Rainfall is the most obvious source of liquid water for Earth’s terrestrial ecosystems. However, some hot deserts experience decadal periods between precipitation events ^[1][34], and some extremely cold deserts, such as the Antarctic McMurdo Dry Valleys, very rarely receive rainfall ^[2][35]. Precipitation events in the latter are in the form of occasional snowfall, where much of the settled snow does not melt but is lost back to the atmosphere via sublimation ^[3][36]. Surface snowmelt may only wet the upper 0.5–1.0 cm of soil. Snow deposition on dry mineral surfaces is thought to provide water for shallow subsurface (2 to 5 mm depth) cryptoendolithic microbial communities ^[4][38].

The extent to which, and duration for which, soils retain water after precipitation is highly relevant to the capacity for soil microbiome functionality. Where present on desert soil surfaces, biological soil crusts (BSCs) are important in retaining precipitation that would otherwise be lost by evaporation or rapid infiltration to subsurface soil ^[5][39]. Both soil structure and composition are important in water retention ^[6][40] where, for example, clays such as sepiolite, palygorskite, and smectite, often associated with arid soil environments and evaporitic rock substrates, exhibit a very high water-holding capacity (250% of wt ^[7][8][9][16,41,42]). Organic substances in soils, such as plant biomass, humic acids, and particularly microbial extracellular polymeric substances (EPSs), are very hygroscopic. Water is less readily lost to evaporation or filtration in saline soils ^[10][43] and, while water in saline soils is generally considered to be less available for plant uptake, the hygroscopic matrices of microbial biofilms may compete effectively for salt- and clay-bound water ^[11][44].

Although long-term precipitation patterns in hyper-arid deserts, coupled with very high evaporation rates, may suggest that persistent microbial communities are not sustainable ^[12][45], there is clear evidence that specialized niches in below-ground soils and endoliths offer ephemeral microbial habitats after significant stochastic rainfall episodes ^{[13][14][15][16]}[14,46,47,48].

Icecaps and other glacial masses might be considered as desert-like ecosystems, in that the consistently low temperatures typically characterizing all but the surface horizons of such environments can ensure that water is rarely, if ever, present in bioavailable liquid form. However, liquid water can exist on both the upper and lower surfaces of many glacial masses. For example, biologically active surface microbial communities are associated with the localized melting of many glacial masses, snow algae and cryoconite communities being two well-known examples. Both communities acquire liquid water from the frozen snow/ice substrate through the same mechanism: a reduction in albedo (reflected incident light) corresponding to an increase in the absorption of solar radiation (solar gain) ^[17][49]. In snow and ice algae (mostly Chlamydomonadales and Zygnematales, respectively), astaxanthin- and purpurogallan-rich pigmented cells absorb solar radiation with an associated heat-gain that melts frozen water (snow/ice) in the immediate vicinity of the cells ^[17][49]. Similarly, cryoconite holes are created by the deposition of dust particles and rock fragments/pebbles on snow and ice surfaces. The dark mineral particles are warmed by the absorption of solar radiation, and generate melt-wells in the ice surface, which rapidly become rich oases of metabolically active microbial life ^[18][50]. On the undersides of glacial masses, heat, generated by friction between the ice mass and the underlying rock surface, generates meltwater; sub-glacial liquid systems support substantial microbial populations ^[19][51].

In desert soil ecosystems, water may be available in the form of dew or fog inputs. Dew formation results from a balance between atmospheric relative humidity and temperature, where condensation occurs on surfaces when the balance exceeds the ‘dew point’ ^[20][52]. Dewfall is a relatively common occurrence in many hot semi-arid and arid (but not hyper-arid) deserts (up to 200 days per annum in some Negev Desert locations ^[21][53]) and is likely to make a significant contribution to the water input budgets of the surface (0–1 cm depth) soils ^[22][54] and their microbiomes. Dew water input in the Badain Jaran Desert (northwest China) over a 5-month period amounted to a total of 3.4 mm, averaging 0.06 mm d^{−1 [23]}[55]. Dew is thought to be an important supplementary water source for desert vegetation. Soil surface lichens and surface microbiological communities (BSCs) have been shown to benefit from dew water inputs ^[24][25][26][56,57,58]. In the Atacama Desert, characteristic microkarstic features found on the surface layers of calcite rocks indicated that dew deposition might be an important source of liquid water for the endolithic communities inhabiting the calcite rock ^[27][30].

Fog water inputs are restricted to a limited number of coastal desert ecosystems, most notably the Namib (Namibia, south-western Africa) and Atacama (north-west Chile) deserts ^[1][28][34,59]. In the Namib Desert, fog generated off-shore by moist air over the Benguela Current is driven inland at night by onshore wind-flows, and can penetrate inland up to around 60 km ^[1][34]. At the coast, fog events are frequent (est. 40% events per annum ^[29][60]) and sustain extensive and well-characterized lichen fields ^[30][61]. This conclusion was recently supported by an extensive remote sensing drone survey of coastal Namib Desert lichen fields using advanced photogrammetry, which showed that Xanthoparmelia and Stellanrangia spp. preferentially colonized ocean-facing rock surfaces, i.e., the direction from which fog originates ^[31][62].

Further inland, water capture from less frequent fog events sustains both specialist plants (e.g., Speargrass (Stipagrostis sabulicola) ^[32][63] and insect species (e.g., dune tenebrionid beetles ^[33][64]). However, salts and clay minerals may remain hydrated. In the Tarapacá Region of the Atacama Desert, gypsum crusts colonized by epilithic lichens and endolithic bacteria benefit from the coastal fog, called ‘camanchaca’, charged with humid air with a relative humidity close to 100% ^[34][65]. These frequent fog events ^[28][59] result in the continuous deliquescence of halite nodules in local salars, providing constant liquid water to the communities inhabiting the salt rocks ^[35][36][66,67].

The role of fog water inputs in supporting soil microbiomes is much less clear. There is growing evidence that fog capture contributes bioavailable water for cryptic hypolithic communities ^[37][69], and supports microbial communities associated with desert plant rhizosheaths ^[38][70]. However, with the exception of some specialized plant species such as S. sabulicola, where captured fog water is channelled by specially adapted leaf structures down to the root zone ^[32][63], fog-derived water inputs are unlikely to penetrate, as liquid, to more than a few millimetres depth in desert pavements.

Even in the driest of deserts, water is present, but not always bioavailable. In cold deserts, permafrost layers, which may exist a few tens of centimetres or a few metres from the desiccated surface, are a potential source of bioavailable water for soil microbiomes ^[39][71]. The hydrologically active zone, the horizon above the permafrost that thaws and refreezes with seasonal cycles, provides a saturated soil profile at some distance below the desiccated surface horizon (Figure 1b). In some Antarctic and Arctic regions, such as the Windmill Islands (Ferrari, pers. comm.) and the Sør Rondane Mountains in East Antarctica, temperature fluctuations are so high during the summer months that surface soils may be subject to freeze–thaw cycles on a daily basis ^[40][72].

The presence of a saturated atmosphere at depth and a low surface relative humidity will generate a strong thermodynamic driver for the upward diffusion of high relative humidity (RH) water vapour ^[41][42][73,74], potentially available to shallow sub-surface microbial communities (see below). In hot deserts, groundwater may be able to fulfil a similar role, even though such subterranean liquid flows may be found tens or hundreds of metres below ground level. Measurements of soil atmospheric relative humidity values in shallow soil depth profiles in the central (hyper-arid zone) Namib Desert and the Atacama Desert are strongly suggestive of the upward transport of subsurface water vapour ^[43][75], whether derived from deep subterranean groundwater or residual water from infrequent rainfall recharge.

Irrespective of the origins of water vapour in desert soils, nocturnal distillation may also provide a mechanism for generating condensed (liquid) water in the upper soil horizons ^[6][40], under conditions where nocturnal soil surface temperatures are sufficiently low to yield measurable increases in shallow sub-surface soil water content ^[6][40].

It has been well-established that fruticose lichens have the capacity to adsorb water from moist air ^[44][76], without the need for condensation processes. The capacity for microorganisms to acquire cellular water directly from the atmosphere (i.e., H₂O(g)) is not well established, although the permeability of the cell membrane to water molecules and the hypertonicity of the cellular cytoplasm suggest that this process is both physically and thermodynamically feasible. The close correlation between desert soil surface and shallow sub-surface relative humidity values suggests that the atmosphere of the near-surface (0 to 2 cm depth) soil horizon is in equilibrium with the above-surface atmosphere, and that water is potentially available to the shallow soil microbial communities that are able to adsorb it. One recent study suggested that, each night, an equivalent of ∼30  μm rainfall may enter the soils of the hyper-arid core of the Atacama Desert via atmospheric water vapour adsorption ^[45][77].

Whether microbial cells directly adsorb atmospheric moisture or not, there is good evidence that the biofilm structures in which most soil organisms reside do have this capacity. Most soil microbial communities are embedded in EPS matrices ^[46][78], composed of compositionally heterogeneous high-molecular-weight glycan polymers ^[47][79] that can constitute up to 90% of the biomass of a biofilm ^[48][80]. Such compounds are rich in free hydroxyl (-OH) and amino (-NH₂) groups, both of which are strongly hydrophilic and contribute to the water-holding capacity of EPSs ^[49][81].

In addition, bacterial production and excretion of EPS constituents is strongly stimulated by exposure to stress, particularly sub-lethal heat ^[50][82] and osmotic stresses ^[51][7], and has been implicated in water retention in desert soil biocrust communities ^[52][83]. It has also been demonstrated that the hygroscopic properties of EPSs facilitate both the acquisition and retention of water from the atmosphere ^[53][54][84,85].

Microorganisms accumulate a wide array of low-molecular-weight organic solutes (including monosaccharide and oligosaccharide sugars, polyols, amino acids and their derivatives, ectoines and betaines ^[55][86]), at least some of which have been implicated in cellular responses to osmotic (including desiccation) stress ^[55][86]. The hygroscopic disaccharide trehalose, which is accumulated intracellularly in many organisms in response to desiccation ^[56][87], is capable of adsorbing and retaining water at atmospheric relative humidity values above 50% ^[57][88]. This mechanism of water acquisition by desiccated microbial cells is, at least theoretically, feasible in shallow subsurface microbial communities in both hot ^[58][89] and cold desert soils.

Mineral (and other) surfaces can acquire and retain thin films of water ^[59][60][61][90,91,92] that may exist from >1 mm depth down to molecular monolayers ^[62][93]. While the water in thicker surface films, typically acquired because of the presence of hygroscopic salts ^[63][94] or solutes ^[64][95], is thought to be bioavailable ^[65][96], it is uncertain whether thin surface water layers (<3 molecules thick) are in a liquid phase ^[65][96].

The presence of minute salt crystals on surfaces, derived from deposited sea spray in coastal deserts, can lead to deliquescence events (i.e., the formation of liquid brine as the salt absorbs water from the atmosphere). The phenomenon of salt deliquescence may be a primary driver for microbial life in much of the Atacama Desert, where cells and communities are typically active in thin layers of brine inside halite rocks ^[66][67][29,97] or in the NaCl-rich subsurface ^[67][97].

There is also recent evidence that endolithic microorganisms can access water from hydrated minerals ^[68][98]. Cyanobacteria growing as biofilms in gypsum (CaSO₄.2H₂O) minerals induce mineral dissolution accompanied by water extraction and the transformation of gypsum to anhydrite (CaSO₄). Given that many coastal desert soils, such as in the Namib and Atacama Deserts, are gypsum-rich, this phenomenon may be of considerable biological importance.

Apart from the various exogenous sources of biologically available water, all microbial cells have the capacity to generate water endogenously. The oxidative heterotrophic metabolism of carbohydrates is hydro-genic (water-generating: C₆H₁₂O₆ + 6O₂ = 6CO₂ + 6H₂O), although it is unclear what proportion of intracellular water this (and related) metabolic processes may contribute. Oxygen stable-isotope analyses have suggested that, in actively growing cells, 70% of microbial intracellular water may be metabolically derived ^[69][99]. Notably, many intracellular metabolic processes are water-generating, including ligation, condensation, polymerization, and related reactions, in addition to oxidative metabolism.

Under desiccated conditions, when many cells may be in an anhydrobiotic state ^[69][70][99,100] with cellular activity limited to basal metabolic processes ^{[70][71][72][73][74]}[100,101,102,103,104], hydro-genesis from carbohydrate metabolism may be limited. However, there is growing evidence that a recently discovered chemotrophic metabolism may even supplement cellular water in desiccated cells. Atmospheric trace gas assimilation, particularly hydrogen oxidation, has been observed in both hot and cold desert soil microbiomes ^{[75][76][77][78][79]}[105,106,107,108,109]. This process, which uses newly described clades of assimilatory Ni–Fe hydrogenases ^[77][107] and is present in a wide range of aerobic soil bacterial phyla ^[78][79][108,109], not only provides the energy and reductant needed to support carbon fixation; the process is also hydro-genic (2H₂ + O₂ = 2H₂O).

The process of trace-gas-dependent hydro-genesis in microbial cells has the potential to realign current paradigms on the functional status and capacity of desert soil microbiomes. Recent transcriptomic data on microbial cellular function in hyper-arid soils suggest that a subset of microbial taxa retains some functionality under desiccated conditions ^[80][81][110,111]. Similarly, metaproteomic analyses of hyper-arid Antarctic Reeve Hill (Casey Station) soils show evidence of active expression of Ni–Fe hydrogenases and RuBisCO ^[75][105].