations, and p _CO2 as well as increased sediment loading.

Therefore, major actions were undertaken to better understand the effects of a change in the main environmental parameters on the survival of reef corals, and especially the effects on their rates of calcification. Effects of eutrophication, of increased p _CO2 levels and seawater temperature on coral calcification have attracted scientists’ attention. Eutrophication corresponds to an increase in nutrient concentrations, in sediment load and in a development of algae [50], all of them reducing light available to the corals and necessary for their calcification. Laboratory and in situ experiments have also reported a decrease in calcification under nutrient-enriched conditions. At the community level, Kinsey and Davies [49] measured a pronounced decrease (50%) in the overall calcification following an 8-month fertilization period of a patch reef system in the Great Barrier Reef. Tomascik [86] and Tomascik and Sanders [87] found a correlation between anthropogenic disturbances and decreased skeletal growth in the coral Montastraea annularis. At the organism level, several studies have monitored a lower calcification under different enrichments in ammonium, nitrate, and phosphorus [28], [29], [55] and [82], up to 50%. An iron enrichment, often seen during El Niño periods, also decreases by 20% the growth of scleractinian corals [29]. The way by which nutrients interact with calcification is still poorly known. They generally highly enhance the growth of the algal component, and this might lead to a disruption of the symbiosis, and especially the coupling between photosynthesis and calcification. Calcification is also affected by an increase in the p _CO2 , which has risen from 280 to 370 μatm since the last century, as a result of human activities and could reach 705 μatm in year 2100 [45]. An increase in seawater p _CO2 induces a change in the other carbon components such as bicarbonate (HCO₃ ^–) and carbonate (CO₃ ^2–) and corresponds to a decrease in CO₃ ^2–. Coral calcification seems however intimately linked to the calcium-carbonate saturation state (ω), which is the ratio of the ion concentration product ([Ca²⁺] × [CO₃ ^2–]) to the solubility product of the mineral deposited. Therefore, a decrease in the concentration of CO₃ ^2– in seawater has been related to a sharp decrease in coral calcification, at the organism or community level [37], [51], [53] and [56]. Together with the change in p _CO2 , the global warming also corresponds to an increase in the mean sea surface temperature, which is expected to rise by 1–3°C over the coming century [68]. This is leading to the frequent El Niño events [93]. The bleaching and death of the coral colonies after an El Niño event is generally followed by a strong coral bioerosion, reducing the 3-D reef structure [43]. When both higher level of p _CO2 and temperatures are combined, coral calcification is lowered by 50% [73].

In conclusion, coral reef ecosystems are under the threats of local direct anthropogenic stresses, and of the more extended stresses of global change, which include, in addition to a raise in temperature and p _CO2 , an elevation of the sea level and in the ultraviolet radiations.

“There has never been any doubt that corals write valuable information into their skeletons; it is their language that has remained blurry and ambiguous” [7].

Anticipation of future environmental and climatic change depends upon knowing and, hopefully, understanding what has happened in the past and is now happening. However, instrumental records of weather and climate go back only about several centuries and that time span is insufficient to judge whether recently observed rises in global temperature are normal or not. Corals can provide climate and environmental records for the world’s shallow-water tropical ocean regions, since these areas are poorly represented by other proxies (tree rings, ice cores, sediments, pollen). Moreover, massive species can provide information both for the last several centuries (from living corals) and for well-dated windows of the more distant past (from well-preserved dead and fossil corals).

Corals build their skeletons from calcium and carbonate extracted from seawater, so their skeleton contains isotopes of oxygen, carbon, as well as trace metals (Sr, Mg, U, Th...). Thus, proxy climate and environmental information are stored in coral skeletons as growth characteristics (e.g., skeletal extension, density and calcification). Examples of information stored in coral skeletons include sea-surface temperatures (SSTs), river flow, rainfall, upwelling, salinity and anthropogenic influences.

The great majority of corals used for palaeo-reconstructions are massive genus (as Porites or Diploastrea). Corals are drilled and the sampling skeleton is cut in order to obtain a slice (1 cm width). Coral skeleton is characterized by annual density bands visible on X-ray radiographies (Fig. 4). The sum of a dark (high-density) and a light (low-density) band represents one year [7]. This characteristic provides a built-in chronometer that enables first-order dating as well as identification of particular years and specific seasons. Once years are located, very small samples of skeleton are taken on a line along the coral growth axis, and isotopic composition of oxygen (δ ¹⁸O) and carbon (δ ¹³C) are analysed with a mass spectrometer.

The skeletal δ ¹⁸O varies with temperature and salinity of the water where the coral grew in. The isotopic composition is expressed in the conventional delta notation relative to a standard, which is PDB for carbonates: However, the promise of coral density bands to yield similar fabulous records to those obtained from tree rings has not been yet realized despite extensive works. One problem is that such proxies need to be calibrated before their use. The great majority of calibrations have been carried out in the field (see for example [17] and [89]); however, calibrations carried out in the laboratory under controlled conditions seem necessary to decipher the effect of each environmental parameter (temperature, salinity, light...). Such an approach has been used successfully with foraminifera (e.g., [81]) but difficulties encountered with culture techniques have precluded the development of experimental calibrations of proxy records in corals. However, recently several works have been done concerning the effects of several parameters (temperature, light, nutrition, p _CO2 ) on coral isotopic composition [70], [71] and [72].

Recently, it has been shown that isotopic record is profoundly influenced by biology (so-called ‘Vital Effect’): different coral colonies cultured in the laboratory under exactly the same temperature conditions have yielded δ ¹⁸O measurements that present a huge variability (2‰) (Fig. 5, Stylophora pistillata). However, when δ ¹⁸O measurements are done on nubbins from the same parent colony, there is no such variability (Fig. 5, Acropora sp.). This suggests that a calibration curve is specific for each colony considered.

We thus concluded that the δ ¹⁸O recorded in coral skeleton is affected not only by temperature and salinity of the seawater, but also by the biological activity. This last point must to be taken into account in future palaeo-reconstructions.

This review highlights the complexity, but also the weakness of our knowledge concerning biomineralisation mechanisms in corals. Formation of a coral skeleton is the result of complex, biologically regulated phenomenon, largely unknown. Due to the fact that coral reefs are presently disturbed by human activities, it is urgent to develop research on coral biology, associating physiologists, molecular and genome biologists, geneticists, ecologists... Beyond the benefit for our environment, a better knowledge of coral calcification will help to understand biomineralisation mechanisms by themselves, which remain poorly understood, even in mammals: “Bone physiology is undoubtedly the source of many concepts of biomineralization, but it remains as one of the few physiological processes for which there is a unified theory and one of the few medical interests that have not benefited from comparative studies” [78, page 292].