Is there a common water-activity limit for the three domains of life?

1Water availability (water activity (aw)) determines both the vitality and functionality of living systems. The majority of microbes cannot multiply below 0.900 aw and for the most extremophilic species, cell division has only been observed down to ∼0.61 aw. The established water-activity window for cell division of archaeal and bacterial life (1–0.755; see; ) is narrower than that of some xerophilic fungi that are even able to grow and/or germinate in the range 0.755–0.605 aw. Hence the maxim that eukaryotic systems have evolved levels of solute tolerance superior to those of prokaryotes.

2Microbes are exposed to hostile conditions because of the spatial heterogeneity of their habitats and the temporal dynamics of environmental stress parameters, as well as collateral damage induced to their macromolecular systems by the solute activities of their own chaotropic and hydrophobic metabolites. Recent studies have addressed the means by which temperature, chaotropicity, hydrophobicity, pH and radiation determine the limits of the functional biosphere (for examples see;;;;;;;;; ). In contrast, there is a paucity of studies to investigate whether physiological processes can occur in low water-activity environments hitherto considered hostile to biological activity. From cultivation-based studies, the majority of Archaea appear to be extremophilic, whereas Bacteria account for the majority of the biodiversity on Earth. It is some members of these domains that hold current records for microbial stress tolerance towards high temperature, chaotropicity, acidity and radiation (;;;;;;; this article is concerned with the ability to retain metabolic activity and undergo cell division rather than the ability to survive in a dormant condition).

3In consequence, we consider it highly unlikely that the cellular biology of these prokaryotes is less capable than that of stress-tolerant members of the Eukarya at high solute concentrations. This study was therefore carried out to determine whether there is a common water-activity limit for the three domains of life; we hypothesised that: (1) halophilic Archaea and Bacteria are capable of cell division below the established 0.755 water-activity limit, that is, this limit is an artefact created by the solubility limit of NaCl rather than a product of their inherent biology; and (2) the most resilient Archaea, Bacteria and Eukarya are equally tolerant to low water-activity.

4A series of experimental, culture-based studies were carried out to determine water-activity limits for Archaea and Bacteria at high-salt concentrations (Figures 1 and 2, Table 1 and Supplementary Table S1), for Eukarya on high-sugar substrates (Figures 3, 4, 5 and Supplementary Table S1) and at high-salt concentrations (Supplementary Tables S1 and S2), and for Bacteria at high-sugar concentrations (Supplementary Tables S1 and S3). Details of microbial strains and culture media are given below; for those microbes where growth rates had been derived previously (see Supplementary Table S1; limits were determined as described in sections on ‘Water-activity measurement' and ‘Determination of water-activity windows for biotic activity').

5An initial assessment of water-activity limits for halophilic species of Archaea and Bacteria was carried out on saline media ranging from 0.803 to 0.642 aw (Figure 1a and Supplementary Table S4) using Halobacterium noricense (DSM 15987T), Halobacterium sp. NRC-1 (i.e. Halobacterium salinarum ATCC 700922), Halococcus morrhuae (NCIMB 787T), Halococcus salifodinae (DSM 13046), Halorubrum saccharovorum (NCIMB 2081T) and Natrinema pallidum (NCIMB 777T). These cultures were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany; Hbt. noricense and Hcc. salifodinae) and the National Collection of Industrial, Food and Marine Bacteria (NCIMB, Aberdeen, UK; Hcc. morrhuae and Hrr. saccharovorum). Hbt. sp. NRC-1 was provided by Helga Stan-Lotter (University of Salzburg, Austria). The composition for the salt-supplemented culture media used to assay these halophiles are detailed in Supplementary Table S4. Cultures were maintained in their respective media and incubated during these experiments at 37 °C. Data obtained were compared with those for Haloquadratum walsbyi (DSM 16790), Pontibacillus (strain AS2), Salinicola (strain LC26), haloarchaeal strains GN-2 and GN-5, Halorhodospira halochloris, Halorhodospira halophila strain (DSM 244T), Halanaerobium lacusrosei (strain DSM 10165T), Actinopolyspora halophila (strain ATCC 27976T), ‘Haloarcula californiae' (strain DSM 8905), ‘Haloarcula sinaiiensis' (strain DSM 8928) and Halorhabdus utahensis (strain DSM 12940T); see Supplementary Materials and Methods).

6Studies of halophilic Archaea and Bacteria capable of growth in MgCl2-rich and/or NaCl-rich media (Figures 1b–f and 2) were carried out using pure cultures of the following strains: Halobacterium strain 004.1 (for source see ), Salinibacter ruber strain DSM 13855T and Salisaeta longa strain DSM 21114T. Cultures of Halobacterium strain 004.1 were maintained in the nutrient media on a rotary shaker at 37 °C (for details see Supplementary Table S5; ). For studies of MgCl2 tolerance, S. ruber was maintained on Modified Yeast Extract, Salts Broth (0.840 aw) containing (per litre): 1.0 g yeast extract (Beckton, Dickinson and Company, Sparks, MD, USA), 5.0 g KCl, 1.25 g CaCl2.2H2O, 0.625 g NaBr, 0.25 g NaHCO3 plus NaCl (3.34 M), MgSO4 (0.101 M) and MgCl2 (0.0801 M), and incubated at 35 °C. For studies of both NaCl and MgCl2 tolerance, S. longa was maintained in (per litre) 2.0 g soluble starch (BDH, Leicester, UK), 1.0 g yeast extract (Difco, Oxford, UK), 1.0 g casamino acids, 5.0 g K2SO4, 0.1 g CaCl2.2H2O plus NaCl (1.71 M) and MgCl2 (0.246 M), and incubated at 35 °C. The composition and water-activity values for culture media used to assay S. ruber and S. longa are detailed in Supplementary Tables S6 and S7, respectively.

7Strains of Aspergillus penicillioides (JH06THH and JH06THJ) were isolated previously, and are available from the corresponding author, and strains of Xeromyces bisporus (FRR 0025, FRR 2347 and FRR 3443) were obtained from CSIRO Food and Nutrition (FRR collection, North Ryde, NSW, Australia). Cultures were maintained on Malt-Extract, Yeast-Extract Phosphate Agar (MYPiA (per litre): 10 g malt-extract w/v (Oxoid, Oxford, UK), 10 g yeast-extract w/v (Oxoid), 15 g agar w/v (Acros, Bridgewater, NJ, USA), 1.0 g K2HPO4 w/v)) supplemented with either 6.52 M glycerol (A. penicillioides) or 2.2 M sucrose (X. bisporus) and incubated in sealed polyethylene bags at 30 °C. The composition and water-activity values for culture media used in stress-tolerance assays of xerophile strains are detailed in Supplementary Table S8.

8Debaryomyces hansenii strain DSM 70590 was used to determine the lower water-activity limit in high-salt media (Supplementary Table S2). Cultures were maintained in Modified Yeast Extract, Casamino Acid, Tri-Sodium Citrate Broth that contained additional NaCl (3.5 M) and MgCl2 (0.399 M); 0.803 aw (see Supplementary Table S4) and incubated at 37 °C.

9Exponential-phase cells of Hbt. noricense, Hbt. sp. NRC-1, Hcc. morrhuae, Hcc. salifodinae, Hrr. saccharovorum, Nnm. pallidum, Pontibacillus strain AS2 and Salinicola strain LC26 were used to inoculate modified Payne's media (250 ml in side-arm flasks; ) of varying water activities (see Supplementary Table S4) at a starting cell density equivalent to 0.2 OD600 nm. Cultures were incubated at 37 °C and cell density was determined by turbidometric estimation using a nephelometer over a period of two months (two replicates; see Table 1 and Figure 1a). Turbidometric readings were used to construct growth curves from which doubling rates (in days) during exponential growth phase were calculated (Table 1 and Figure 1a).

10Exponential-phase cells of Halobacterium strain 004.1 (1.6 ml of a preculture, OD560 nm=0.5) were used to inoculate media (160 in 250 ml Erlenmeyer flasks) with salt concentrations at increasing multiples of the control medium (see Figure 1d and Supplementary Table S5). Flasks were incubated on a rotary shaker (37 °C; 225 r.p.m.) and cell density monitored turbidimetrically (OD560 nm). Cell concentration was determined with a Helber haemocytometer (Hawksley, Sussex, UK) and plotted against the water activity (see below).

11Exponential-phase cells of S. ruber were used to inoculate complex media (100 in 250 ml Erlenmeyer flasks) supplemented with water from the Dead Sea, Israel (35% w/v total salts; see Figure 1e and Supplementary Table S6) to determine tolerance to MgCl2/NaCl mixtures in relation to water activity. Cultures (three replicates) were incubated at 35 °C over 6 days during which OD600 nm was regularly determined. Mean values (OD units per day) derived from exponential-phase measurements were plotted against media water-activity (Figure 1e). Exponential-phase cells of S. longa were used to determine tolerance to MgCl2/NaCl mixtures (100 in 250 ml Erlenmeyer flasks) in relation to water activity (Supplementary Table S7).

12For the xerophilic fungi assayed, agar plugs (4-mm diameter) were taken from the periphery of actively growing cultures of A. penicillioides and X. bisporus and inoculated onto MYPiA supplemented with a range of stressors (Supplementary Table S8). Petri plates were sealed using Parafilm (Neenah, WI, USA) and plates of the same water activity were incubated in bags of polyethylene that allows gaseous exchange but minimises water loss. Colony diameter was measured daily (in perpendicular directions) and used to calculate rates for radial hyphal extension in mm per day (see ).

13Growth of Debaryomyces hansenii (Supplementary Table S2), cultured in Modified Yeast Extract, Casamino Acid, Tri-Sodium Citrate Broth supplemented with NaCl and MgCl2 (see Supplementary Table S4), was assessed by via nephelometer (see above) for a period of two months at 37 °C. Turbidometric readings were used to construct growth curves, and doubling times during exponential-growth phase were calculated in order to determine the lower water-activity limit (Supplementary Table S2). For details relating to utilisation of growth-related data from previous studies (Supplementary Table S1), see Supplementary Materials and Methods.

14Water-activity values of media were measured using a Novasina Humidat IC-II water-activity machine (Axair Ltd, Pfäffikon, Switzerland) at the same temperature at which the relevant microbial culture was incubated. The apparatus, fitted with an alcohol-resistant humidity sensor and eVALC alcohol filter that prevents interference by volatiles such as ethanol and glycerol (; Axair Ltd), was calibrated using saturated salt solutions of known water-activity. Water-activity determinations were carried out three times using replicate solutions made up on separate occasions; variation between replicate values was within ±0.002 for determinations in the range 1 to 0.900 aw, and within ±0.001 for those within the range 0.900 to 0.600 aw (see below).

15Media used for stress-tolerance assays of Pontibacillus strain AS2, Salinicola strain LC26 and Hqr. walsbyi (Table 1 and Supplementary Table S1;; ), the halophilic bacterial community (largely Salinibacter) from crystalliser pond CR-30 (Braç del Port, Alicante, Spain) (Supplementary Table S6 and Figure 1b; ) and S. ruber (Supplementary Table S6 and Figure 1f;; ) were remade for the current study, and then water-activity determinations carried out as described above. The ionic composition of brines used to culture haloarchaeal strains GN-2 and GN-5 (which was determined to an accuracy level within the low-mM range; ) was used to calculate the water activity of these media (Figure 1a). The values obtained were consistent with the conversion of their brine density (Bé°) to water activity by.

16Lower water-activity limits for cell division in some of the most halophilic eukaryotes were determined by plotting extant data sets for cell division at high-salt concentrations in relation to water activity (Supplementary Tables S1 and S2). For culture media of algal species Dunaliella parva (UTEX 1983), Dunaliella peircei (UTEX 2192) and Dunaliella salina (UTEX 200), and the fungi Eurotium halotolerans (MZKI A-560), Wallemia ichthyophaga (EXF-994), Wallemia sebi (FRR 1473) and Wallemia muriae (EXF-951;;; ), the water-activity values that correspond to the limit for cell division were calculated according to the concentrations of NaCl and other components of their respective culture media (Supplementary Table S2) by reference to standard water-activity curves that had been plotted using data points derived at the relevant temperature. Lower water-activity limits for cell division of the nanoflagellate species Euplaesiobystra hypersalinica (CCAP 1528/1), Halocafeteria seosinensis (EHF34) and Pleurostomum flabellatum (CCAP 1959/1) were determined by recreating their respective culture media (see,, ), followed by empirical determinations of water activity (see below) at their respective incubation temperatures (see Supplementary Table S2).

17Water-activity limits for cell division in some of the most xerotolerant Bacteria were determined by plotting extant data sets for cell division at high concentrations of sugar (or other non-ionic solutes) in relation to water activity (Supplementary Tables S1 and S3). Water activity values that corresponded to the cell-division limits for Tetragenococcus halophilus (strains T11 and T15) were determined by empirical measurements of water activity (see below) of sucrose solutions (30 °C) that were made up based on the °Bx values from, ). Water-activity limits for Asaia bogorensis (JCM 10569T), Gluconacetobacter diazotrophicus (DSM 5601T), Saccharibacter floricola (DSM 15669T), Pseudomonas putida (DSM 6125h), Halotalea alkalilenta (AW-7T) and Rosenbergiella nectarea (8N4, LMG 26121, DSM 24150) were determined empirically (see below) using sugar solutions that were remade, and then measured at their appropriate incubation temperatures (;;;; Supplementary Table S3).

18There are various theoretical and empirical ways in which to determine water-activity values (see, for example,;;;,;,; ). For undefined media and/or culture media that have been sterilised, poured into Petri plates and/or stored, water activity cannot be predicted due, in part, to water loss as vapour and hence must be determined empirically. Water-activity values obtained via any available technologies are associated with some degree of variation because of the respective limits of uncertainty and accuracy of the latter (see ). Documented water-activity values for saturated salt solutions that can be used for the calibration of instruments are relatively accurate (see; ). For example, demonstrated overall uncertainties for water-activity measurements of only ±0.0012 to ±0.0018, and accuracy levels of 0.0002 to 0.0005. Some of the new generation of commercially available instruments, such as the Novasina Labmaster series (Axair Ltd), are limited by software that can only be calibrated at a small number of fixed water-activity values and temperature points. The Novasina Labmaster series have been programmed using theoretical water-activity values that were derived using data pooled from a small number of empirical measurements and/or a series of (untested) assumptions, rather than those collated from a large number of (methodologically diverse) sources—and therefore more accurate—by.

19We took precautions to minimise potential errors that can arise from sampling protocol, instrument resolution and discrimination threshold, inaccuracy of measuring equipment and values for constants or other parameters obtained from external sources and so on (including standard values of saturated salt solutions used for calibration;;; ). An earlier model of water-activity apparatus was used—the Novasina Humidat IC-II (see above)—that allows for manual calibration along continuous scales of water activity and temperature, as used in our earlier studies (see, for example,,,,,;;;;; ); filters were fitted to prevent interference from volatiles or particulates, and these were installed taking care to prevent tearing (as described previously, ); saturated salt solutions were made up using highest-grade salts and these (along with the medium sample which was contained in a closed sample cup, and the Novasina Humidat IC-II sensor) were left to equilibrate at the temperature at which water-activity determinations were carried out for two to three weeks as described by ); determinations were carried out in a Binder MK53 Environmental Simulation Chamber (Binder GmbH, Tuttlingen, Germany) to minimise temperature variation; measurements were invariably made when the sensor reading was below the water activity of the sample to avoid making the sensor desaturate during a reading; for solid media, Petri plates were poured at a temperature of only 5–7 °C above the medium gel-point (which can differ by up to 10 or 15 °C depending on stressor type and concentration; see ) and lids were immediately placed onto plates to minimise loss of water vapour, avoid condensation and minimise any potential difference between water lost upon pouring between the first and last plates of each batch (for the same reason, agar-containing media were made in small volumes where possible and poured rapidly as practically possible); water-activity readings were taken frequently throughout each sample equilibration period in order to enable the construction of curves that help to establish levels for accuracy of values, consistency and sensor functionality; the Novasina Humidat IC-II sensor was allowed to desaturate immediately after each water-activity measurement to avoid growth of xerophilic fungi on the sensor or sensor housing; and other precautions that are listed in the following section. Uncertainty analyses for water-activity determinations using Aqualab equipment (which is based on dew-point measurements rather than humidity readings) that were carried out by established that equipment errors were less than those stated by the manufacturer (that is, ±0.003). This is consistent with our finding that the variation of our values using the now-obsolete Novasin Humidat IC-II (±0.001 in the range 0.900 to 0.600 aw) was less than that claimed by the manufacturer for current Labmaster models that use the same sensor technology (that is, ±0.003).

20Assessments of microbial growth rates were made on media over a range of water-activity values in order to facilitate the construction of curves to determine water-activity windows (Supplementary Table S1 and Figures 1, 2, 3;;,;; ). As a precaution, water-activity values of media were checked post-incubation (none of these differed from preinoculation values by more than ±0.002). Growth curves were extrapolated where necessary (Supplementary Tables S1–S3 and Figures 1, 2, 3, 4) to predict the failure point of the window. This has been carried out for data obtained from (1) planktonic cultures of diverse microbes in liquid media; (2) spot-test assays for filamentous Bacteria on solid media; and (3) radial extension of mycelia for filamentous fungi on solid media (Hallsworth and Magan, 1999;;;; ). Some of these studies demonstrated/validated theoretical determinations for water-activity minima with an accuracy consistent with the level of sensitivity of the microbial cell; that is, ±0.002. However, where growth curves continue in an asymptotic manner, extrapolation may give a conservative estimate for water-activity minima for both fungi (see Figures 1b, 3 and 4) and planktonic cultures of prokaryotic cells. For studies of temperature, modelling of microbial growth curves can also result in conservative predictions of the actual limits. In the current study, linear, polynomial or Gausian regression analyses were carried out to predict water-activity minima; for each data set, the regression analysis that gave the highest regression coefficient was employed (Supplementary Table S1). In order to compare water-activity limits for the most resilient microbes known, we selected the 30 eukaryotic microbes (see Supplementary Tables S1 and S2 and Figures 3 and 4) and 30 Archaea or Bacteria (see Supplementary Tables S1 and S3 and Figures 1 and 2) with the highest recorded tolerance to ionic or non-ionic solutes (Figure 5;; ). These data were compiled to determine only the lower water-activity limits of microbial growth windows, and were based on biotic activity including germination rates and growth rates, as there are no data sets derived using culture-independent techniques that reveal metabolic activity at such low water-activity values to our knowledge.

21A number of halophilic Archaea and Bacteria are not only capable of cell division in salt-saturated substrates, but can have substantial growth rates under these conditions. NaCl is soluble to ∼5.2 M, which is equivalent to 0.755 aw at temperatures in the range of 25–35 °C (the water activity of saturated solutions of NaCl ranges from 0.765 to 0.745 between 2 °C and 50 °C; ). We selected nine extremely halophilic species of Archaea and Bacteria that belong to a range of phyla, and determined whether they would grow in liquid media supplemented with various stressors to obtain a range of water-activity values from 0.803 to 0.642 (Figure 1a, Table 1 and Supplementary Table S4). At low water-activity values, cell-replication times for the vast majority of halophiles become exponentially long. Remarkably, however, the majority of the empirically determined doubling times for these microbes ranged from <7 to ∼21 days at water-activity values of ≤0.717, and four of the selected species grew at even lower water-activity values: 0.709 for Hqr. walsbyi, 0.693 for Hcc. salifodinae, 0.687 for Hbt. noricense and 0.681 for Nnm. pallidum (Figure 1a and Table 1). The relatively high growth rates of Hqr. walsbyi, Hbt. noricense and Nnm. pallidum, despite this low water-activity range (0.709 to 0.681), suggest that their actual water-activity minima are <0.681 (likely in the range 0.650 to 0.605; Figure 1a). For some of the media assayed, the NaCl concentrations were relatively low (between 3.3 and 1.5 M) and MgCl2 concentrations were relatively high; two of the media contained ethylene glycol or glycerol (Figure 1a, Table 1 and Supplementary Table S4). When compared on a w/v basis, MgCl2 depresses water activity more effectively than NaCl; it is, nevertheless, the potent chaotropicity-mediated stress induced by MgCl2 that typically limits microbial systems which are exposed to this salt. It is for this reason that we substituted MgCl2 with ethylene glycol and glycerol for some treatments. Glycerol can protect against chaotrope-induced stresses in phylogenetically diverse microbes. This polyol can, however, itself act as a chaotropic stressor at molar concentrations. For Hbt. noricense and Hcc. salifodinae, the partial substitution of salts by either ethylene glycol or glycerol facilitated growth at lower water-activity values than occurred on media with only salts added (Table 1 and Supplementary Table S4). Given the ability of MgCl2 to reduce water activity to <0.755 (several studies suggest that MgCl2 concentrations in the range 2–3 M can be biologically permissive for some halophiles), we sought evidence that halophilic Archaea and Bacteria can be active at water activities <0.681 in magnesium-rich media or substrates including bitterns, that is, crystalliser ponds that have naturally elevated magnesium concentrations (Figures 1b–f).

22Two strains of haloarchaea that originated from Mexican solar salterns (GN-2 and GN-5) were previously cultured in bittern water. We determined the water activity of this substrate (see Materials and methods) and plotted growth data for these strains against water activity (Figure 1b), indicating that although optimal growth was observed between 0.800 and 0.845 aw, growth rates were remarkably high at 0.755 aw; that is, 61% and 67% of that of the optimum for GN-2 and GN-5, respectively. Despite the short incubation time (6 days), both strains were active even in the most high-salt medium (equivalent to a brine density of 32 Bé° ) at a water activity of 0.635 (Figure 1b). In common with other studies of water-activity and temperature windows for microbial growth (or those for other stress parameters; see Materials and methods), we carried out regression analyses to obtain theoretical water-activity minima for strains GN-2 and GN-5; these were 0.615 and 0.611, respectively (Supplementary Table S1 and Figure 1b). This represents a substantial increase in relation to the previously accepted water-activity window for cell division of archaeal or bacterial halophiles.

23We also sought evidence of microbial activity at sub-0.755 water-activity values for halophilic Bacteria that are known to inhabit bittern brines. Water from a crystalliser pond (Pond CR-30, Braç del Port) was previously used to inoculate a synthetic seawater medium supplemented with NaCl, which was then monitored over time by quantifying total cell counts and using molecular probes to identify halophilic Bacteria belonging to the Salinibacter assemblage within the community (; Figure 1c). We determined the water activity of these media (see Materials and methods) and plotted these values against growth-rate data for this Bacterial assemblage (Figure 1c). Growth rates were optimal at 0.841 aw (doubling time=12 h) and reduced by only 50% at 0.755 aw. Extrapolation of this curve suggests an actual water activity limit in the range 0.675–0.670 (Supplementary Table S1 and Figure 1c). These values were derived from a combined data set representing a number of bacterial populations, and hence it is likely that one or more populations and species is/are capable of growth at lower water activities than this extrapolation suggests.

24The biotic windows were also determined for individual Archaea and Bacteria found in magnesium-rich habitats and/or known to have a degree of tolerance towards MgCl2 (strains of Halobacterium, Salinibacter and Salisaeta; Figures 1d–f). Halobacterium strain 004.1, isolated from a brine pool within a subsurface salt deposit (UK, ), shares traits in common with strains of Hbt. noricense, a species that is frequently found in crystals of buried halite. This strain remained highly active at 0.728 aw (∼70% of the optimum growth rate) and extrapolation of the curve suggested a lower theoretical limit of 0.658 aw (Supplementary Table S1 and Figure 1d). S. ruber can form up to 27% of halophile communities in high-magnesium habitats such as crystalliser ponds in solar salterns. This bacterium has extraordinary adaptations that enable it to dominate high-salt habitats, including light-activated protein pumps that generate proton-motive force and thereby boost energy generation, use of ions—rather than organic compounds—for osmotic adjustment and cytosolic proteins with physicochemical properties that enable function at high ionic strength. S. longa, a close relative of Salinibacter spp., is also found in hypersaline marine habitats. However, we found that high-magnesium media (supplemented with water from the Dead Sea; Supplementary Tables S6 and S7) were not biologically permissive for S. ruber and S. longa in as much as their growth windows did not extend below 0.755 (Figures 1e and f).

25Given that the chaotropicity of MgCl2 may have curtailed the growth of S. ruber and S. longa (Figures 1e and f), we also determined growth rates for these strains in high-NaCl media (Figure 2). Bacteria notorious for obligate extreme halophilicity at high NaCl concentrations have been studied previously: Actinopolyspora halophila, Halanaerobium lacusrosei (formerly Haloanaerobium lacusroseus), Halorhodospira halochloris (formerly Ectothiorhodospira halochloris) and Halorhodospira halophila (formerly Ectothiorhodospira halophila;;;;;;; ), as well as the exceptionally halophilic Archaea ‘Haloarcula californiae', ‘Haloarcula sinaiiensis' and Halorhabdus utahensis. We therefore quantified the water-activity limits for each species based on previous empirical growth determinations (Figure 2 and Supplementary Table S1). Two of these species, A. halophila and Hlr. halochloris, exhibited exceptionally high biomass/growth rates at 0.757 and 0.774 aw, respectively, in salt-saturated media that were ∼95% of those corresponding to their water-activity optima (Figures 2b and d, respectively). Unlike S. ruber, A. halophila utilizes organic compounds for osmotic adjustment and is known to accumulate high levels of compatible solutes such as ectoine and trehalose. Extrapolations of growth curves (Supplementary Table S1) indicated theoretical limits of 0.747 for S. longa (Figure 1f), 0.725 for S. ruber (Figure 1e), 0.710 for ‘Har. sinaiiensis' (Figure 2e), 0.704 for ‘Har. californiae' (Figure 2e), 0.680 for Hlr. halochloris (Figure 2a), 0.668 for H. lacusrosei (Figure 2c), 0.660 for A. halophila (Figure 2d), 0.647 for Hrd. utahensis (Figure 2f) and 0.623 for Hlr. halophila (Figure 2b). The data presented in Figures 1 and 2 are consistent with the high growth rates of other archaeal and bacterial species known to be extreme obligate halophiles (for example, Actinopolyspora iraqiensis (syn. Saccharomonospora halophila) strain IQ H2,; ). Furthermore, the majority of the halophilic prokaryotes represented in Figures 1 and 2 achieved optimum growth rates in the range 0.845 to 0.765 aw. This water-activity range is lower than, or equivalent to, that of the most xerophilic fungi thus far reported, with the exception of Xeromyces bisporus (see below).

26Before our recent studies involving manipulation of the chao-/kosmotropicity of culture media, the established water-activity limit for mycelial growth in xerophilic fungi was 0.656 (for X. bisporus, ). In the current article, we attempted to extend biotic windows of xerophilic fungi via manipulations of medium composition and other environmental conditions, and in this way to emulate the study of halophilic prokaryotes described above. Determinations were made for water-activity windows for hyphal extension (rather than germination) to enable comparisons with the stress tolerance of the Archaea and Bacteria (Table 1 and Figures 1 and 2). Furthermore, vegetative growth of mycelia is qualitatively distinct from spore germination in terms of cellular ultrastructure and physiology, growth kinetics and water-activity limits. Whereas fungal propagules are renowned for their inherent robustness (for example, during desiccation and at extreme pressures and temperatures), this tenacity is only observed during dormancy. For this study, we selected five strains of extremely xerophilic fungi: X. bisporus FRR 0025 (reported to grow down to 0.656 and germination down to 0.605 aw; ); X. bisporus strains FRR 2347 and FRR 3443 that are capable of hyphal growth at water activities comparable to strain FRR 0025; and Aspergillus penicilliodes strains JH06THH and JH06THJ that have been reported to be more xerophilic than X. bisporus (the former were capable of mycelial growth down to 0.647 aw; ).

27The water activity for optimum growth of A. penicillioides was found to be between 0.800 and 0.820 (Figure 3); lower than that for X. bisporus, other species of xerophilic fungi (Figure 4; ) and most halophilic Archaea and Bacteria (Figures 1 and 2). Radial growth rates for A. penicillioides and X. bisporus at ∼0.656 aw (that is, ≤0.074 and 0.125 mm per day, respectively) were two orders of magnitude slower than those recorded under optimal conditions, yet strains of both species were able to grow down to 0.640 aw (Figures 3 and 4), a hitherto unprecedented limit for mycelial growth of xerophiles. Indeed, at 0.640 aw, A. penicillioides strain JH06THH and X. bisporus strain FRR 3443 grew at rates equivalent to 3.43 and 13.0 mm per year, respectively. In the context of the high-solute and desiccated habitats of xerophiles (some of which can, over periods spanning years or decades, establish populations in environments such as dust and ancient papers and fabrics;;;; ) this is a remarkable rate of growth. Extrapolations indicate lower water-activity limits of 0.636 and 0.632 for X. bisporus FRR 3443 and A. penicillioides JH06THJ, respectively (Supplementary Table S1 and Figures 3 and 4).

28The lowest empirically determined water activity for vegetative growth of a eukaryotic microbe is, therefore, 0.640 and the theoretical lower limit is 0.632 (Figure 4); for Archaea or Bacteria the lowest empirically determined value is 0.635, with theoretically determined minima of down to 0.611 (Figures 1 and 2). Planktonic growth of the yeast Zygosaccharomyces rouxii was recorded at 0.620 aw (with a theoretical limit of 0.616; ), conidial germination of Aspergillus echinulatus at 0.620 and germination of X. bisporus at 0.605, although neither development of mycelium nor sporulation of X. bisporus occurred at 0.605 (and none of these achievements have been equalled or exceeded during the subsequent decades, to our knowledge).

29A comparison of 60 of the most solute-tolerant microbes (Figure 5) indicates modest water-activity limits for osmotolerant Bacteria and halotolerant/philic eukaryotes relative to those for halophilic Archaea and Bacteria or xerophilic eukayotes. Empirical determinations of water-activity minima for halophilic prokaryotes indicate that numerous strains are not capable of multiplication below 0.755 aw; but that some can do so below 0.650 (Figure 5; ). Up to now, there has been a paucity of studies carried out to establish the true water-activity windows of extreme halophiles by circumventing the solubility limit of NaCl (for example, Table 1 and Figure 1a). Nevertheless, the theoretical determinations suggest parity between the most xerophilic members of the Archaea, Bacteria and Eukarya that are virtually equivalent in their water-activity limits (Figure 5). It is noteworthy that halophilic Archaea and Bacteria are active at water-activity values that are less than both the previously established and revised limits for hyphal growth of fungi (0.632, Figure 3) on high-sugar media (Figure 5). If theoretical values (those derived by extrapolation) are not included, the water-activity minimum for multiplication of haloarchaeal strains GN-2 and GN-5 (0.635 aw) are marginally lower than those for hyphal growth of xerophilic fungi (Figures 3 and 4). In nature, extremely halophilic Archaea and Bacteria are not only found in salterns but are also present within, and can dominate, microbial communities located in the hypersaline fluid inclusions of salt crystals in evaporite deposits that underlie a considerable portion of the Earth's surface. The osmophilic fungal xerophiles X. bisporus and Z. rouxii (see Figures 4 and 5) inhabit high-sugar environments such as dried fruits (for references, see ). Highly xerophilic strains of A. penicillioides, Aspergillus echinulatus, Eurotium amstelodami and Eurotium chevalieri have been isolated from both high-solute and other desiccated habitats. Eurotium species, Bettsia fastidia, A. penicillioides and W. sebi are particularly common in grains, nuts and spices and, indeed, A. penicillioides may be the pioneer species in such habitats. These fungi are also common spoilage species in many low water-activity baked foods, dried meats and fish, whereas Xerochrysium xerophilum, Eurotium repens, Eurotium halophilicum and W. sebi species are most commonly associated with high-sugar habitats, particularly confectionery, chocolate, jams, maple syrup, dried substrates such as hay, dry beans and grains, and dried fruits. Water activity can act as a determinant of community composition and ecosystem function for diverse ecophysiological groups. We investigated the limits for algae, fungi and nanoflagellates in saline substrates (Supplementary Table S2) and those for Bacteria in high-sugar or high-polyol substrates (Supplementary Table S3).

30The water-activity minima for most halophilic members of the Eukarya ranged between 0.743 and 0.712 (for the fungi Polypaecilium pisce at 0.741, Wallemia ichthyophaga at 0.720 and Basipetospora halophila at 0.712; for the algae Dunaliella peircei at 0.743 and Dunaliella salina at 0.739; Supplementary Table S2) that is substantially higher than those of comparable species of halophilic Archaea and Bacteria (Figure 5). Basipetospora halophila and Wallemia ichthyophaga are commonly found in salty environments. Most isolates of Basipetospora halophila have been isolated from salted and dried fish and also from dried seaweed food and sea salt. Wallemia ichthyophaga occurs on salted, dried meat, hypersaline waters of salterns, salt crystals and MgCl2-rich bitterns (Jančič et al., unpublished data). The hypersaline waters of salterns are also an important habitat for xerophylic Wallemia sebi and Wallemia muriae. Dunaliella species are highly prevalent in microbial communities of salt-saturated salterns as well as other niche habitats such as spider-web silk in desert environments. The water-activity limits for three nanoflagellates that are also found in salterns at or close to saturated NaCl (0.782 for Euplaesiobystra hypersalinica; 0.767 for Pleurostomum flabellatum; and 0.757 for Halocafeteria seosinensis; see Supplementary Table S2) are lower than those of some other eukaryotes, but are nevertheless exceptional for grazing species that are thought to be absent from most salt-saturated habitats. The xerotolerance of the most sugar-tolerant Bacteria (in the range 0.849 to 0.800 aw for Mycobacterium parascrofulaceum, Mycobacterium smegmatis, Saccharibacter floricola and Tetragenococcus halophilus) was inferior to that for the nanoflagellates E. hypersalinica, H. seosinensis and P. flabellatum as well as fungal comparators (Figure 5). Mycobacterium species, including those that were capable of growth at high concentrations of glycerol or PEG 400 (down to 0.800; Supplementary Table S3; Santos et al., unpublished data), can be isolated from the surface film of sphagnum moss, algal communities and other habitats that (like salterns) may have high concentrations of glycerol and/or other organic low molecular mass solutes that reduce water activity. There are various low water-activity, sugar-rich substances of biotic origin such as dried or high-sugar fruits, honey, maple syrup and sugar-beet juice that can delay or prevent microbial colonisation and the formation of microbial biomass as they depress water activity to values outside the growth windows of most, if not all, osmotolerant and osmophilic Bacteria and yeasts (they may also contain antimicrobials and/or constituents—such as fructose and ethanol—that are chaotropic; for references, see ). It is noteworthy that the data presented for biotic activity at extremely low water-activity (Figure 5) come exclusively from culture-dependent studies as there is a paucity of evidence from culture-independent studies to demonstrate microbial processes at equally low water-activities.

31The findings demonstrate that some species of halophilic Archaea and Bacteria are active at water activities considerably below 0.755, suggesting that—based on extant data sets—microbial xerophilicity ultimately converges on a narrow range of water activity (∼0.650–0.600), and possibly even a common value—of ∼0.61—for all three domains of life. Given that saline, rather than sugar-rich, habitats were most common on the early Earth, this finding has implications for the origins of terrestrial life. Furthermore, we know much about the water activity of potential microbial habitats in extraterrestrial locations, some of which could potentially be inhabited by prokaryotic halophiles capable of multiplication below 0.755 aw. There are implications of the findings of the current study, therefore, in relation to planetary protection. The net effect of multiple physiological factors and diverse stress parameters is known to determine the extent of microbial growth windows. Conversely, it appears that water activity is the ultimate determinant for biotic activity and cell division of numerous extremophiles (for example, haloarchaeal strains GN-2 and GN-5, A. penicillioides and X. bisporus; Figures 1b and 3) and is likely to be equally true for such microbes in their natural habitats, possibly even for microbes not located in high-solute habitats or those that access water from the vapour phase. Given the fundamental roles of water as a ‘chaperone' and in macromolecular hydration, generation of hydrophobic forces, and other interactions within and between cellular macromolecules, we consider water to be the most potent force to shape the functional biosphere on Earth (see also;,;;;;;; ). Converging lines of evidence indicate that the ultimate limits for solute tolerance of xerophilic fungi and halophilic Archaea and Bacteria are determined by a prohibitively high energy expenditure that is required for stress adaptation. It is therefore noteworthy that Aspergillus species, which appear to have an extraordinary capacity for energy generation, also have exceptional tolerances to diverse stresses and an ability to out-compete other microbes, thereby dominating their respective habitats. In other words, common physicochemical and/or thermodynamic constraints determine this limit, irrespective of phylogeny.

32It may be that a greater understanding of microbial stress biology in relation to low water-activity habitats can lead to further improvements in food preservation, biological control, management of soil microbiology in arid regions and interventions to enhance crop plant/mycorrhizae (and other plant/microbe) interactions or plant–insect interactions mediated by osmophilic microbes inhabiting plant nectar; thereby enhancing plant conservation, crop production and, ultimately, global food security. The majority of studies carried out to determine limits of microbial solute tolerance have focused on culturable species of weed-like and/or copiotrophic microbes (Figures 1, 2, 3;; ). As a consequence, little is known about water-activity limits for oligotrophic and/or slow-growing species that may undergo a single cell division over a period of decades or longer. Other questions remain outstanding, such as: what are the metabolic factors that ultimately limit energy generation as well as the synthesis and retention of compatible solutes; can in vitro studies of transcription or culture-independent techniques to detect metabolic activity offer insights into the in-situ water-activity limits for microbial communities in hostile environments; why are halophilic eukaryotes and osmotolerant/philic prokaryotes relatively intolerant to high solute-concentrations; what are the phenotypic differences between strains of a single bacterial species isolated from salt-rich environments, that are unable to grow at high-sugar concentrations, and those isolated from sugar-rich substrates that can do so; how do the limitations of certainty and accuracy of water-activity quantification compare with those of chaotropicity, pH and temperature determination in relation to the sensitivity of the cellular system; can further manipulations of environmental chemistry (e.g. ) lead to microbial life at solute concentrations currently thought to be prohibitive; and could synthetic biology be utilised to obtain microbial cell(s) capable of completing life cycles under hitherto nonpermissive conditions? We are intrigued to see whether further studies to address some of these questions can potentially lead to the documentation of life processes at <0.600 aw.