The past decade has seen a rekindling of interest in microorganisms isolated from ancient evaporites. From a practical point of view, there is concern that viable microbial communities in salt deposits could lead to biodeterioration of containers holding toxic and nuclear waste. Additionally, many studies have focused on understanding the origins of halophilic microorganisms from rock salt, in particular the haloarchaea. A most intriguing question is whether these halophiles are the descendants of populations that became trapped in salt when the evaporites formed millions of years ago. If this was the case, then we need to explain how they have been surviving or even thriving in their salty subterranean world. The discovery of life in evaporites has, in turn, resulted in similar environments on Mars and Europa being considered as targets in which to seek extraterrestrial life. In this review, we consider the evidence for and against long-term survival of halophilic microorganisms in ancient salt deposits.
Hypersaline environments and their inhabitants
Although NaCl is the main salt in nearly all hypersaline waters, the ionic composition of inland saline lakes can vary considerably depending largely on the surrounding geology. For example, the Dead Sea is enriched in Mg2+ and Ca2+, which in turn influences the microbial flora ( Oren, 1988). The salinity of sea water varies slightly across the globe (e.g. 34–36 g l−1 in the Pacific), and its composition appears to have been stable over 2 billion years, although this is still a matter of debate ( Schreiber, 1986). Most hypersaline environments are derived from the evaporation of sea water, encouraged by restricted flow, high temperatures, low rainfall, low humidity and high wind speed. It takes ≈ 1000 m depth of sea water (35 g l−1) to produce 14 m of evaporites, most of which would be halite ( Schreiber, 1986). Given that some halite beds are several hundred metres thick and that about one-quarter of the land mass is underlain by evaporites ( Blatt et al., 1980 ), it is clear that hypersaline environments have been significant reservoirs for the long-term evolution of specifically adapted microorganisms. During certain geological periods, such as the Permian, extremely saline seas were much more extensive than today ( Javor, 1989).
Organisms from the three domains of life have adapted to grow in high-salt environments but, as the concentration of salt increases, the overall diversity of physiological groups decreases ( Oren, 1999). Some physiological groups, such as nitrifiers, appear to be absent at salt concentrations above 15% (w/v), probably because their dissimilatory metabolism results in only a small amount of energy, and the extra energetic costs imposed by having to produce compatible solutes makes the process unfavourable ( Oren, 1999). This, however, does not mean that hypersaline environments are nutrient poor. In most cases, quite the opposite is true – evaporation results in the concentration of nutrients as well as salts. Sonnenfeld (1984) cites many examples of increased organic carbon, productivity and heterotrophic activity in hypersaline lagoons.
Haloarchaea and halite precipitation
The haloarchaea (order Halobacteriales, also called halobacteria) have typical archaeal characteristics such as ether-linked lipids. They have an obligate requirement for high salt concentrations and are generally unable to grow in less than 10% (w/v) salt. The osmotic pressure of surrounding brines is balanced by the accumulation of K+ ions inside their cells. As a consequence, the biochemistry of haloarchaea is specifically adapted and reliant on high salt concentrations ( Oren, 1999). To date, there are 14 genera of haloarchaea ( Fig. 1), reflecting their considerable ecophysiological diversity. For example, some species require high pHs for growth, whereas others need high concentrations of magnesium, largely reflecting adaptation to the environment in which they live ( Oren, 1994).
From about 20% (w/v) NaCl concentration up to halite saturation [≈ 32% (w/v)], haloarchaea become the dominant microorganisms ( Benlloch et al., 1996 ). They can occur at such high cell densities that they cause brines to turn red, which further encourages evaporation by trapping solar radiation. This red colour is caused by carotenoid pigments that protect cells from the harmful effects of ultraviolet light ( Shahmohammadi et al., 1998 ). As NaCl starts to precipitate, most haloarchaea become trapped inside fluid inclusions, which can constitute 2–6% (w/w) of freshly harvested solar salt ( Lefond, 1969). This phenomenon has been observed in solar salterns throughout the world ( Norton and Grant, 1988; Castanier et al., 1999 ). Moreover, the presence of haloarchaea affects crystallization in several ways: it leads to an increase in the size and number of crystals ( Lopez-Cortes et al., 1994 ); results in bigger fluid inclusions ( Norton and Grant, 1988); encourages dendritic crystals ( Norton and Grant, 1988; Lopez-Cortes et al., 1994 ); accelerates crystal formation ( Norton and Grant, 1988); and sometimes encourages the formation of salt ooids ( Castanier et al., 1999 ). Haloarchaea have been shown to be motile for several weeks and remain viable inside fluid inclusions for at least 6 months ( Norton and Grant, 1988). Bain et al. (1958) recovered viable ‘pink bacteria’, presumably haloarchaea, from solar salt that had been in storage for 4 years, while Dombrowski (1966) recultivated bacteria encased in salt after 5 years.
The role of microorganisms in the diagenesis of evaporites and their short-term survival inside salt crystals are not in question. The evidence for long-term survival in salt deposits is more difficult to assess and requires an understanding of the geology and geochemistry of evaporites as well as information about isolated microorganisms.
Isolations of microorganisms from brines in salt mines
The flora of brines found in salt deposits and mines could provide valuable information about the origins of microorganisms isolated from rock salt. This is because these brines could serve as a source of contamination for rock salt, or the microorganisms in them could have originated from the rock salt. There are several possible processes by which brines form: (i) natural recrystallization of hydrated minerals, forming residual brines; (ii) influx of surface water or ground water; (iii) condensation of air used for ventilation; (iv) locally introduced water, spills, etc.; and (v) water deliberately injected into salt horizons for extraction as brine (solution mining). The possibility of external contamination in each type of brine will differ. Namyslowski (1913) was the first to reveal that brines in Wieliczka salt mine in Poland were teeming with halophilic microorganisms, and Nehrkorn and Schwartz (1961) tentatively assigned isolates from Braunschweig mine brines to genera now known to belong to the haloarchaea. Vreeland and Huval (1991), investigating brines formed by dissolution of salt deposits by surface water, isolated many strains with a range of salt tolerances, such as haloversatile types [0– > 17% (w/v) NaCl], moderate halophiles [2–20% (w/v) NaCl] and extreme halophiles [12–32% (w/v) NaCl]. It seems unlikely that the haloversatile microorganisms are anything other than adaptable microorganisms from surface waters.
Triassic Northwich halite is exploited at Winsford salt mine (UK). In this mine, ventilation air is driven through a series of dewatering tunnels, in which NaCl-saturated brines form as a result of condensation on the cool mine walls ( Norton et al., 1993 ). Total counts in five of these brines ranged from 2 to 4 × 107 cells ml−1, but viable counts on halophile medium ( Norton and Grant, 1988) were up to 200 times lower. Given that haloarchaea normally dominate salt-saturated brines almost to the exclusion of Bacteria, it was surprising to find that halophilic Bacteria were dominant, contributing 86–100% of the colonies (T. J. McGenity and W. D. Grant, unpublished). It is probable that the low temperature (12–14°C) of Winsford salt mine enabled Bacteria to outcompete Archaea, as observed experimentally by Rodríguez-Valera et al. (1980) . Surprisingly, 47 out of 48 bacterial strains from Winsford brine pools required at least 10% (w/v) NaCl for growth, and could grow with 30% (w/v) NaCl; more than half of them required 12.5% (w/v) NaCl (T. J. McGenity and W. D. Grant, unpublished). In other words, they were extreme halophiles, a feature that has not been described previously in aerobic, heterotrophic, halophilic Bacteria (see Ventosa et al., 1998 ). These findings raise a number of interesting questions concerning the evolution of extreme halophily. How long does it take for moderate halophiles to adapt to become extreme halophiles? Do these salt mine Bacteria maintain an osmotic balance with organic solutes? Alternatively, have they evolved the highly specialized system of using K+ ions, as found in haloarchaea and Haloanaerobiales ( Oren, 1999)? Salt deposits, unlike surface brines that are periodically refreshed by rain, are permanently salt-saturated environments, in which there would be no advantage to grow at low salt concentrations. Therefore, if we assume that it takes millions of years of adaptation to become an extreme halophile, it would be more likely that the Bacteria isolated from brine pools in Winsford salt mine have been evolving inside the salt formations. However, a better understanding of the molecular basis of adaptation to high salt concentrations is needed before this assumption can be validated.
Brines in the Boulby salt mine (UK), in which Permian Zechstein salts are exploited, have diverse origins, reflected in their different chemistries. Those that are believed to be residual (G. M. Jones, personal communication) had the largest halophilic communities, up to 1.9 × 106 viable cells ml−1, all of which were haloarchaea (T. J. McGenity and W. D. Grant, unpublished). NaCl-saturated, solution-mined brine from Northwich halite supported 2.7 × 103 viable cells ml−1, and again all were haloarchaea ( Norton et al., 1993 ).
It should not be assumed that halophiles from brines in salt deposits are necessarily derived from the salt beds, and neither should the information provided by these samples be ignored when considering their origin. The presence of a large halophile community in mine brines and associated efflorescences of recrystallized salts has led to suspicions that they may be the source of halophiles in supposedly original rock salt samples. Consequently, great care must be taken when trying to isolate halophiles from rock salt.
Isolations of microorganisms from ancient rock salt
The various factors that need to be taken into account when sampling from rock salt have been reviewed recently ( Grant et al., 1998 ; Vreeland and Powers, 1999; Vreeland and Rosenzweig, 1999). A common cause for concern is that the mine air may harbour haloarchaea. Three studies have failed to isolate haloarchaea from mine air. In Winsford salt mine, no haloarchaeal colonies appeared on halophile agar plates moved through the air for 5 min and then left open for a further 30 min. Similarly, no colonies were found in the New Mexico mine investigated by Vreeland et al. (1998) , even when plates were placed near the fan driving air through the mine. An air-sampling system used by Stan-Lotter et al. (2000) again resulted in no haloarchaeal colonies after depositing the particle content of several thousand litres of mine air onto agar plates.
Heinz Dombrowski and Paul Tasch were the first to isolate microorganisms from rock salt. Summaries of their rigorous technique are found in Dombrowski (1963) and Tasch (1963). Reiser and Tasch (1960) succeeded in isolating diplococci, which they had observed in fluid inclusions in Permian salt (Kansas, USA). These diplococci were able to grow from 0 to 30% (w/v) NaCl and could not be isolated from mine sumps or mine air. Dombrowski performed similar work on Permian rocks in Europe and older rocks worldwide (see Dombrowski, 1963). Enrichments in salt-saturated medium yielded a strain from several sites resembling Bacillus circulans ( Dombrowski, 1963). Another isolate called Pseudomonas halocrenaea ( Dombrowski, 1963) was subsequently shown to be indistinguishable from Pseudomonas aeruginosa ( DeLey et al., 1966 ), and must therefore be considered as a contaminant. Schwartz and colleagues, although able to see fossil bacteria, failed to isolate microorganisms from 30 samples of rock salt ( Bien and Schwartz, 1965). Bibo et al. (1983) attempted to repeat Dombrowski's isolations with numerous controls and tests of surface sterilization. They isolated extreme halophiles, described as Gram-positive cocci and rod-shaped spore formers, from ‘primary’ Permian Zechstein salt cores. It would certainly be valuable to characterize these isolates using modern techniques.
Norton et al. (1993) isolated seven haloarchaeal strains from rock salt in Winsford salt mine; three belonged to the genus Halorubrum and four to Haloarcula. Two separate enrichments of the same core of potash from Boulby salt mine gave rise to two identical strains, which had novel polar lipid profiles and were phylogenetically distinct (represented by strain 2Bbr13.2 in Fig. 1). It should be emphasized that the underground caverns are vast, and these samples were freshly blasted or cored from dry parts of the mines. All samples were surface sterilized by soaking in ethanol and flaming ( Norton et al., 1993 ).
The number of isolations in the study by Norton et al. (1993) was small, about two positive enrichments kg−1 rock salt, despite many areas of the salt mines being sampled on different occasions. This is in stark contrast to recent reports, in which up to 6.9 × 106 colonies kg−1 were isolated in the Permian Salado formation, New Mexico ( Vreeland et al., 1998 ), and in which 1.3 × 105 colonies kg−1 were found from the Permian Zechstein deposit mined at Bad Ischl ( Stan-Lotter et al., 2000 ). In both cases, no prior enrichment was used, and the distribution was very heterogeneous. In the New Mexico mine, most salt crystals gave rise to less than 104 colonies kg−1 ( Vreeland et al., 1998 ). Vreeland et al. (1998) did not attempt to sterilize the surface of the salt crystals, because their purpose was not to demonstrate an ancient origin of the isolates. Stan-Lotter et al. (2000) used large pieces of salt of about 500 g, which were surface sterilized by flaming with ethanol. In the study by Stan-Lotter et al. (2000) , only haloarchaea were isolated, and preliminary characterization places them in the genera Halococcus, Haloarcula and Halorubrum ( Stan-Lotter et al., 2000 ). The latter two were also found in rock salt from Winsford ( Norton et al., 1993 ) and, with Halobacterium species, are usually the most abundant cultivated haloarchaea in salt-saturated surface environments.
It is not immediately clear why such large numbers of halophiles were isolated in the formations studied by Vreeland et al. (1998) and Stan-Lotter et al. (2000) compared with previous investigations (e.g. Norton et al., 1993 ). There were no major differences in the media. Perhaps the surface sterilization procedures of pre-1998 studies were too harsh, resulting in high temperatures inside the crystals. Alternatively, the geology or geochemistry of the two recently investigated deposits may have provided more favourable conditions for preservation and/or growth of viable haloarchaea.
Isolations of microorganisms directly from fluid inclusions
The repeated isolations from rock salt discussed previously suggest that haloarchaea are genuinely present in salt deposits. However, they are not necessarily as old as the formation, particularly if there has been brine movement and secondary crystallization of the salt. Indeed, these processes would encourage growth, multiplication and movement of haloarchaea by supplying new nutrient resources, thereby providing a mechanism for long-term survival of halophilic communities. The Triassic Northwich halite studied by Norton et al. (1993) had not been deeply buried, and care was taken to select samples of salt between marl striations. Although such features are indicative of primary halite, it is extremely difficult to confirm that a crystal is primary. When microorganisms are isolated from bulk rock salt samples, they could have originated from primary or secondary fluid inclusions, intercrystalline boundaries or intracrystalline fractures. In order to demonstrate the age of encased microorganisms, it is necessary to extract brine directly from fluid inclusions inside single, dated, primary crystals, taking great care to avoid contamination. Such an approach was taken by Fredrickson et al. (1997) on halite crystals from a core beneath the Badwater salt pan, Death Valley, USA. Initial attempts using the polymerase chain reaction (PCR) to amplify haloarchaeal 16S rRNA genes were unsuccessful ( Fredrickson et al., 1997 ). However, they have subsequently isolated a strain closely related to Halobacterium salinarum from a crystal, U/Th-dated at ≈ 97 000 years (M. Mormile, personal communication). Recently, a Bacillus species was isolated directly from a fluid inclusion in a crystal of halite defined as primary and from the Permian period (R. H. Vreeland, personal communication). This is not the first time that isolations of strains from this genus, which contains spore formers, have been claimed from ancient salt ( Dombrowski, 1963), and also from ancient amber ( Greenblatt et al., 1999 ).
There are few good negative controls in studies of ‘ancient life’. Dombrowski (1966) failed to recover microorganisms from salts that had been subjected to temperatures of about 160°C. R. T. Gemmell and W. D. Grant (unpublished) did not isolate halophiles from a brine that had originated from a highly deformed potash seam within a salt diapir, which had previously been deeply buried and at a temperature of at least 80°C. Although this brine appeared to be sterile, it was found to be capable of supporting the growth of representatives of several haloarchaeal genera.
Relationship of subsurface haloarchaea to surface isolates
Several of the studies mentioned previously have involved a detailed taxonomic analysis of halophilic isolates, comparing strains from salt deposits with those from surface environments. Phylogenetic analysis of 16S rRNA sequences ( Fig. 1) confirms that many salt deposit isolates belong to the genus Halorubrum, and that many are Haloarcula species ( Gemmell et al., 1998 ), as suggested by Norton et al. (1993) . It is very contentious to equate 16S rRNA dissimilarity with time, because there is a great deal of variation in the molecular clock of different organisms. Moreover, slow growth over geological time would be expected to decrease the mutation rate, while stressful conditions may accelerate it. Putting aside these reservations for a moment, Moran et al. (1993) proposed an evolutionary rate of 1.5 nucleotide substitutions per 100 bases in 16S rDNA per 50 million years on the basis of co-evolution of aphids and their bacterial endosymbionts. This calibration can be used to indicate when two strains last shared a common ancestor. The most closely related pair of haloarchaea from the surface and a salt deposit is Halobacterium saccharovorum and strain 54R. When this value is applied to strain 54R (isolated from ≈ 240 million-year-old rock salt) and its closest relative, it can be seen that they last shared a common ancestor about 10 million years ago. Therefore, either strain 54R entered the salt deposit relatively recently or, as discussed above, the molecular clock is not equally applicable to all organisms and may vary by an order of magnitude. Despite their high 16S rRNA similarity, strain 54R and Hr. saccharovorum have only 48% DNA:DNA homology ( Norton et al., 1993 ), indicating that their 16S rRNA clock may indeed be slow. The possibility that Hr. saccharovorum, which was isolated from crystallizer ponds near San Francisco, was derived from rock salt of Triassic age is probably too far-fetched.
Other salt deposit haloarchaea are quite distinct from surface strains ( Fig. 1), and even form phylogenetic clusters. This does not necessarily mean that they do not exist in surface environments, such as the Dead Sea or Great Salt Lake, merely that we have not looked for them sufficiently.
In Fig. 1, it is notable that Halococcus salifodinae is represented by three strains with identical 16S rRNA sequences: (i) strain BIp from Permian Zechstein rock salt, mined at Bad-Ischl, Austria ( Stan-Lotter et al., 1993 ); (ii) strain Br3 from solution-mined, Triassic Northwich halite ( Norton et al., 1993 ); and (iii) strain BG2/2 from a core of Permian Zechstein salt, Berchtesgaden, Germany. Denner et al. (1994) described BIp as a new species and observed similarities with strain Br3. A detailed characterization of the three independently isolated halococci revealed that they were extremely similar and should be considered as strains of the same species, distinct from Hc. saccharolyticus ( Stan-Lotter et al., 1999 ). During the Permian period, a large hypersaline sea (Zechstein Sea) covered an area of about 250 000 km2 over much of northern Europe. This sea would have provided a connection between the areas from which the three strains of Hc. salifodinae were isolated ( Stan-Lotter et al., 1999 ). It is conceivable that Hc. salifodinae was abundant in the Zechstein Sea and became trapped in the evaporating salts. Again, it is important to emphasize that the presence of Hc. salifodinae in surface environments may have been overlooked. Like all haloarchaea, these three strains need 15% (w/v) NaCl for growth; however, like all halococci, they do not lyse in low salt concentrations, so they could survive outside the salt deposits. However, Berchtesgaden and Bad Ischl are distant from the sea and at high altitude, which would make dispersal by sea water or wind problematic.
Dispersal of haloarchaea
Little is known about the transport of haloarchaea outside hypersaline environments, but this is crucial in understanding the potential for contamination of salt deposits. The review by Grant et al. (1998) should be consulted for more details of haloarchaeal dispersal. Haloarchaea from several genera have been isolated from relatively low-salt environments (see McGenity et al., 1998 ), and so sea water is an obvious means of dispersal. Indeed, haloarchaeal 16S rDNA clones were found in a salt marsh on the east coast of England ( Munson et al., 1997 ), but their sequences did not correlate with those of strains from salt deposits. Wind-blown salt is a potential source of mine contamination. For example, Wheeler (1985) found cubes of halite among Saharan sand that had blown over eastern Britain in 1984. To date, no-one has looked for haloarchaea inside such crystals. The possibility of mine contamination by people and other animals should not be overlooked. Leaks along mine shafts and accidental flooding can lead to the transport of microorganisms from surface water or groundwater. All these possible sources of contamination can be minimized by sampling dry parts of the mine with no history of flooding and by taking crystals from the inner face of freshly blasted, surface-sterilized rock salt.
Long-term survival of haloarchaea inside salt crystals
A large body of evidence, discussed in previous sections, points towards the long-term survival of haloarchaea inside fluid inclusions in halite. It is not known, however, whether such cells are dormant. It is essential to consider how haloarchaea and other halophiles could survive over geological time. There is no evidence of haloarchaea forming spores, but cyst-like structures have been observed in a number of species (see Grant et al., 1998 ). For example, when a haloarchaeon isolated from an Upper Devonian oilfield brine was grown on agar, crystals formed around colonies after 8–10 days ( Zvyagintseva et al., 1998 ). At this point, there was a ‘drastic increase’ in the number of cells with thickened cell walls (210–230 nm), dense spiral packing of DNA and electron-dense cytoplasm ( Zvyagintseva et al., 1998 ).
In order to grow, haloarchaea require a source of carbon, energy, nutrients and a terminal electron acceptor. Fluid inclusions observed in nature ( Norton and Grant, 1988; Castanier et al., 1999 ) can be packed with cells and, therefore, during the early stages of burial, there is likely to be no shortage of usable substrates, largely in the form of necromass. Over geological time, however, it is more difficult to envisage where energy sources will come from. As mentioned earlier, recrystallization may replenish dormant microorganisms. Fluid inclusions migrate, perhaps tapping new energy sources. Can haloarchaeal activities encourage this migration?
The distribution of organic matter in evaporites is very heterogeneous. For example, between 0.1% and 7% organic carbon was found in salts from the Bresse and Mulhouse basins in France, most of which was associated with carbonate clay facies ( Pironon et al., 1995 ). The nature of organic matter also varies widely and can consist of microbial cells, wood debris, pollen and amorphous material. The association between evaporites and petroleum has been known for a long time ( Sonnenfeld, 1984) . Indeed, hydrocarbons are found occasionally within fluid inclusions, and so it was interesting to note that the diplococci isolated by Tasch (1963) could grow on a variety of hydrocarbons. Similarly, Hc. salifodinae strain Br3 grew with branched and straight-chain alkanes as sole carbon and energy source (R. S. Greedy and W. D. Grant, unpublished). The considerable input of plant material in many present-day hypersaline environments may explain why Vreeland et al. (1998) isolated a large number of cellulose-degrading halophiles from the Permian Salado formation.
Fluid inclusions containing respiring microorganisms quickly become anaerobic. Haloarchaea grow at relatively low oxygen concentrations caused by poor solubility of oxygen in brines. Several species can use alternative terminal electron acceptors. such as nitrate, dimethyl sulphoxide, trimethylamine N-oxide, fumarate, elemental sulphur and thiosulphate (see Oren, 1994). Fermentation has also been demonstrated in haloarchaea (see Oren, 1994). The presence of suitable electron acceptors and the ability of salt deposit isolates to use them are worthy of detailed study.
All the factors discussed above, morphological plasticity, physiological versatility, together with their inherent protection against DNA damage, make haloarchaea ideal candidates for long-term survival.
Conclusions and the way forward
From the studies reviewed here (and elsewhere), increasing evidence is emerging for the presence of viable microorganisms in geological formations that are millions of years old. It is still not known, however, whether ancient salt deposits are merely a repository for dormant microorganisms, or whether they provide a subsurface habitat in which halophilic microorganisms can grow and multiply, perhaps interspersed with relatively short periods of dormancy. The possibility that haloarchaea (or other halophilic microbes) could survive in a state of dormancy over geological time periods remains to be proven unequivocally. Were it to be so, all kinds of interesting questions about evolutionary rates and physiological adaptation could be addressed in a unique way. We could perhaps use revived microorganisms as living biomarkers to determine palaeotemperatures. In the meantime, long-term dormancy cannot definitely be ruled out, and the Mars exploration programme continues to give the topic an impetus beyond the normal scientific curiosity. Biological explanations for long-term survival could, in principle, be tested in the laboratory, providing plausible explanations for the hypothesis of longevity. Repeated isolations however, using protocols that would reasonably exclude contamination, coupled with definite depositional dates for the target microhabitats would appear to be the best evidence that might be obtainable.