Diversity and Ecology of Viruses in Hyperarid Desert Soils

Abstract

In recent years, remarkable progress has been made in the field of virus environmental ecology. In marine ecosystems, for example, viruses are now thought to play pivotal roles in the biogeochemical cycling of nutrients and to be mediators of microbial evolution through horizontal gene transfer. The diversity and ecology of viruses in soils are poorly understood, but evidence supports the view that the diversity and ecology of viruses in soils differ substantially from those in aquatic systems. Desert biomes cover ∼33% of global land masses, and yet the diversity and roles of viruses in these dominant ecosystems remain poorly understood. There is evidence that hot hyperarid desert soils are characterized by high levels of bacterial lysogens and low extracellular virus counts. In contrast, cold desert soils contain high extracellular virus titers. We suggest that the prevalence of microbial biofilms in hyperarid soils, combined with extreme thermal regimens, exerts strong selection pressures on both temperate and virulent viruses. Many desert soil virus sequences show low values of identity to virus genomes in public databases, suggesting the existence of distinct and as-yet-uncharacterized soil phylogenetic lineages (e.g., cyanophages). We strongly advocate for amplification-free metavirome analyses while encouraging the classical isolation of phages from dominant and culturable microbial isolates in order to populate sequence databases. This review provides an overview of recent advances in the study of viruses in hyperarid soils and of the factors that contribute to viral abundance and diversity in hot and cold deserts and offers technical recommendations for future studies.

INTRODUCTION

Over recent decades, the critical roles that viruses play in the environment have become increasingly recognized by the research community (1). It has been estimated by direct counts of extracellular (“free-floating”) virus-like particles (VLPs) that the global “virosphere” may contain up to 10³¹ viral particles (2), suggesting that viruses may be the most abundant biological entities on the planet and, potentially, the greatest reservoir of genetic diversity (3–5). Appreciation of the ecological importance of viruses on a global scale has predominantly emerged from studies of marine and freshwater microbial communities (6–12), where viruses have been linked to core processes such as biogeochemical nutrient cycling (6, 7, 10), microbial population control through viral lysis (7, 8), and microbial evolution via horizontal gene transfer (11). Research on the virus ecology of soil environments has progressed more slowly and has received proportionally less attention (12–14). However, enumeration of virus particles by electron microscopy (EM) performed on several soil types (15–17) has shown high viral abundance values ranging from 1.5 ×10⁸ to 6.4 ×10⁸ per gram (dry weight) of soils. Soil ecosystems are subject to unique abiotic ecological pressures, in part due to their wide compositional spectrum and spatial heterogeneity in terms of physicochemical properties (18, 19). Environmental stresses are even greater in extremely arid soil systems, where soil organisms and communities may be simultaneously exposed to pulsed-water events and to the effects of desiccation-, solute-, and UV-B radiation-induced oxidative stresses (20, 21). Deserts represent the single largest terrestrial ecosystem type on Earth, covering ∼33.6% of the global land mass, excluding Antarctica (22), and are classified in terms of their aridity index, a ratio between precipitation (P) and potential of evapotranspiration (PET) (23). This results in four desert categories: dry-semiarid (0.5 < P/PET < 0.65), semiarid (0.2 < P/PET < 0.5), arid (0.05 < P/PET < 0.2), and hyperarid (P/PET < 0.05). Hyperarid deserts generally receive annual precipitation of ≤70 mm and are often associated with intrinsic characteristics such as high (∼7 to 9) pH, high salinity levels, high surface radiation fluxes, long periods of desiccation, and low water activity (24). Desert soil microbial ecology research has primarily focused on bacterial communities, which have been shown to be largely responsible for primary production and provision of key ecosystem services (25–28). Soil virus populations and functions are seldom taken into consideration, thereby omitting a crucial variable within ecological models designed to predict microbial population dynamics. As a result, the ecological roles, survival mechanisms (against biotic and abiotic factors), and spatial and temporal changes in viral community structures (virus biogeography) and viral phylogenetic diversity are still poorly understood in desert soils. Within the field of soil virus ecology (13), several desert soil ecosystems have been recently investigated (Table 1). With the advances in next-generation-sequencing (NGS) technologies, culture-independent methods have become the standard for determinations of viral diversity (41). However, the rapidly growing volume of viral environmental sequence data has revealed that most sequences (∼70%) have no homologs in public databases, and such sequences are typically labeled “viral dark matter” (42, 43). Here, we discuss the current understanding of hot and cold desert soil virus diversity and function, propose alternative technical approaches to virus concentration methods, and identify key areas of future research.

TABLE 1.

Medium-to-high-throughput soil-based studies pertaining to viral ecology since 2005

Authors, yr of publication (reference no.)	Soil type(s)	Location(s) of sample collection	Method(s) useda
Prigent et al., 2005 (29)	Hot desert surface sand	Sahara Desert in Morocco and Tunisia	EM, PFGE, lytic induction
Williamson et al., 2005 (16)	Agricultural, forest	Delaware, USA	Epifluorescence microscopy, EM
Fierer et al., 2007 (30)	Hot arid desert, tallgrass prairie, tropical rainforest	USA, Peru	Sanger sequencing of random viral metagenomics clones
Williamson et al., 2007 (17)	Loamy and sandy soils, agricultural, forested wetlands	Antarctica (Tom and Obelisk Pond); USA (Delaware)	Induction assays, epifluorescence counting
Prestel et al., 2008 (31)	Surface sand	Namib Desert	EM, PFGE, Sanger sequencing of cloned DNA fragments (LASL)
Swanson et al., 2009 (32)	Dystric-fluvic Cambisol soil	Dundee, Scotland	EM, epifluorescence counting
Meiring et al., 2012 (33)	Soil under hypoliths	Miers Valley, Antarctica	Lytic induction, EM, phage isolation from culture
Pearce et al., 2012 (34)	Surface soil	Alexander Island, Antarctica	Shotgun metagenome pyrosequencing
Swanson et al., 2012 (35)	Surface soil (Antarctica)	Antarctica	EM, lytic induction, phage isolation
Prestel et al., 2013 (36)	Dune surface sand	Mojave Desert, USA	EM, random amplification for viral DNA (Sanger)
Srinivasiah et al., 2013 (37)	Surface soil (Antarctica); silt loamy soil (USA)	Antarctica (Tom and Obelisk pond); Delaware, USA	RAPD viral community fingerprinting
Adriaenssens et al., 2015 (38)	Soil-associated rocks (hypoliths)	Namib Desert	Shotgun viral metagenome sequencing (Illumina)
Zablocki et al., 2014 (39)	Antarctic surface soil and hypoliths	Miers Valley, Antarctica	Shotgun viral metagenome sequencing (Illumina)
Srinivasiah et al., 2015 (40)	Silt loamy soil	Delaware, USA	Microcosms, RAPD viral community fingerprinting, epifluorescence counting

^aEM, electron microscopy; PFGE, pulse field gel electrophoresis; LASL, linker amplified shotgun library; RAPD, random amplified polymorphic DNA.

DIVERSITY AND ABUNDANCE OF VIRUSES IN DESERT SOILS

Hot deserts.

Viral community analyses have been conducted on surface soil samples from three hot hyperarid deserts: the Sahara (29), Namib (31, 38), and Mojave (30, 36). In each of those studies, difficulties in detecting extracellular VLPs by electron microscopy (EM) or pulse field gel electrophoresis (PFGE) profiling were reported, suggesting a very low viral abundance within these soils. However, the inclusion of a lytic induction step (prophage excision stimulated by the addition of mitomycin C [44]) in the soil extraction protocol substantially increased the recovery of virus particles (29, 31). For Sahara Desert surface sand samples, induced phage genomes were estimated to range in size from 45 to 270 kb. EM of the induced phage fraction showed a majority of tailed virus morphotypes belonging to the Myoviridae family, some of which showed peculiar ribbon-like structures located at the tail tip of the virions (45). In the Namib Desert soil samples, 20 distinct morphotypes were identified, all members of the Myoviridae and Siphoviridae families with no apparent Podoviridae-like virions (31). PFGE profiles from Namib soils indicated an average genome size of 55 to 65 kb, with several genomes of up to 350 kb in size (31). EM visualization of Mojave Desert sand samples showed 11 distinct tailed morphotypes, belonging to all three families of the Caudovirales (36). Sanger sequencing of randomly selected cloned phage fragments from the Mojave Desert soil virus communities showed that 36% of the sequenced clones had no homologs in public sequence databases (36). Within the identified virus sequences, the majority were homologous to bacteriophages infecting common soil bacteria such as members of the Proteobacteria, including Bacillus and Rhizobium. From the same samples, 38 bacterial isolates were grown in pure culture and 84% were shown to harbor at least one SOS-inducible phage. A similar study on loamy sand from a different area of the Mojave Desert showed that a large majority of randomly selected metaviral clone sequences had no database homologs (30). Of those clones with significant sequence identity (tBLASTx search using an E-value cutoff value of 10⁻³), phages associated with Actinoplanes, Mycobacterium, Myxococcus, and Streptomyces were the most common. Other virus signals detected included archaeal (Haloarcula phage) and herpes-like viruses. Using a similar methodology, 50% of the viral sequences from three Namib Desert surface sand samples had no homologs in public sequence databases, with most positive hits showing homology to Siphoviridae phages linked to Gram-positive bacteria (31). Most recently, a shotgun NGS approach was used to investigate the metavirome of Namib Desert hypoliths (38), cyanobacterium-dominated microbial niche communities on the ventral surfaces of translucent rocks (46). The most abundant sequences belonged to Geobacillus- and Bacillus-infecting phages, while cyanophage markers were unexpectedly found only in low numbers. The distinct phylogenetic clustering of assembled phoH genes (a cyanophage marker [47]) suggested that desert soil cyanophages were related only distantly to their well-studied marine counterparts (38) and that the dominance of marine cyanophage sequences in sequence databases might account for the low cyanophage hit rate of homologous sequences in the Namib Desert hypolithon metavirome. This observation has wider implications for studies of soil metaviromics, where an underestimation of cyanophage abundance and diversity may skew estimates of the functional importance (and population dynamics) of soil cyanobacteria, arguably the most important taxonomic group in desert soil microbial communities (27, 28).

Cold deserts.

Studies of viral communities in cold hyperarid desert soils have been almost exclusively conducted in the major ice-free regions of Antarctica (e.g., the East Antarctic McMurdo Dry Valleys). Direct viral counts by epifluorescence microscopy (17) showed high VLP densities, in the range of 2.3 × 10⁸ to 6.4 × 10⁸ extracellular VLPs per gram of dry soil. The prevalence of bacterial lysogens within these soils was between 4.6% and 21.1%, a much lower occurrence level than has been estimated for bacteria in hot desert soils (84% [30]). Using epifluorescence direct counts of extractable bacteria and extracellular virus particles, virus-to-bacteria ratios (VBR) ranging from 170 to 8,200 were calculated, the highest recorded for any soil ecosystem (17). Antarctic soil bacterial isolates have yielded several unique virus genomic structures. The distinct temperate siphoviruses (SpaA1 and BceA1) isolated from Staphylococcus pasteuri and Bacillus cereus both contained almost-complete additional phage genomes (MZTP02) (35). This “Russian doll” gene arrangement had not been previously described for soil bacteriophages and has led to speculation that it may represent a “fast-track” route for virus evolution and horizontal gene transfer, with a possible role in host range expansion. Pyrosequencing of Antarctic soil metagenomic DNA has identified a wide diversity of bacteria, archaea, microeukaryotes, and viruses (34). From the total sequence data set, 494 phage-related hits (0.18% of the total number of sequences) were identified. Top BLAST hits in public databases were related to phages known to infect Mycobacteria, Burkholderia, Bordetella, Pseudomonas, Enterobacteria, Flavobacterium, Myxococcus, Synechococcus, Prochlorococcus, and Sinorhizobium. However, viral DNA was not specifically enriched in this study, and this may have resulted in an underestimation of viral diversity. The spatial composition and dynamics of viral communities along an Antarctic soil transect have been recently reported (37). Using random amplified polymorphic DNA (RAPD-PCR) assays, viral community fingerprints were used to assess short-term changes in the composition of viral communities. To maximize the number of virus sequences amplified, RAPD-PCR primer design was based on the identification of recurring dodecamer sequences (G+C content, ≥70%) within 22 selected viral metagenomes. Qualitative comparisons of the Antarctic fingerprint patterns demonstrated that heterogeneous soil conditions and associated environmental factors (e.g., carbon levels, moisture content, pH, and light exposure frequency) impacted the composition of viral assemblages across geographic distances of as little as 20 meters. The RAPD-PCR fingerprint data also suggested that virus assemblages were not present as inactive, inert particles but were dynamically involved in infection of coexisting microbial hosts. Furthermore, the authors suggested that environmental pressures (e.g., low moisture levels) known to influence bacterial community structures in the Antarctic desert (17) were shown to have a similarly influential role in virus community dynamics. Abundance estimates (17) suggest that Antarctic desert soils contain a substantially higher proportion of free extracellular VLPs than hot hyperarid desert soils, where a lysogenic lifestyle appears to be prevalent (29, 31, 36). A sequence-based metagenomic comparison of viral assemblages (single- and double-stranded DNA viruses only) in surface soils and hypolithic communities in the Antarctic McMurdo Dry Valleys (39) demonstrated that bacteriophages constituted the majority of the identified viruses, representing all Caudovirales families. Mycobacterium phage sequences were the most highly represented in the viral fraction (34). No archaeal virus sequences were recorded, in line with previous observations that archaea are either absent in this environment or present in very low numbers (26, 48). Within the hypolith metavirome data set, the fraction of cyanophage sequences was underrepresented, with low sequence similarities to known cyanophages. Dry Valley surface soils also contained a number of other virus signatures, including phycodnaviruses, mimiviruses, and virophage capsid protein genes (39), many of which are most commonly identified in aquatic systems.

FACTORS SHAPING VIRAL COMMUNITY STRUCTURES IN DESERT SOILS

Soil virus populations display dynamics different from those in marine and freshwater systems (37) (Fig. 1). In marine systems, two major factors influence viral abundance: the biological productivity of the system and microbial diversity and abundance (3, 5). Viral abundance has been shown to increase as bacterial productivity in a system increases (49). Cooccurring virus host communities also influence viral abundance, as in the case of microbial bloom events, which increase the number of lytic infections, thereby releasing additional phage particles (50, 51). Marine environment-associated abiotic parameters such as temperature, salinity, and pH are stable on relatively large spatial scales (13) and do not appear to significantly affect viral abundance. Soil systems, particularly desert soils, are inhomogeneous, in that soil particles are semidiscrete. Extended periods of desiccation and oligotrophy are typical characteristics of hyperarid desert soils. Under these environmental constraints, microbial populations often form discrete biofilms, where cells embedded in extracellular polymeric substance (EPS) matrices are adsorbed to particle surfaces (52–54). The EPS matrix serves a protective role, sequesters nutrients, and provides a defense barrier against virulent phages (55). Temperate phages in their prophage state have been shown to stabilize biofilms, whereas a switch to the lytic cycle aids in biofilm dispersal. Within the biofilm, temperate phages have been shown to contribute to the life cycle of biofilm by aiding in biofilm dispersal (55). While this has been described for Pseudomonas aeruginosa biofilms (56), we argue that a similar mechanism may be present in hyperarid desert soil biofilms. Such a mechanism would drive the positive selection of temperate phages in this ecosystem and negatively influence the presence of virulent phages. This suggestion is consistent with the observation that extracellular VLPs are readily extracted only after induction of prophages in hot desert soils (29, 31, 36, 57). In Antarctic hyperarid desert soils, where biofilm communities also frequently occur (58), high VLP counts have been recorded. We suggest that the effects of temperature may explain the apparent differences between hot and cold desert soil systems. Temperature has been shown to be one of the major factors controlling viral survival rates in soils (59, 60), with lower temperatures enabling higher survival rates, extended latent periods, and reduced burst sizes. Warmer temperatures have been associated with reduced virus proliferation and higher inactivation rates (61, 62). Thus, in Antarctic soils, colder temperatures may allow the preservation of extracellular VLPs, making them more abundant and detectable (63). In contrast, the high-temperature settings (e.g., maxima of ≥50°C in the Namib Desert [48]) of hot deserts may increase the rate of degradation of extracellular virus particles.

FIG 1.

Virus community dynamics in aquatic (A) and soil (B) ecosystems. Marine and freshwater systems can be regarded as homogenous systems, where the distributions of virus particles (e.g., phages) and host organisms (e.g., bacterioplankton) are relatively even. (A) Such a continuous medium allows rapid phage-host dispersion and increases the rates of phage-host collisions, leading to high infection rates. (B) In contrast, hyperarid soil microbial communities exist as discrete systems, embedded in protective biofilms. The level of virus-host interactions (VHI) within and between individual biofilm communities remains an open issue, but diffusion rates are expected to be low on both small and large spatial scales.

Viral operational taxonomic unit (OTU) abundance estimates from low-throughput Sanger sequencing of metaviromes have provided some insights into the factors that shape the diversity of viral communities in desert soils (30). Comparisons of viral community compositions across three contrasting soil ecosystems (prairie, desert, and rainforest) have demonstrated that microbial communities were both locally and globally diverse. Comparative phylogenetic analyses showed little taxonomic overlap of soils sampled from the three different habitats, as well as low values of identity to annotated sequences in public databases. However, the factors that may be responsible for the observed niche specialization are as yet unknown.

Similar habitat-specific viral community compositions that have been determined through the use of hierarchical clustering of metaviromes, based on dinucleotide frequencies, have been reported (64). This method is especially useful for gaining ecological insights from metagenomic data sets containing a majority of unaffiliated reads that have been provided to public databases. Dinucleotide frequencies within metaviromes have shown distinct virus community clustering within single habitat types such as desert soils. Although reported from a single study which analyzed two sets of pooled samples, cluster analysis of hypolith and open-soil metaviromes from Antarctic and Namib Desert soil samples has shown that the two hypolith metaviromes clustered at a single node whereas, in contrast, the two open-soil metaviromes displayed identical patterns (65). Despite great geographic distances or differing environmental conditions, similar habitat types harbored more closely related viral communities. The most obvious common factor in the two contrasting deserts is very limited water availability, which may be a key driver of community speciation and recruitment in these soils.

TECHNICAL RECOMMENDATIONS AND FUTURE RESEARCH

Research on desert soil viruses is technically challenging, partly due to the physical properties of soil. Desert soils frequently produce suboptimal (≤10 ng/μl) viral DNA yields (66), forcing the inclusion of a random PCR amplification step for NGS library construction. The use of whole-genome amplification (WGA) by multiple-displacement amplification (MDA) or random-priming, sequence-independent, single-primer amplification (RP-SISPA) (67) almost certainly results in biased amplification of certain virus groups (68–71) and prevents the accurate determination of viral abundances and diversity. While viral amplification is widely accepted as a necessity in metaviromic studies (72), we argue that amplification of virus metagenomic DNA should be avoided where possible. It would be preferable to focus efforts on improving virus concentration methods in order to reach the minimum concentrations required for sequencing. Sequential rounds of centrifugation and the pooling of samples should increase the number of viruses recovered. Methodological improvements in virus concentration would also allow more-precise virus counts using microscopy (73). Thus, the development of more-efficient and -effective metaviromic DNA extraction technologies, so as to obviate the use of WGA, would represent a substantial advance in the field. Efforts to achieve this goal are further facilitated by recent technical improvements in sequencing chemistries where, for example, Illumina paired-end sequencing library construction kits (ThruPLEX; Rubicon Genomics) have reduced the minimum genomic DNA requirement to around 50 pg.

Sequence-based identification of viral communities, using either multiple gene markers (74) or full-virome sequencing (75), is becoming more routine. In marine virus ecology, the use of conserved viral marker genes such as DNA polymerases (76), ribonucleotide reductases (77), and T4-related structural proteins (78, 79) has provided detailed data on viral biodiversity, on intra- and interviral evolutionary relationships, and on oceanic viral turnover rates. The use of these methods to study virus diversity and biogeography in desert soils is relatively new and most commonly involves the sequencing of whole metaviromes (37–39). However, metaviromic approaches generally result in a large number of unknown sequences (43). In addition, we warn that the taxonomic affiliation of single genes and/or virus genome fragments (using BLAST analysis of public databases) in metavirome data sets may not be evidence for the presence of these viruses in the sample (80), and results should be carefully inspected by additional read mapping to a reference genome. While a metaviromics approach provides the opportunity for virus discovery, it may also be valuable to use, in parallel, a high-throughput sequencing approach focusing on conserved signature genes. Such a combinatorial strategy could provide both informative data on viral richness and insights into the functional roles of viruses in soil ecosystems.

A common feature of many desert ecosystems is the occurrence of hypolithic niches (81). These rock-associated cryptic microbial communities are usually dominated by photosynthetic cyanobacteria but contain a wide diversity of members of the phyla Actinobacteria, Acidobacteria, Bacteroidetes, and Proteobacteria (26, 82–84). Cyanobacteria are of particular importance, due to their key roles in primary productivity and nitrogen input in depauperate ecosystems (27, 28). To date, no fully characterized desert soil-associated cyanophage isolates have been reported. Preliminary metagenomic data on Antarctic and Namib hypoliths (38, 39) have shown evidence of novel soil cyanophage lineages, the sequences of which have low identities to characterized marine cyanophage genomes. As cyanobacteria are readily amenable to culturing (85), this provides opportunities for the isolation of their phages and access to full-length soil cyanophage genomes. Such data would support downstream applications such as primer design for targeted amplification of related taxa and monitoring of these assemblages within desert soil ecosystems.

CONCLUSION

Research on phage ecosystem ecology in hyperarid desert soils has demonstrated that desert soil viruses are numerous and diverse and carry novel genes whose functions are yet to be determined. In order to understand how viruses contribute to desert soil ecosystem functioning, critical research questions, addressing both micro- and macroscale issues, must be addressed. The microscale complexity of the soil matrix drives the distribution, maintenance, metabolic state, and biodiversity of microbial communities (86). Consequently, investigating the dynamics of virus-host interactions at the microscale level will contribute significantly to our understanding of the factors which determine virus distribution and diversity of bacteriophages in soil systems. In addition, the effects of extreme physicochemical conditions (e.g., intense UV radiation and temperature) on the preservation of virion particles and the kinetics of virus decay in hyperarid deserts remain unexplored. At the macroscale level (i.e., the ecosystem level), the contributions of viruses to ecosystem services such as nutrient cycling and energy flow and the sequestration of nutrients remain open issues. It would also be highly informative to understand the kinetics and scales of virus transport processes within and between hyperarid ecosystems, potentially important factors in understanding phage phylogeography.

Biographies

Olivier Zablocki received his B.Sc. and M.Sc. in Microbiology from the University of Pretoria, South Africa. While pursuing his M.Sc., he trained as a plant virologist and participated in the implementation of NGS for plant disease diagnosis for the South African citrus industry, with a focus on Citrus tristeza virus. In 2013, he started a Ph.D. program under the supervision of Don Cowan, the current director of the Centre for Microbial Ecology and Genomics at the University of Pretoria. For his thesis, he used both metaviromics and soil physicochemical analyses to assess virus community structure and dynamics in hyperarid desert soil ecosystems. He has recently joined the Institute for Microbial Biotechnology and Metagenomics at the University of the Western Cape (South Africa) to continue research in virus ecology using “omics” strategies.

Evelien Adriaenssens received her B.Sc. and M.Sc. from the faculty of Bioscience Engineering of the University of Leuven (KU Leuven), Belgium. During her Ph.D. research, she worked on the isolation and characterization of bacteriophages of the potato pathogen Dickeya solani for applications in plant protection. This was a collaborative effort between the Laboratory of Gene Technology of Rob Lavigne (KU Leuven), the Plant Production Laboratory of Maurice De Proft (KU Leuven), and the Unit Plant-Crop Protection of the Institute for Agricultural and Fisheries Research with Martine Maes. After obtaining the degree in 2012, she joined the Centre of Microbial Ecology and Genomics at the University of Pretoria, South Africa, in 2013 to work as a Postdoctoral Fellow heading the viral metagenomics project, investigating viral communities in hot and cold deserts.

Don Cowan was educated in New Zealand at the University of Waikato and completed a period of postdoctoral study there before moving to University College London (United Kingdom) as a Lecturer in 1985. After 16 years in London, he accepted the position as Professor of Microbiology in the Department of Biotechnology at the University of the Western Cape, Cape Town, where he was a Senior Professor and Director of the Institute for Microbial Biotechnology and Metagenomics. In May 2012 he moved to the University of Pretoria as director of both the Genomics Research Institute and the Centre for Microbial Ecology and Genomics. His research activities encompass a wide range of projects in the field of ecogenomics: the use of genomic and metagenomic methods to understand the diversity and function of microorganisms in different environments.