Hypolithic microbial communities: between a rock and a hard place

Arid environments and the hypolithic habitat

Aridity has been a widespread global phenomenon since approximately 1.8 Ga and arid environments (here termed deserts or drylands) currently comprise over 30% of the Earth's land mass, representing the most extensive terrestrial biome (Fig. 1) (Thomas, 1997; Laity, 2008). A widely accepted definition for deserts reflects water deficit by defining deserts as experiencing a precipitation to potential evapotranspiration ratio (P/ETP) of less than 1 (United Nations Environment Program, 1992). Four zones of aridity are recognized using this criterion: sub‐humid P/ETP 0.5–< 0.65; semi‐arid P/ETP 0.2–< 0.5; arid P/ETP 0.05–< 0.2; hyper‐arid P/ETP < 0.05. Hot deserts display a mean annual temperature of > 18°C, and cold deserts < 18°C, while polar deserts exhibit year‐round cold climates with maximum temperatures below freezing (polar frost), or from 0 to 10°C (polar tundra) (Peel et al., 2007). Arid environments are also typically low‐energy systems with low soil‐nutrient levels, and higher plant and animal life tends to be ephemeral, or undetectable in extreme hyper‐arid deserts (Thomas, 1997). In such challenging environments edaphic microbial colonization may represent a major biotic component of the ecosystem (Cary et al., 2010; Pointing and Belnap, 2012), along with patchy distribution of invertebrate colonization (Convey and Stevens, 2007).

Desert pavement, comprising surface pebbles or rocks embedded in and covering the soil surface, accounts for over 50% of all desert terrain (Laity, 2008) (Fig. 2). Translucent pebbles, such as quartz and marble, have been recorded in pavements of every major desert on Earth (Bahl et al., 2011). The ventral surfaces of these rocks provide a substrate for development of cryptic microbial biofilms, and these are termed hypoliths (Golubic et al., 1981). These are viewed as distinct from endolithic colonization that occurs within the pore spaces(cryptoendolithic) or cracks and fissures (chasmoendolithic) of porous and weathered rocks (Golubic et al., 1981), and are not the subject of this review article. Hypolithic colonization can be viewed as a stress avoidance strategy, where the overlying mineral substrate provides protection from incident UV radiation and excessive photosynthetically active radiation (Schlesinger et al., 2003; Cowan et al., 2010; Wong et al., 2010), thermal buffering in hot (Warren‐Rhodes et al., 2006) and cold desert (Broady, 1981), protection from freeze–thaw events (Cockell and Stokes, 2006), physical stability (Wong et al., 2010), and by enhancing moisture availability over the surrounding soil (Warren‐Rhodes et al., 2006). Studies have shown that quartz effectively filters UV irradiance while still providing sufficient light for photosynthesis (Schlesinger et al., 2003; Cowan et al., 2010; Wong et al., 2010). The average size of individual colonized pebbles increases as aridity increases, suggesting a role in water retention (Warren‐Rhodes et al., 2006). Hypoliths typically appear as a light to dark green biofilm on upturned pebbles and are readily identified visually in the field (Fig. 2).

Biodiversity of hypoliths

Hypolithic microbial colonization was first recorded in desert pavements during the 1960s and 1970s (Cameron and Blank, 1965; Friedmann and Galun, 1974). Subsequent studies have used microscopy and culture‐independent methods to identify cyanobacteria as the most common phylum observed in hypolithic colonization in deserts worldwide, including the Atacama Desert (Warren‐Rhodes et al., 2006; Pointing et al., 2007), Namib Desert (Budel and Wessels, 1991), Negev Desert (Berner and Evanari, 1978) and Taklimakan Desert (Warren‐Rhodes et al., 2007), plus semi‐arid regions in Australia (Tracy et al., 2010) and the Mojave Desert (Schlesinger et al., 2003). Cyanobacterial hypoliths have also been recorded in arid Arctic tundra (Cockell and Stokes, 2006), Tibetan tundra (Wong et al., 2010), Antarctica's McMurdo Dry Valleys and other ice‐free inland areas (Broady, 1981; 2005; Smith et al., 2000; Wood et al., 2008; Pointing et al., 2009) and maritime Antarctic sites (Cockell and Stokes, 2004). A recent worldwide survey using a multi‐locus phylogenetic approach for environmental sequences demonstrated that cyanobacterial hypoliths occurred on quartz in major deserts spanning every continent on Earth (Bahl et al., 2011). Species richness estimates are significantly higher in non‐polar as compared with polar deserts (Caruso et al., 2011).

Microscopic observations and analysis of environmental sequences from DGGE, ARISA and clone libraries have suggested an interesting dichotomy in cyanobacterial composition, where hypoliths from warmer deserts appear to comprise largely coccoid Pleurocapsalean cyanobacteria of the genus Chroococcidiopsis (Friedmann and Galun, 1974; Warren‐Rhodes et al., 2006; 2007; Bahl et al., 2011) and harbour greater cyanobacterial diversity, whereas those in extreme cold and polar deserts appear to support a higher abundance of filamentous oscillatorian cyanobacterial morphotypes and support lower cyanobacterial richness (Smith et al., 2000; Broady, 2005; Cockell and Stokes, 2006; Pointing et al., 2009; Wong et al., 2010) (Fig. 3). Recently, rRNA‐directed massively parallel pyrosequencing of hot and cold desert hypoliths from two locations in China has provided strong support for the validity of this delineation (Bahl et al., 2011), although interrogation of additional sites is required to confirm this hypothesis on a broader scale. As this pattern occurred across deserts experiencing similar levels of xeric (moisture limitation) stress but greatly contrasting thermal stress, this may in part reflect selection based on thermal and/or freeze–thaw tolerance, although this hypothesis too remains untested.

Diazotrophic cyanobacterial phylotypes have been recovered from most non‐polar deserts investigated, mainly indicating Nostoc sp. (e.g. Warren‐Rhodes et al., 2006; Pointing et al., 2007; Tracy et al., 2010; Wong et al., 2010). Diazotrophic cyanobacteria appear to be generally absent from extreme polar desert hypoliths, although Nostoc phylotypes have recently been recovered from hypoliths in low‐lying wetter Dry Valleys sites in Antarctica (Cowan et al., 2011a) and Nostoc‐dominated microbial mats occur with patchy distribution around the edges of some Antarctic lakes (Wharton et al., 1983). It is also possible that the keystone hypolithic taxon Chroococcidiopsis may be capable of diazotrophy, given laboratory demonstration of acetylene reduction by endolithic strains of this cyanobacterium (Boison et al., 2004). An alternate but significant source of diazotrophy in polar deserts may be non‐cyanobacterial and derive from alphaproteobacteria (Pointing et al., 2009).

Analysis of bacterial assemblages based upon matching 16S rRNA gene tRFLP profiles to extensive sequence libraries created for hypolithic communities from 92 samples derived from deserts worldwide revealed the Cyanobacteria comprised 47–96% of all recoverable bacterial phylotypes (Fig. 3) (Caruso et al., 2011), and hot deserts supported significantly higher abundances of heterotrophic bacteria relative to photoautotrophs than cold deserts. Assuming no differences in exogenous nutrient input, this implies that productivity is higher in hot deserts and therefore capable of supporting greater biomass and trophic complexity than in cold deserts. This hypothesis could be tested using in situ measurements and offers a potentially fruitful avenue for future research. Culture‐independent molecular studies have revealed a phylogenetically diverse heterotrophic bacterial component, but some clear trends are apparent. The Actinobacteria, Alphaproteobacteria and Gammaproteobacteria (along with photoautotrophic cyanobacteria) appear ubiquitous to all hypoliths (Pointing et al., 2007; 2009; Wong et al., 2010), and these may represent keystone taxa essential to community assembly in hypolithic communities.

Other taxa appear to be more abundant or specific to either hot/cold or polar deserts, for example the Deinococci appear to occur frequently in warm and hot deserts but are not common in polar hypoliths (Warren‐Rhodes et al., 2006; Pointing et al., 2007; 2009; Lacap et al., 2011), whereas lichen associates have only been encountered in polar hypoliths [some fungal phylotypes were recorded for Mojave Desert hypoliths (Schlesinger et al., 2003), but these cannot be assigned as lichen mycobionts]. Hypoliths support up to 16 bacterial phyla, but others are encountered with relatively low frequency. It is of interest that in a comparative study of hot and cold deserts in China, phylogenetic analysis revealed lineages of Deinococci related to mean annual temperature (Pointing et al., 2007), and this may further suggest thermal tolerance is a selection criterion for hypolithic taxa.

Few studies have attempted multi‐domain diversity assessments. The presence of eukaryotic algal morphotypes (Broady, 2005; Cockell and Stokes, 2006) and phylotypes (Smith et al., 2000; Pointing et al., 2009) has been recorded for polar hypoliths, and culture‐independent molecular studies have also revealed fungal phylotypes in arid Mojave Desert hypoliths (Schlesinger et al., 2003). A real‐time quantitative PCR approach was employed to quantitatively estimate the entire hypolithic community in Tibetan and Antarctic Dry Valleys hypoliths. This revealed that eukaryal and archaeal taxa comprise less than 5% of recoverable phylotypes after correcting for rRNA copy number of target taxa between domains (Pointing et al., 2009; Wong et al., 2010). Eukaryotic signatures could not be recovered from hypoliths of the most extreme hyper‐arid Atacama and Taklimakan Deserts (Warren‐Rhodes et al., 2006; Pointing et al., 2007). Similarly, the archaea have not been encountered in hypoliths from the most hyper‐arid non‐polar deserts (Warren‐Rhodes et al., 2006; Pointing et al., 2007) or from inland Antarctic Dry Valleys locations (Pointing et al., 2009), although they have been recovered from Tibetan hypoliths (Wong et al., 2010) and from maritime influenced Antarctic Dry Valleys sites (Wood et al., 2008). Whether the observation that archaea are not recoverable from the hypolithic niche in the more extreme arid locations is applicable on a global scale remains untested, but would provide an important insight into the limits of life for a domain that has been regarded as poorly tolerant to xeric stress (Rothrock and Garcia‐Pichel, 2005) relative to the Bacteria.

It is also possible that under certain conditions hypoliths may develop with a phototrophic taxon that is non‐cyanobacterial, perhaps reflecting variations in micro‐niche abiotic variables. A recent study of hypoliths from the hyper‐arid core of the Atacama Desert in Chile revealed that a reddish tinted hypolithic biofilm was dominated by anoxygenic phototrophic Chloroflexi, although the cyanobacterial taxon Chroococcidiopsis and other common hypolithic bacterial phylotypes were also recovered (Lacap et al., 2011). Under certain conditions hypoliths in Miers Valley (Antarctic Dry Valleys) become dominated by moss biomass, although the cyanobacteria‐dominated biofilm also remains in close contact with the rock surface (Cowan et al., 2010).

Physiological challenges

The physiology of hypolithic systems is still poorly understood. Higher taxonomic ranks retain some coherence in an ecological context (Phillippot et al., 2010) and thus broad taxonomic identification from environmental rRNA genes offers some insight to physiology that may guide future research efforts, particularly the degree to which different community assemblies share similar functional capabilities in hot, cold and polar deserts. Hypolithic cyanobacteria have been cultivated (Smith et al., 2000; Warren‐Rhodes et al., 2007), but have not been subjected to rigorous physiological examination, largely due to problems in obtaining axenic cultures of hypolithic cyanobacteria. Hypolithic communities may also harbour heterotrophic taxa with novel phenotypic traits (e.g. Le Roes‐Hill et al., 2009). As a result, there is significant opportunity for physiological studies to improve our understanding of hypolithic taxa. In the meantime, analyses of environmental samples using ‘omics’‐based approaches, at the genomic, transcriptomic and proteomic levels, are ongoing in several laboratories. Such studies aim to address fundamental issues related to how hypoliths function in the environment and respond to seasonal and diurnal stress, and to stochastic and chronic disturbance.

The major stressor thought to impact hypolithic organisms is moisture stress, and studies have indicated that the keystone taxon Chroococcidiopsis may possess exceptional ability to withstand and recover from desiccation. This has generally been investigated by measuring response to ionizing radiation, as it is thought to elicit the same cellular repair pathway (Billi et al., 2000; Billi and Potts, 2002; Cockell et al., 2011). The Deinococci are known to be the most radio‐tolerant bacterial group (De Groot et al., 2005; Daly, 2009) and other taxa commonly recovered in hypolithic communities such as Rubrobacter also have cultivated radio‐tolerant strains (Ferreira et al., 1999). Desiccation tolerance and repair pathways may emerge as a common trait in hypolithic organisms as our knowledge improves. Elucidating the pathways involved may shed new light on tolerance and repair related to xeric (and radiation) stress.

Hypolithic organisms must also be well adapted to tolerate freeze–thaw events, extreme heat, and be able to exploit relatively narrow windows of favourable growth conditions, with rapid desiccation and rehydration responses. Indeed, in the hyper‐arid core of the Atacama Desert it has been estimated that hypoliths experience < 75 h a year on average when moisture, temperature and light intensity are sufficient to allow photosynthesis (Warren‐Rhodes et al., 2006). In semi‐arid regions of Australia the conditions for photosynthesis may exist for a 10‐fold longer duration each year (Tracy et al., 2010). In polar deserts hypoliths must also be able to tolerate long winters without light at very cold temperatures (Wynn‐Williams, 1990). As a result, it is estimated that hypoliths may be very long‐lived biofilms. Indeed, a steady‐state carbon isotope probing of Atacama Desert hypoliths revealed a positive correlation between aridity and age of biofilms, implying that hypoliths experiencing less moisture were longer lived, with age estimates of approximately 12 000 years for the most extreme arid location on Earth (Warren‐Rhodes et al., 2006).

A further adaptation that may be key to successful hypolithic colonization is the secretion of extracellular polymeric substances (EPS). Hypolithic biofilms in both hot (Warren‐Rhodes et al., 2007) and cold polar deserts (Pointing et al., 2009; Cowan et al., 2010; Wong et al., 2010) are typically cemented to the ventral surface of the substrate by an EPS matrix (Fig. 4). While the composition of the EPS has yet to be fully characterized, observations of natural and cultivated biofilms indicate that it has a strongly hygroscopic nature (Liu et al., 2008), and the properties of this EPS may contribute to water acquisition and retention and thus confer an adaptive advantage. It may also be fruitful to investigate adaptation at the cellular level, including the production of osmolytes or other stress‐response molecules.

Ecology of hypoliths

Several studies indicate that hypolithic communities in both hot and cold deserts are clearly distinct from the surrounding soil communities (Smith et al., 2000; Schlesinger et al., 2003; Warren‐Rhodes et al., 2006; Pointing et al., 2007; 2009; Wong et al., 2010). Their ecological importance is therefore related to the frequency of colonization in a given desert pavement niche. Certainly on a landscape scale, the role of hypoliths may be greater than previously considered by desert ecologists. A survey of hypolith coverage across three maritime Antarctic dry valleys showed that these communities comprised only 0.024% of total surface area and yet represented a significant proportion of total soil biomass (Cowan et al., 2011a). This pattern is also reflected in studies of hyper‐arid hot and cold deserts (Warren‐Rhodes et al., 2006; 2007).

It is likely that hypolithic biomass may act as a reservoir of cyanobacterial and other microbial innocula for dispersal to other edaphic niches. We have observed in field studies that hypolithic biofilms can develop quite extensively relative to observable colonization in surrounding soil and rocky niches. In addition, we have observed that under moisture sufficiency, hypolithic biomass extends into soil and acts as a cohesive force to stabilize the soil matrix. This appears partly due to growth of eukaryal taxa, particularly mosses in cold and polar deserts (Cowan et al., 2010; Wong et al., 2010), but may also be contributed by EPS secreted by cyanobacteria (Fig. 5). An interesting question therefore arises as to whether hypoliths are important in deserts undergoing transition due to climate change, where they might act as a positive feedback in enhancing soil stability. This would be of great relevance to deserts where land‐use change is planned or under way, and may shed light on whether hypoliths act as sites of ‘nucleation’ for transitions to more stable soil environments.

In the most extreme hyper‐arid deserts hypoliths are the only detectable source of primary productivity, and are thus a key to understanding carbon flux under extreme xeric stress (Warren‐Rhodes et al., 2006; 2007). In less arid conditions colonization frequency is orders of magnitude greater, but higher plants may also be present (Schlesinger et al., 2003; Tracy et al., 2010) and this reduces the relative contribution of hypoliths to desert carbon flux. Measurements of primary productivity by maritime polar hypoliths, however, indicate that their rates at least equal that of above‐ground plant biomass (Cockell and Stokes, 2004). A laboratory study of Mojave Desert hypolithic cyanobacteria indicated that net carbon exchange declined dramatically when relative humidity fell below 93% (Schlesinger et al., 2003). A field and laboratory study of Australian hypoliths indicated that soil moisture levels of 15% (by mass) were necessary to support hypolithic photosynthesis, within a temperature range of 8–42°C (Tracy et al., 2010). Clearly, there are significant gaps in our knowledge, particularly with regard to field measurements of productivity and other geobiological processes on a temporal scale with changing environmental conditions. Hypolithic communities from the Mojave were concluded to have a role in dinitrogen fixation in hot deserts based upon laboratory measurements of acetylene reduction (Schlesinger et al., 2003). A recent study of acetylene reduction by Antarctic hypoliths in situ suggested that hypolithic communities probably constituted the major source of N input into lake‐free polar deserts (Cowan et al., 2011a). Further work to resolve hypolith activity under natural conditions is identified as a major goal for future studies of these systems.

The source of moisture input to hypolithic systems likely varies depending on the environment. In the Atacama and Namib Deserts, where the majority of hypolith studies have been carried out, topography and a coastal location make fog a major source of moisture and in some cases fog may be the only detectable source of moisture to hypoliths (Azua‐Bustos et al., 2011). Elsewhere in inland deserts the bioavailable water budget may be derived solely from infrequent rainfall events, but also possibly via upward transport of water vapour from groundwater although this latter source remains unproven. Possibly, one of the greatest enigmas is the source of moisture to hypoliths in the hyper‐arid McMurdo Dry Valleys of Antarctica. As snowfall is very infrequent and sublimes rapidly, a significant source of moisture may also be upward transfer of water vapour through soils from permafrost (Stomeo et al., 2012). Whether the quartz substrate and/or the EPS content of the hypolith biofilm play a role in attracting or retaining moisture has yet to be resolved, but such processes are probably key to understanding how these communities develop where surrounding soils cannot support significant biomass.

Information on how quickly hypolithic colonization proceeds and what successional patterns occur is currently lacking. Some studies have indicated that environmental conditions necessary to support photoautotrophy may exist for very limited periods as a result of interactions between temperature, light and moisture availability (Warren‐Rhodes et al., 2006; Tracy et al., 2010). A series of field‐based in situ monitoring studies for primary productivity (using for example PAM fluorometry) would add further insight.

Further understanding of the physiology of hypolithic systems may also be gained by considering those occurring under less stressful conditions. For example hypolithic biofilms also develop on various translucent substrates in tropical terrestrial and aquatic environments under moisture sufficiency and without exposure to freeze–thaw events (C. P. McKay and S. B. Pointing, field observations). Resolving patterns of community assembly in the absence of obvious stressors will help in understanding the ecology of hypolithic biofilms.

Ecological applications

Hypoliths as biomarkers of aridity and climate change?

A number of studies report estimates for colonization by hypoliths, usually expressed as a percentage of colonizable translucent pebbles. Up to 100% colonization frequencies have been reported from ‘casual’ surveys in the Mojave Desert (Schlesinger et al., 2003), Australia (Tracy et al., 2010) and maritime Arctic and Antarctic (Cockell and Stokes, 2006). Detailed landscape‐scale ecological surveys have also been reported, and these encompass hot, warm, cold and polar deserts. Such surveys typically include hundreds to thousands of samples recorded in spatially randomized surveys covering hundreds of square kilometres (Warren‐Rhodes et al., 2006; 2007; Pointing et al., 2009; Cowan et al., 2011b). A clear trend of decreasing colonization frequency with increasing aridity for all deserts regardless of thermal stress has emerged from these surveys. For example, in the hot Atacama Desert colonization decreases from 28% to less than 1% in a latitudinal gradient from arid to hyper‐arid (Warren‐Rhodes et al., 2006). The arid high‐altitude tundra of Tibet supported 36% colonization (Wong et al., 2010). Colonization in the hyper‐arid Taklimakan Desert and Turpan Depression in China revealed colonization of 12–< 1% along an aridity gradient with a concomitant reduction in species richness (Pointing et al., 2007; Warren‐Rhodes et al., 2007), and in hyper‐arid polar desert colonization was approximately 5% (Pointing et al., 2009). The latter study also revealed that isolated areas of stochastic moisture input due to snowmelt increased colonization levels to those encountered in other less arid regions. A synthesis of all published values for percentage colonization by hypoliths (including quartz and other substrates) reveals a strong trend, where onset of hyper‐aridity is marked by a linear relationship between available liquid water and colonization (Fig. 6). The linear nature of this relationship opens interesting possibilities as to whether hypolithic colonization may be a useful and predictive bio‐indicator for aridity. This has high immediacy given that deserts are the most impacted ecologically by climate change of any terrestrial biome (Safriel, 2006). Hypolith colonization patterns may have potential as a useful predictor in situations where onset of wetter conditions (i.e. increased colonization) occurs, although the longevity of hypoliths may limit their use as biomarkers of increasing aridity (i.e. extinction) in the short term. A further outcome of considering landscape scale hypolith colonization may be that arid areas with different surface geology within the same climatic region may potentially influence differential responses of edaphic biota to climate change.

Concluding remarks

The importance of arid landscapes in terms of their abundance and relevance to future land use has led to increased awareness of desert systems. Arid ecosystems are more at risk from climate change than any other biome and so improving understanding of organisms and communities existing at the xeric limit for life should be given high priority. Omics‐based technologies have the potential to yield major advances in our understanding of the functional capacity of microbial communities and their adaptive potential. Such studies must be complemented by rigorous field‐based investigations of microbial distribution, activity and contribution to biogeochemical processes, in order to fully understand ecosystem function. Existing studies have contributed significant amount of genetic and ecological information to public databases and it is hoped that future studies will continue to do so and allow hypoliths to become increasingly used as a model system to test hypotheses in microbial ecology.