INTRODUCTION
Biological soil crusts (biocrusts) are communities anchored by primary producers such as cyanobacteria, mosses, algae, and lichens, which are accompanied by diverse bacteria, archaea, and fungi (
1). In deserts and drylands, biocrusts occupy the first few millimeters of the soil surface, where they perform multiple functions, including nutrient capture and erosion control (
2,
3). Globally, biocrusts cover approximately 12% of the Earth’s terrestrial surface (
4) and contribute significantly to soil stability, hydrology, and carbon and nitrogen cycling at ecosystem scales (
1). At the microhabitat scale, drylands sometimes support hypolithic niches on the ventral side of semitranslucent stones (usually quartz) embedded in the soil surface (
5). Hypoliths can occur in hyper extreme habitats too harsh to support exposed biocrusts (
6–9), but they are also found as dispersed microsites within areas supporting more extensive biocrusts (
10).
Dryland soil organisms are physiologically specialized for survival in polyextreme environments characterized by challenges such as high (and low) temperatures, desiccation, intense UV radiation, and nutrient limitation (
11). Though both biocrusts and hypoliths experience extreme conditions, the environment in hypolithic microhabitats is buffered compared to biocrusts. In the Mojave Desert, quartz stones reduce light transmission by ~98%, decrease daytime high temperatures by ~2°C, and increase relative humidity by nearly 100% (
12). To survive environmental extremes, organisms in the community are typically poikilohydric, capable of equilibrating to the ambient relative humidity of their environment and suspending all metabolic activity in a dried and quiescent state. Once water is reintroduced, poikilohydric organisms resume metabolic activity almost instantaneously through a combination of cellular protective mechanisms deployed during drying (e.g., reactive oxygen species (ROS) scavenging, compatible solutes, mRNPs) and repair mechanisms initiated upon rehydration (
13–16). For larger biocrust organisms (e.g., mosses) that may require extensive cellular repair upon rewetting from the desiccated state, the process of rehydration is energetically costly and creates a carbon deficit that must be recovered through a period of photosynthetic activity (
17,
18). Thus, while biocrusts are physiologically specialized for environments with low precipitation, they are sensitive to the frequency, timing, and duration of hydration events (
19).
In habitats where biocrusts occur, drying events happen quickly relative to the time required for poikilohydric organisms to launch extensive cellular protective processes. Thus, biocrust organisms tend to rely heavily on cellular repair during rehydration as their strategy for tolerating desiccation (
20). Although these repair mechanisms are highly efficient (
21) larger biocrust organisms such as mosses lose some cellular contents during the process of membrane repair during rehydration, which in turn may provide a nutritional resource to support a diverse community of heterotrophic microbes, a phenomenon coined the ‘bryotic pulse’ (
22).
Photoautotrophs (cyanobacteria, mosses, and lichens) anchor biocrust communities, both physically (i.e., soil aggregation, hydrological controls) and through primary production. Typically, biocrusts are dominated by one type of photoautotroph, which in turn influences the diversity and abundance of other organisms in the community (
10,
23–25). The identity of the dominant photoautotroph also influences biocrust multifunctionality and community stability in the presence of climate perturbations (
26–29).
The identity of the dominant photoautotroph and taxonomic composition of the rest of the community is at least partially dictated by predictable successional processes (
30,
31). Bare soils are first colonized by filamentous cyanobacteria such as
Microcoleus, which aggregates soil particles with its polysaccharide sheaths and generates organic carbon to support a diverse community of heterotrophic bacteria, including diazotrophs, within the cyanosphere (
32). Later successional stages are characterized by darkly pigmented nitrogen-fixing cyanobacteria like
Scytonema, followed eventually by mosses and/or lichens (
23). While cyanobacteria are typically the major photoautotroph found in hypolithic communities (
5,
33), some hypoliths support mosses (
5,
10,
34–36). Previous taxonomic work indicated some compositional overlap in microbial communities supported by hypoliths and moss-dominated biocrusts (
10), but the extent to which hypolith communities may be functionally distinct from surrounding biocrusts is unknown.
Building on previous work demonstrating that biocrust photoautotrophs affect the taxonomic composition of their associated microbial communities (
10,
23,
25) and biocrust ecophysiology/multifunctionality (
23,
26,
27,
37), we investigated the degree to which niches (biocrust or hypolithic microsites) harbor communities with distinct functional repertoires using a comparative metagenomics approach. We also assessed the effect of dominant photoautotroph (moss, cyanobacteria) on microbial traits to assess the degree to which the photoautotroph anchor might support communities with distinct functional pathways. Specifically, we set out to test the following hypotheses: (1) hypolithic microsites within regions supporting biocrusts should harbor their own distinct microbial communities enriched in pathways reflective of lower levels of heat and desiccation stress relative to surrounding biocrusts, and (2) the presence of moss in biocrusts creates an important nutritional resource due to the ‘leakiness’ of gametophyte tissues, and moss biocrusts should support communities with pathways that reflect the utilization of diverse substrates provided by moss leakage in an oligotrophic environment. We sampled replicate cyanobacteria- and moss-dominated biocrusts and hypoliths from two distinct habitats in the Mojave Desert of California. Metagenomic sequence data generated from these samples were then analyzed to compare functional potential across different biocrust and primary producer types to identify adaptive strategies related to survival in extreme dryland environments.
DISCUSSION
Our investigations revealed that the functional repertoire of surface communities in dryland ecosystems is strongly shaped by ecological niche (biocrust versus hypolith) and, to a lesser degree, dominant primary producer (moss versus cyanobacteria). Relative to niche and primary producer, location and season (Sheep Creek Wash collected in March and Granite Mountains Reserve collected in August) had a substantially reduced effect on functional potential. This observation suggests that the functions identified here likely play conserved roles in the ecology of the different environmental niches distributed in dryland soils. As hypothesized, genes and pathways enriched in biocrusts relative to hypoliths reflect adaptation to heat and desiccation stressors. These communities showed an increased capacity for DNA repair, motility, environmental stimuli sensing and response, and interactions with other community members. On the other hand, hypolithic communities were enriched in antibiotic and secondary metabolite synthesis pathways. Moss-dominated samples showed an increased abundance of genes for the uptake of monosaccharides, amino acids, and osmoprotectants relative to cyanobacteria-dominated samples, which may reflect leakage of these substrates by moss gametophyte tissues (the “bryotic pulse”) (
22).
Given that severe prolonged water deprivation exerts extreme stress on microbial communities, we expected to find differences in samples collected in March, where rainfall had occurred within 2 to 3 days prior to collection, and August, where communities had experienced months of extreme heat without recent precipitation. We also predicted that geographic location might affect community functional potential due to climatic differences between sites. The Wrightwood site experiences cooler annual temperatures and higher precipitation (average high and low annual temperatures, 16.8°C and 1.7°C; average annual precipitation, 49.4 cm) compared to the Granite Mountains site (annual high and low temperatures, 26.5°C and 3.5°C; average annual precipitation, 22 cm) (Wrightwood Weather Station, NOAA National Climatic Data Center; Granites Weather Station, UC Natural Reserve System). Instead, collection month and location had minor effects compared to niche and dominant primary producers. This suggests that biocrust and hypolithic communities are resilient to the stressors imposed by environmental extremes and that these taxa have high degrees of physiological flexibility that enable them to maintain consistent abundances during seasonal fluctuations (
56–58). This is opposed to a model where taxa adapted to specific seasonal environmental conditions (and their genes) change in abundance with yearly cycles. In the future, the resilience of these communities and their physiological plasticity could be further investigated by tracking concurrent changes in taxonomy, function, and functional potential across seasons. Our data also suggest that factors characteristic of the two niches we investigated are more important than geographic distance and broad climatic similarities in determining functional potential. This observation has played out on an even larger scale, where studies have demonstrated that hypolithic communities in cold and hot desert environments share more similarities with each other than with nonhypolithic soils (
8).
DNA repair.
Desiccation and high UV exposure induce multiple types of DNA damage, which is countered by a variety of repair mechanisms. Hypolithic communities colonize the ventral sides of semitranslucent stones, which filter UV radiation and increase moisture availability. Hypolithic communities also experience an attenuation of daily high and low temperature extremes (
12), whereas biocrust communities must persist without this protective buffer. Our data show that biocrust communities have an increased capacity for DNA damage repair, likely to counteract the effects of UV and desiccation. We speculate that enrichment of DNA repair genes may be due to increased copy numbers in biocrust genomes. Previous work from Negev Desert biocrusts demonstrated that taxa highly specialized for the desert crust environment contained multiple copies of double-stranded break repair genes in their genomes (
37), which may enhance expression or produce proteins with alternative activities or specificities. An analogous scenario holds for the genome of the common biocrust moss,
S. caninervis, which contains a highly expanded repertoire of protective early light-induced protein (ELIP) genes (
40), a signature of physiological desiccation and UV tolerance in land plants (
60). The higher abundance of DNA repair genes may also be because biocrust taxa on average possess more repair mechanisms and pathways than hypolith taxa. Previous studies have shown uneven distributions of DNA repair pathways across taxa and suggest the number of repair systems may be related to desiccation and UV tolerance (
61). Future work should enable further investigations into these explanations through genomes assembled from metagenomic sequence data.
Intercellular competition and antibiotic synthesis.
The differential abundance of bacterial secretion systems, quorum sensing genes, biofilm formation genes, and antibiotic synthesis pathways suggests intercellular interactions play an important role in niche specialization in dryland communities. Competition for finite resources through eliminating competitors appears to be crucial in both biocrust and hypolith communities (
62), but the strategies for doing so have diverged. Biocrusts have a greater capacity to use the T6SS as a weapon of interbacterial competition, whereas hypoliths have an increased ability to produce multiple classes of antibiotics. T6SSs deliver toxic effector proteins to the cytoplasm of target cells through a tubular device that extends to puncture the cell envelope (
63). Such interactions require direct cell-to-cell contact, suggesting higher encounter rates between cells. This is consistent with previous observations that T6SS-bearing cells are more abundant in environments with closer cell proximities (
64). Conversely, the production of antibiotics may reflect a more open system and higher moisture content, enabling metabolites to diffuse away from cells. The greater abundance of biofilm-related genes in biocrusts than hypoliths is consistent with increased opportunities for direct cellular interactions. Cells are packed densely in biofilms (
65), which may facilitate the direct contact necessary for the T6SS to deliver toxins to neighboring cells. In hypoliths, increased moisture availability via condensation and slower rates of evaporation (
12) may facilitate the diffusion of compounds between cells, favoring the use of antibiotics. To a lesser degree, we also observed a significant enrichment of antibiotic synthesis pathways in samples containing moss as the primary producer. The association of antibiotic synthesis with the presence of moss might also reflect the increased availability of moisture to enable diffusion of antibiotics, as mosses possess morphological features (e.g., leaf and branch architecture, leaf papillae, and leaf hair points) that are specialized for the sequestration, transport, and retention of external water (
66,
67). We note that the enrichment of antibiotic synthesis pathways in hypoliths compared with biocrusts does not imply biocrusts lack this ability. Indeed, biocrusts and other arid soils may contain more of these genes and pathways that other soil types (
68,
69). Biocrusts were previously found to harbor a diversity of biosynthetic gene clusters, which were crucial for niche differentiation and maintenance (
70).
Canonically, bacterial antibiotic production has been viewed as a weapon in competitive interactions (
71), but alternative hypotheses suggest that subinhibitory concentrations may play a role in intercellular signaling (
72,
73). Recent studies designed to distinguish between the hypotheses strongly indicate that antibiotics act as weapons of interbacterial competition (
74,
75). Regardless, the diversity and abundance of antibiotic synthesis genes suggest this environment is a large potential untapped resource that could aid in addressing the mounting public health crisis of widespread antibiotic resistance in pathogens. Bacterial soil isolates represent a major source of modern antibiotics and other metabolites useful in medicine and biotechnology. However, the use of environmental bacteria for antibiotic discovery has slowed due to high rates of antibiotic rediscovery and because only a small fraction of isolates produce useful metabolites (
76,
77). Here, deep sequencing of the uncultured majority provides a resource that could be used to overcome this hurdle either through targeted cultivation or synthetic biology, potentially revealing novel compounds useful in the clinic and beyond (
76,
78).
Members of the phylum Actinomyceota (formerly Actinobacteria) are known for their ability to produce a diversity of secondary metabolites, including antibiotics. A previous study that surveyed community composition based on 16S rRNA amplicons at one of our sample sites (Sheep Creek Wash) found that four Actinomyceota genera (
Solirubrobacter,
Rubrobacter,
Conexibacter, and
Angustibacter) were significantly more abundant in in hypoliths versus biocrusts (
10). None of these are particularly notable for producing secondary metabolites (
79). The apparent disconnect might be explained by more limited knowledge of these genera compared with known antibiotic producers such as
Streptomyces (
80,
81), differences between the studies (e.g., one study site rather than two and the use of different samples), or a distribution of biosynthetic genes in taxa that extends beyond well-known secondary metabolite producers. Further investigation using assembly and binning methods will likely shed light on this topic revealing the taxa harboring these pathways, their phylogenetic distribution, and potentially other factors such as the relationships between genome size and antibiotic production (taxa with larger genomes tend to have more biosynthetic gene clusters [
82]).
Environmental sensing and response via the two-component regulatory system.
Two component systems are found in nearly all bacterial genomes, but those that inhabit rapidly changing or diverse environments typically encode a large number of two component system genes, suggesting that organisms expand their repertoire to adapt to environmental challenges (
51,
83). The high prevalence in biocrust communities, which are exposed to more extremes than hypolith communities, is consistent with this observation suggesting that sensing and responding to local conditions and stressors play a key role in adaptation to the biocrust ecological niche. Since the input signal and cellular response of a given system can often be predicted based on DNA sequence, the specific two-component system genes that are enriched in a particular environment can be used to infer which environmental stimuli microbial communities are attuned to and the subsequent downstream response (
49). In the case of biocrust communities, these adaptations include motility and chemotaxis, surface adhesion and biofilm formation, redox conditions, nutrient limitation, and environmental stressors, all of which may be particularly important for organisms inhabiting oligotrophic environments with transient pulses of nutrient availability and metabolic activity (
58,
84,
85). Biocrusts are poikiloydric communities with the ability to desiccate completely during extended dry periods and quickly resume metabolic activity when moisture becomes available (
1). Heterotrophic organisms in these communities must thus maintain the capacity to respond quickly to changes in their environment and exploit resources generated by primary producers during brief pulses of hydration and metabolic activity. Moss-dominated biocrusts produce vertical strata, with gradients of light, moisture, UV exposure, and nutrients lost from moss leaf cells during rehydration (
22). It is likely critical for microorganisms to optimize and maintain their position relative to the spatial distribution of these variables within the moss biocrust. Similarly, the microbial community associated with the common early-successional cyanobacterium
Microcoleus (the “cyanosphere,” [
32]) is also likely to harbor adaptations to chemotaxis and motility, as
Microcoleus moves vertically within the soil surface in response to moisture availability (
86).
Methylotrophy, photosynthesis, and CO2 fixation in hypolithic communities.
The enrichment in genes related to methylotrophy in hypoliths compared to biocrust communities could reflect a higher abundance of moss-associated Methylobacteria in hypoliths. Methylobacteria are known to live as epiphytes on plants, including mosses, where they utilize methanol emitted as a by-product of pectin degradation in cell walls during cell division and growth (
87). We note that if the moss-associated methylotrophs were the sole explanation for the high abundance of methylotrophy genes in hypoliths, we should also have observed a corresponding enrichment of the methane metabolism pathway in moss- versus cyanobacteria-dominated samples, which was not apparent at the conservative threshold we used to identify differentially abundant genes (
P < 0.01). However, when a threshold of
P < 0.05 was applied, most of the methotrophy genes that were enriched in hypoliths versus biocrusts were also enriched when comparing moss versus cyanobacterial samples. This suggests the enrichment observed in hypoliths may be due to higher abundances of moss-associated methylotrophs in hypolithic microsites plus other yet unknown metabolic processes resulting the availability of C1 substrates to methylotrophic communities. In hypolithic communities, we also found that photosynthesis and CO
2 fixation pathways were more abundant than in biocrusts, which might be explained by a higher abundance of cyanobacteria relative to other microbial taxa in hypoliths (
10).
Other factors controlling functional potential in hypolith and biocrust communities.
Previous studies have observed that differences in carbon and nitrogen content affect microbial communities in biocrusts at varying successional states, with pH playing a weaker role (
23,
59). At our sites, pH values were weakly acidic to neutral (6.15 to 7.52) and differed between primary producer (cyanobacteria, 7.08; moss, 6.37), niche (hypolith, 6.48; biocrust, 6.96), and sample site (Wrightwood, 6.49; Granite Mountains, 6.96) (
Table S2). Though pH levels differed between niches, the magnitude of difference between primary producers and sample sites was roughly the same. Because sample site and primary producer had a reduced effect on functional potential compared to niche, it is likely that pH gradients may influence functional potential but play a secondary role compared to other environmental factors. Further investigations including other physicochemical analyses and additional pH measurements should shed light on this issue.
We note that we did not observe substantial differences between samples and bare soil controls (
Fig. 1), which is contrary to our expectations and prior observations (
59). We posit three explanations. First, we sampled the surface (1–2 cm) in areas directly adjacent to biocrusts (within 1 m). It is likely that we captured transient communities seeded by the surrounding biocrusts/hypoliths. Second, it plausible that bare soil contains a subset of biocrust/hypolith communities given the shared stressors and environmental conditions. Finally, we collected many fewer bare soil controls compared with samples. Additional controls may have improved our ability to detect differences.
Approaches to analyzing metagenomes.
This study represents an overview of the functional potential of biocrust and hypolith microbial communities. By using reads rather than assembled data, we were able to capture sequences from low abundance taxa and avoid biases introduced by assembling and binning metagenomic sequence data (
88), thus providing quantitative information on gene relative abundances. We mitigated issues associated with read-based analyses such as mis-annotations due to short read length by using an extremely conservative approach, setting a low
P-value threshold (
P > 0.01) and requiring that many genes within a pathway reach significance.
Unlike 16S rRNA amplicon data or assembly approaches, our analysis strategy yields limited information regarding taxonomy or the phylogenetic distribution of traits. The exception is instances where taxonomy and functional genes are tightly linked, as shown by the enrichment of cyanobacteria metabolic pathways in samples where cyanobacteria are the dominant primary producer. Ongoing work that includes assembly and constructing metagenome assembled genomes will provide complementary information, revealing how functions are partitioned among community members and enabling us to address questions generated by read-based analyses such as whether enhanced DNA repair abilities in biocrust communities are due to copy number variation, more DNA repair pathways, or a combination of the two.
Concluding remarks.
In dryland regions where hypoliths and biocrusts are intermingled on the soil surface, both niches share superficial similarities, such as primary producers, which motivated us to question how the microbial communities associated with these proximate yet distinct microsites might differ. We found that niche (biocrust or hypolith) had a stronger influence on functional pathway differences than primary producers (moss or cyanobacteria). The importance of environmental stressors such as desiccation, extreme daily temperature fluctuations, and UV exposure was reflected in the significant enrichment of pathways associated with responding to and mitigating these stresses in biocrusts as opposed to hypoliths. Contrasting strategies for competition that we observed in our comparisons may reflect different conditions promoted by niche and primary producers and highlight hypoliths and moss biocrusts as potential sources of novel antibiotics. Notably, the functional signal generated by niche and primary producers greatly overshadowed the influence of spatial (collection site) or temporal (season) variations, highlighting the deterministic nature of these communities. The consistency of functional pathways across divergent environmental conditions suggests that the communities we sampled may be relatively stable, relying on physiological plasticity and/or intermittent quiescence (dormancy) for survival as opposed to compositional turnover.
ACKNOWLEDGMENTS
Metagenomic sequencing for this project was provided by the Department of Energy Joint Genome Institute (JGI) through a Community Science Program award (CSP 504034) to K. Fisher and P. Vaishampayan, and R. Mackelprang. This work was also supported by a National Science Foundation Dimensions of Biodiversity Program award to K. Fisher (DOB 1638996).
The work (proposal: 10.46936/10.25585/60000926 conducted by the U.S. Department of Energy Joint Genome Institute (
https://ror.org/04xm1d337), a DOE Office of Science User Facility, is supported by the Office of Science of the U.S. Department of Energy operated under Contract No. DE-AC02-05CH11231.
Jameka Jefferson and Amy Vasquez assisted with the preparation and DNA extraction of samples used for this study. Thank you to Jameka Jefferson and Jenna Ekwealor for field assistance, and the Sweeny Granite Mountains Desert Research Center for facilitating sample collection.
We declare no competing interests.