A global atlas of soil viruses reveals unexplored biodiversity and potential biogeochemical impacts

Historically neglected by microbial ecologists, soil viruses are now thought to be critical to global biogeochemical cycles. However, our understanding of their global distribution, activities and interactions with the soil microbiome remains limited. Here we present the Global Soil Virus Atlas, a comprehensive dataset compiled from 2,953 previously sequenced soil metagenomes and composed of 616,935 uncultivated viral genomes and 38,508 unique viral operational taxonomic units. Rarefaction curves from the Global Soil Virus Atlas indicate that most soil viral diversity remains unexplored, further underscored by high spatial turnover and low rates of shared viral operational taxonomic units across samples. By examining genes associated with biogeochemical functions, we also demonstrate the viral potential to impact soil carbon and nutrient cycling. This study represents an extensive characterization of soil viral diversity and provides a foundation for developing testable hypotheses regarding the role of the virosphere in the soil microbiome and global biogeochemistry.

Viral contributions to soil ecology are largely unknown due to the extreme diversity of the soil virosphere.Despite variation in estimates of soil viral abundances (10 7 to 10 10 viruses per gram of soil), it is clear that soils are among the largest viral reservoirs on Earth [1][2][3] .Early metagenomics investigations have revealed high genetic diversity in soil viruses, with putative impacts on global biogeochemistry 1,2,4,5 .Still, less than 1% of publicly available viral metagenomic sequences are from soil 6 , reflecting the lack of knowledge about soil viruses and their ecological roles 4,7 .
High soil viral diversity may be due to the structural and/or physico-chemical heterogeneity of soil compared to other ecosystems 1,[8][9][10] , as well as the high diversity of their microbial hosts.Indeed, soil viral abundance and composition vary with factors such as pH, temperature, moisture, chemistry, and habitat [10][11][12] .Much of this viral diversity is contained within DNA viruses, though RNA viruses also have the potential to influence soil processes 13,14 .
While less is known about soil viral activity, a recent study of peatlands reported that close to 60% of soil viral genomes may be involved in active infections 15 , consistent with high activity observed in marine and other systems 4,[16][17][18] .
Whether common macroecological patterns apply to the soil virosphere remains an open question; initial studies of the soil virosphere indicate that the ecology of viruses is at least partially decoupled from other microorganisms 8,10,19 .A major finding is that soil viral community turnover may occur over shorter spatial and temporal scales than microbial communities 8,10,19 .For instance, spatial viral turnover has been shown to be over 5 times higher than microbial community turnover across an 18m soil transect 8 , and only 4% of peatland 'viral operational taxonomic units' (vOTUs) are shared across continents 20 .Other studies note the possibility for long-distance soil viral dispersal through atmospheric 21 or aquatic transport 22 consistent with low turnover.These contrasting results indicate a lack of consensus surrounding the spatial and temporal patterns of soil viruses and the need for large-scale surveys of the soil virosphere.
Importantly, soil viruses can influence biogeochemical cycling, antibiotic resistance, and other critical soil functions by releasing carbon and nutrients during host infection and/or by altering host metabolism via auxiliary metabolic genes (AMGs) 9,15,18,[23][24][25][26][27][28] .While soil AMG characterization is nascent 14 , marine systems demonstrate the breadth of functions ripe for discovery in soil 24 .More than 200 viral AMGs encoding functions related to carbon and nutrient cycling; stress tolerance; toxin resistance; and other processes have been detected in marine systems 24 .In contrast, only a handful of these functions have been identified as soil viral AMGs 12,14,15,22,29,30 .AMGs encoding carbohydrate metabolism in particular may be present in soils, including a few that have been experimentally validated 9,10,15,[29][30][31] .
Accordingly, understanding the role of viruses in soil ecosystems is one of the most pressing current challenges in microbial ecology 32 .Despite the expansion of studies characterizing soil viruses 4,12,29,30 , a comprehensive description of the global soil virosphere has yet to be performed.Such a description is necessary to begin to address questions regarding the spatiotemporal dynamics, physicochemical interactions, host organisms, and food web implications of the soil virosphere.We present a meticulous compilation of the Global Soil Virus (GSV) Atlas based on previous metagenomic investigations of worldwide soils.This atlas represents the most extensive collection of soil metagenomes to date, encompassing contributions from prominent repositories, ecological networks, and individual collaborators.

Global Soil Virus (GSV) Atlas.
For a description of the files contained by the GSV Atlas, please see 'Data Availability'.
We amassed 1.25 × 10 12 of assembled base pairs (bp) across 2,953 soil samples, including 1,552 samples that were not previously available in the United States Department of Energy's (DOE), Joint Genome Institute's (JGI) IMG/M database (Fig. 1 and 2).These samples were screened for viruses, yielding 616,935 uncultivated virus genomes (UViGs) of which 49,649 were of sufficiently high quality for further investigation (see methods).To quantify the extent of new viral diversity encompassed by the GSV Atlas, we compared sequences from samples not already in IMG/VR to those that were previously deposited.Newly contributed sequences clustered into 3,613 vOTUs of which only 317 clustered with existing viral sequences in IMG/VR.The vast majority associations with IMG/VR were with sequences previously uncovered from soil habitats (Fig. 2b).
The 49,649 UViGs of sufficient quality for downstream analysis clustered into 38,508 vOTUs at the species-like level 34 , of which 3,296 were previously unrepresented in IMG/VR (Fig. 2a-b).Only 13.9% of the GSV Atlas' vOTUs appeared in more than one sample, and less than 1% were present in more than 5 samples.At higher taxonomic levels, we found 21,160 and 7,598 clusters at the genus and family level respectively 35 .This equates to an average of 40.01 (range: 1 -2,124), 35.48 (range: 1 -1,651), and 24.91 (range: 1 -896) unique viral clusters per sample at the species, genus, and family levels.38,278 out of 38,508 vOTUs (99.4%) had at least one member assigned to a taxon by geNomad.
In contrast to the saturation observed in rarefaction curves for microbial taxonomy and genes annotated by Pfam, rarefaction curves of soil viruses for individual samples (vOTUs and and viral clusters) and their genes (annotated by Pfams) did not reach an asymptote (Fig. S1).
The number of new and unique vOTUs and viral clusters at the family level (Fig. S1b and S1c) was linearly related to sequencing depth, while viral Pfams displayed slight curvature.These linear relationships were observed when considering 2TB of metagenomic sequencing--4-fold more sequencing depth than any other soil metagenome in this study and 40-fold more than the Joint Genome Institute (JGI) recommended sequencing depth for soil samples (45 GB).When considering cumulative unique attributes versus sequencing depth (Fig. S2), relationships in vOTUs and viral clusters displayed slight curvature, while viral Pfams neared saturation.

Microbial Hosts of Soil Viruses.
We connected 1,450 viruses to putative hosts of 82 different bacterial and archaeal orders with CRISPR spacers.This equates to 2.78% of QA/QC-ed viral contigs that were associated with CRISPR spacer hits; roughly 70% more host assignments than in another recent assessment 4,36 .While we observed a maximum of 73 vOTUs per host (i.e., CRISPR spacer), the mean overall vOTU per host ratio was 0.42 (median = 0), reflecting the predominance of unique host associations for individual vOTUs.
Out of 1,223 samples with at least one vOTUs assigned to a host, only 72 samples had an average of more than one host sequence per vOTU, underscoring the low abundance of detected hosts across all soils.An average of 0.64 unique host orders were detected per sample, with a maximum ratio of CRISPR spacer hits to viral sequences of 73.Further, samples with a high ratio of vOTU:host almost exclusively were matched to host sequences from a single microbial order, reflecting high phylogenetic conservation of host associations.Of the 10 samples with the highest CRISPR spacer sequence to viral sequence ratio, only 1 contained a CRISPR spacer matching more than one microbial order.
The most prevalent host taxa were distributed across distantly-related phyla, including members of prominent soil orders such as Pseudomonadales, Burkholderiales, Acidobacteriales, and Bacteroidales (Fig. 3b).The frequency of CRISPR hits associated with Acidobacteriales, Oscillospirales, Pedosphaerales, and Geobacterales were positively correlated to soil nitrogen, organic carbon, and CEC; while Enterobacterales, Obscuribacterales, Mycobacteriales, Pseudomonadales, and Streptomycetales were positively correlated to soil bulk density, and to a lesser extent, pH and clay.

Metabolic potential encoded by the soil virosphere.
Because viral gene annotations were sparsely distributed across many functions, we screened all viral genes (regardless of assignment by the AMG pipeline) against KEGG pathways to understand relationships among genes in the context of known metabolic processes.
Across the entire soil virosphere, we uncovered portions of KEGG pathways that were mostly complete, including functions involved in amino acid and sugar metabolism and biosynthesis, antibacterial mechanisms, nucleotide synthesis, and other viral functions (e.g., infection strategies, Supplemental Material and Fig. 4).Of note, we found nearly complete portions of energy-generating pathways including the pentose phosphate pathways and F-type ATPasemediated portions of photosynthesis.Lipopolysaccharide pathway-related genes (LPS) that may be important as host receptors for bacteriophage and prevention of superinfection were also prevelant 37,38 .

Global Soil Virus (GSV) Atlas.
The GSV Atlas demonstrates the immense, unexplored taxonomic and functional diversity of the soil virosphere.Viral diversity in the GSV Atlas appeared to be largely distinct from other global habitats.Nearly 80% of GVS Atlas sequences that clustered with existing sequences in IMG/VR were attributed to soil or soil-like habitats [i.e.'other terrestrial' or 'plantassociated' (rhizosphere)], underscoring the unique composition of the soil virosphere relative to more well-studied marine and human environments.Additionally, few shared vOTUs and viral clusters between samples may indicate high spatial turnover (i.e., changes in soil virosphere composition through space).Recent studies have estimated soil viral diversity is high both relative to other viral habitats and relative to soil microbial diversity 7,8,10,22 .However, these estimates have been limited by copious viral and microbial 'dark matter' for which no functional or taxonomic assignment is known 14,23,32 .Towards this end, the diversity encompassed by the GSV Atlas can serve as a community resource for characterizing this unknown fraction of the soil virosphere.
Analysis of the GSV Atlas suggests that extreme spatial heterogeneity may be a key feature of the soil virosphere at the global scale.While rapid viral spatial turnover was recently observed across short spatial scales (<10-20m) 8,10 , there has been no such demonstration of viral biogeography across the world.We propose that high rates of spatial turnover could result from low dispersal rates or distinct temporal dynamics of viral communities relative to other organisms.For example, while dormant microorganisms and relic DNA can persist for months or more [39][40][41] , the burst of viral cells associated with active infections may generate short-lived pulses of distinct viral communities that do not contribute to relic DNA due to their comparatively small genome sizes versus microorganisms.Additionally, the apparent discrepancies between microbial and viral dispersal processes could be due to the presence of free viruses that are not actively involved in microbial infection 14 , smaller viral genomes that could facilitate physical protection, differences in traits that facilitate dispersal between viruses and microorganisms, variation in bioinformatic pipelines, and/or other ecological differences between viruses and microorganisms.
Together, these factors make characterizing the soil virosphere an immense challenge for the coming decade.When examining individual soil samples, the number of new and unique viral attributes (e.g., vOTUs, clusters, Pfams) was linearly related to sequencing depth, suggesting that new viral discoveries are likely to continue with increasingly deep sequencing (Fig. S1).This contrasts with rarefaction curves of the soil microbiome and of microbial hosts of soil viruses, which both asymptoted well before sequencing depths of typical soil microbial investigations.Still, when looking at the cumulative number of unique viral attributes detected in all samples collectively (Fig. S2), many viral attributes began to saturate with sequencing depth.This suggests that while individual samples do not capture soil viral diversity, we can begin to constrain the extent of diversity when sequences from thousands of existing samples are aggregated.
Assembling the GSV Atlas enables us to begin to describe the enormous functional diversity in the soil virosphere.After filtering viral genes to putative AMGs (see methods), some KEGG pathways represented by the most putative AMGs were associated with major soil carbon cycling processes (map00052, Galactose metabolism; map00500, Starch and sucrose metabolism).Likewise, at the level of gene annotations, the most common putative AMGs suggested a role for viruses in soil carbon cycles.Glycosyltransferases 4 (GT4) and glycosylhydrolases 73 (GH73) encode enzymes involved in the metabolism and/or production of sugars common to soils including sucrose, mannose, glucosamine, and maltose 42,43 ; and Carbohydrate Binding Module 50 (CBM50) has been previously linked to chitin -one of the most abundant carbon molecules in soil 44 .
Collectively, these genes encode the decomposition of chemically complex organic material for microbial energy generation and growth.Our results are consistent with previous work from single locations that have hinted at a wide range of possible soil viral AMGs, including glycoside hydrolases, carbohydrate esterases, and carbohydrate-binding modules 15,23,31 .
Given that a large proportion of soil microorganisms are infected by viruses at any given time 45 , AMGs encoded by soil viruses have the potential to impact global biogeochemical cycles 15,22,23,31 .The thousands of putative AMGs identified here represent the most extensive survey to date and further impress the importance of the soil virosphere as a reservoir for biogeochemical potential.

Microbial Hosts of Soil Viruses.
Unraveling relationships between viruses and their host communities is imperative to understanding the impact of the virosphere on soil processes.Host presence should be tightly coupled to viral abundance, and in turn, these linkages are mediated by spatial, temporal, and environmental factors 15,46,47 .These linkages are also dependent on viral host range (i.e., host specificity) --higher host specificity should lead to stronger coupling between microbial and viral abundance and community composition.Viral host specificity is also associated with ecological factors that impact microbial community composition and may result in tradeoffs between viral growth and the breadth of the host range 11,[48][49][50] .These dynamics are unresolved in soils, leading to open questions regarding associations between viral host specificity and microbial community functions.
Our analysis of the GSV Atlas indicates that limited host infection range may be a dominant trait across soil viruses.Across the GSV Atlas, there were few hosts per vOTU on average (mean = 0.42), and of vOTUs associated with multiple host sequences, the vast majority were linked to multiple hosts of the same phylogenetic clade.While high host specificity has historically been the prevailing paradigm, our work contrasts recent studies suggesting that some soil viruses may have broader host ranges than viruses in other habitats 51,52 .
As a whole, we identified taxonomically diverse soil microorganisms as viral hosts (Fig. 3).Host sequences spanned nearly every major soil microbial clade, supporting recent studies that detected soil viral hosts classified as Acidobacteria, Verrucomicrobia, Deltaproteobacteria, and Gammaproteobacteria 22,23,31 .The ultimate impact of viral predation on soil functions is at least partially associated with the taxonomic distribution of hosts--viral impacts may be larger when the hosts contribute to key ecosystem functions.The taxonomic breadth of the microbial hosts of soil viruses also suggests a role for the soil virosphere in most soil habitats on Earth, as the microbial hosts found here occupy a wide range of ecological niches.Moreover, some hosts were susceptible to changes in the environment, underscoring the role that soil heterogeneity may play both in viral diversity and in shaping viral-host dynamics.
Relationships between environmental factors and host taxa may reflect environmental filtering on host communities (which in turn, determine the amount and type of viruses present) or on viruses directly which subsequently impacts host community composition 15,27,53,54 .Viral infections have been previously linked to soil parameters including moisture 12,30 and carbon and nitrogen content 9 .In our analysis, bulk density may serve as a proxy for hydrologic connectivity in the soil matrix.For example, low hydrologic connectivity may create 'spatial refuges' for soil bacteria from viral infections 8 , influence the virus-host encounter rates, and thus structure the soil virosphere and its hosts.Nutrient amendments are also considered to be drivers of the soil virosphere, supporting the relationship we observed between carbon, nitrogen, and host taxa.

Metabolic Potential Encoded by the Soil Virosphere.
We detected many hallmarks of viral activity in the soil virosphere, including genes associated with cell lysis, DNA repair/replication, and other infection signatures.For instance, across the entire soil virosphere, genes associated with DNA mismatch repair (map03430), homologous recombination (map03440), and base excision repair (map03410) were prevalent (Fig. S3).Folate biosynthesis was also common in the soil virosphere.Folate and other B vitamins are key to bacterial growth (map00670, map00790, Fig. 4) 55 and therefore may be logical targets for pathogens 56 .Bacterial secretion systems (map03070, Fig. S4), which may be evolutionarily derived from phages 57 , were also found in many soil viruses.Type IV secretion systems in particular can be used by bacteria to secrete toxins 58 or as a method for DNA transfer through membranes 59 .Finally, the Caulobacter cell cycle (map04112, Fig. S5), which has a distinct division pattern 60 , was rife across soils and may be a promising indicator of viral infections.The prevalence of viral genes associated with central microbial functions highlights the potential importance of viral activity in soils and the need for targeted approaches to quantify the extent and impact of viral gene expression.
The GSV Altas also contains many viral amino acid biosynthesis/degradation pathways that could be critical in viral life cycles(e.g., map00250, map00260, map00270, map00330, map00340, Fig. S6).Amino acids are building blocks for cellular material and also support soil biogeochemical cycles, as they can enhance carbon cycling through priming effects and/or enhanced nutrient availability 61,62 (e.g., map00500, map00052, map00051, Fig. 4).Collectively, these pathways demonstrate several possibilities for soil viral impacts on processes that are central to microbial metabolism and biogeochemical cycling of elements in soil.
Beyond these pathways, we highlight three KEGG pathways with near-complete portions represented in the GVS Atlas: F-type ATPase (map00190), pentose phosphate pathway (map00030), and LPS (map00540).Five of seven subcomponents of the F-type ATPase were detected in the soil virosphere, while no V-or A-type ATPases were found.Given the evolutionary similarities between V-and F-type ATPase in particular 63 , the lack of any V-or Atype ATPase components is notable in light of the near complete F-type ATPase.Though there is some basis for F-type ATPases in viral replication 64 , we also note the possible involvement of Ftype ATPase in photosynthetic energy generation 65 .Given the prominence of photosynthetic marine AMGs 26,66 , we highlight the possibility of a viral F-type ATPase as a soil AMG.The pentose phosphate pathway is also a prevalent and important AMG found in marine ecosystems, where viral infection diverts carbon towards the pentose phosphate pathway as an 'express route' of energy generation, at the expense of host carbon metabolism (reviewed in 66 ).Finally, we observed nearly complete LPS pathways in the GSV Atlas.Phages often carry depolymerases and other enzymes that target LPS or similar outer membrane components to facilitate binding and entry 37 .However, the representation of the LPS synthesis pathway by putative soil AMGs indicates that phage may work to change the function of the pathway post-infection, potentially to prevent superinfection 38 .Collectively, we propose that F-type ATPase, pentose phosphate pathway, and LPS may be interesting pathways for more targeted investigations into the role of the virosphere in soil microbiome function.

Future Recommendations.
The field of soil viral ecology is poised for rapid expansion; yet several challenges remain in fully characterizing soil viral diversity and function.Overcoming these methodological and ecological hurdles will require broad participation from global researchers.Below, we present a summary of issues, from our perspective, facing the current generation of soil viral ecologists and suggestions for surmounting them.
First, we propose methodological investments to improve viral detection and resolve genomic 'dark matter'.Metagenomic sequencing enables the detection of thousands of viruses per soil sample, yet the number of viruses detected in soil metagenomes has remained relatively flat over time 4 .In part, this is because soil metagenomics from shotgun sequencing is highly fragmented, leading to lower quality UViGs 67,68 .Identifying novel viral sequences and assigning viruses to microbial hosts are also limited by the extent of our knowledge of viral diversity; thus expansion of the known virosphere is needed.Technical advances may improve soil virus identification and host-linkage predictions from shotgun metagenomics, long read sequencing, and/or targeted sequencing approaches.Promising new methods include experimental verification of viral activity 29 , size fractionation ('viromics') 7,8,15 , viral isolation 69 , optimized viral nucleic acid extraction 70 , microscopy 29 , combined metagenomic assembly 4 , and long-read and/or single-cell sequencing 71,72 .
Knowledge about soil viral diversity and function is also limited by gaps in field and laboratory experiments.The GSV Atlas demonstrates that extensive, spatially-explicit sampling is needed to capture the high spatial turnover of the soil virosphere.The spatial coverage of most 'global' ecological studies, including this one, often suffers from large data gaps 73 .Concerted efforts are needed to sample wide spatial domains, including historically undersampled regions, given the high viral diversity uncovered by the GSV Atlas.Expansion of the known virosphere in this way will also help to facilitate tool development.Although we did not assess temporal dynamics, temporally-explicit approaches are likewise needed to characterize temporal dynamics in soil viral communities.Further, our functional annotation of viral contigs revealed diverse genes associated with functions relevant to both viral and microbial communities, and it is impossible to know the true functions of viral genes without targeted functional assays.We therefore propose that experiments targeting the expression and auxiliary metabolic function of viral genes are needed to properly assess AMGs in viral communities.
Finally, we still know relatively little about the ecological drivers of soil virus distribution or how to represent these mechanisms in process-based models.Extreme soil virosphere diversity renders some common microbial ecology statistical methods unfeasible, including those often used to test ecological principles (e.g., ordinations, distance-decay, richness, etc.).This highlights the need for innovative statistical approaches to interpret the soil virosphere and to develop new theories surrounding their ecological roles.These advances can help aid development of process-based models, which have made tremendous improvements in representing soil carbon cycles but are missing dynamics involving the soil virosphere.

Conclusions.
The GSV Atlas is a new public resource that can help generate hypotheses and provide insight into some of the most pressing challenges in soil viral ecology.We uncovered 616,935 UViGs from global soil samples to show the extreme diversity, spatial turnover, and functional potential of the soil virosphere.This includes a wide taxonomic array of microbial hosts of soil viruses, key functions associated with soil carbon cycles, and an assortment of viral metabolisms that may be critical to deciphering viral ecological principles in the soil ecosystem.We specifically highlight F-type ATPase, the pentose phosphate pathway, and LPS-related genes, as well as enzymes involved in carbohydrate metabolism, as fruitful areas for further investigation.
Our work scratches the surface of the soil virosphere and serves as a basis for tool, theory, and model development to further advance soil ecology, biogeochemistry, ecology, and evolution.
Correspondence should be addressed to emily.graham@pnnl.gov.FIGURES Fig. 1.Data collection and workflow.a, Global distribution of samples, scaled by assembled base pairs.In order horizontally, histograms of (b) mean soil bulk density (kg/dm 3 ), cation exchange capacity (cmol(c)/kg), clay content (%), total nitrogen content (g/kg), pH, and soil organic carbon (g/kg) associated with our samples from the SoilGrids250 database (0-5 cm).c, Sequence processing pipeline.

Data Collection and Curation.
We collected a total of 2,953 soil metagenomic samples from soils from major repositories and ecological networks including the JGI Integrated Microbial Genomes and Microbiomes (IMG/M) platform, MG-RAST metagenomics analysis server, Global Urban Soil Ecological Education Network (GLUSEEN), Earth Microbiome Project (EMP), and National Ecological Observatory Network (NEON) plus submissions from individual collaborators.This included 1,552 samples not previously included in IMG/M (Fig. 1 and 2).
For samples collected via JGI IMG/M, we retrieved all studies with GOLD 74 ecosystem type of "Soil" as of August 2020.We manually curated metagenomic sequences to remove misclassified data as follows.We removed samples from studies with the following: ( 1 In parallel, we retrieved mean values for soil parameters from the SoilGrids250m database from 0-5 cm 33 .SoilGrids250m is a spatial interpolation of global soil parameters using ~150,000 soil samples and 158 remote sensing-based products.Here, we focus on six parameters often associated with soil microbial communities: bulk density, cation exchange capacity, nitrogen, pH, soil organic carbon, and clay content.Because our focus on spatial dynamics and soils were collected at various times, we did not include temporally dynamic variables such as soil moisture or temperature in our set of environmental parameters, though we acknowledge they may have profound impacts on the soil virosphere.

Assembly and Annotation of Samples added to IMG/M.
To standardize data analysis across all samples, the 1,552 soil metagenomic samples not collected from IMG/M were analyzed using the JGI's Metagenome Workflow 75 .In brief, samples were individually assembled using MetaSpades v. 3.1.1,476 of the 1,552 assembled soil samples passed default quality control thresholds 76 , yielding 133 gigabases of assembled DNA in 241,465,924 contigs.Additionally, three very large metagenomes (>1TB each) were assembled separately due to computational limitations in standard workflows 77 .The resulting assemblies were assigned GOLD identification numbers and imported into IMG/M and processed using

Fig. 2 .
Fig. 2. Data description.a, shows the count of each category across the full dataset.b, shows

Fig. 3 .
Fig. 3. Relationships between soil viruses and their hosts.a, Cumulative Distribution

Frequency
e c i e s ) V i r a l c o n t i g s A n n o t a t e d v i r a l g e n e s V i r a l b a s e p a i r s V i r a l g e n e e m b l e d b a s e p a i r s H o s t ( o r d e r )V i r a l c l u s t e r s ( g e n u s ) v i r a l c l u s t e r s ( f a m i l y )

UU
M a g n e to c o c c a le s D S M − 1 6 0 0 0 R h o d o s p i r i l l a l e s S p h i n g o m o n a d a l e s R h o d o b a c t e r a l e s R h i z o b i a l e s P a r v i b a c u l a l e s A c e t o b a c t e r a le s X a n t h o m o n a d a l e s A c i d i f e r r o b a c t e r a l e s M e t h y l o c o c c a l e s T h io t r ic h a le s N it r o c o c c a le s B er ki el la le s Ps eud om ona dal es Enterobacterales C a rd io b a c te ri a le s B u r k h o l d e r i a l e s Po lya ngi ale s S ac ch ar im on ad al es S u n g b a c te ra le s M o r a n b a c te r a le s L e p t o ly n g b y a le s O b s c u r i b a c t e r a l e s A n a e r o li n e a le s T e p id if o rm a le s K te do no ba ct er al es UB A5 177 L im n o c y li n d r a le s A c i d i m i c r o b i a l e s A c t i n o m y c e t a l e s S t r e p t o s p o r a n g i a l e s M y c o b a c t e r i a l e s S t r e p t o m y c e t a l e s P r o p i o n i b a c t e r i a l e s J i a n g e l l a l e s C o r i o b a c t e r i a l e s S U R F − 1 2 I s o s p h a e r a le s S p h i n g o b a c t e r i a l e s F l a v o b a c t e r i a l e s B a c t e r o i d a l e s C y t o p h a g a l e s K a p a b a c t e r i a l e s S yn tr op ho ba ct er al es De sul fob acc ale s S y n tr o p h a le s G e o b a c te r a le s S y n t r o p h o r h a b d a le s Myxococcales H o l o p h a g a l e s B r y o b a c t e r a l e s A c i d o b a c t e r i a l e s g u o b a c t e r a l e s B a c i l l a l e s _ B B a c il la le s B a c i l l a l e s _ A R F 3 9 S ta ph yl oc oc ca le s L a c to b a c il la le s B r e v i b a c i l l a l e s P a e n i b a c i l l a l e s T h e r m o a c t i n o m y c e t a l e s Acidaminococcales Pe pto stre pto coc cal es T is si er el la le s C h ri s te n s e n e ll a le s L a c h n o s p ir a le s S a c c h a r o f e r m e n t a n a le s A c e t i v i b r i o n a l e s