Scorpion Group 1 discovered in the venom of Anuroctonus phaiodactylus

Scorpion Group 1 (SG1) represents a novel, highly divergent lineage of symbiotic bacteria within the class Mollicutes (phylum Mycoplasmatota). Currently, SG1’s distribution remains restricted - it has been identified exclusively within the scorpion family Vaejovidae, specifically in the genera Vaejovis and Mesomexovis. To date, SG1 has not been detected in other major scorpion families, suggesting a specialized host-symbiont relationship. In this study, we utilized the long-read capabilities of Oxford Nanopore Technology to sequence the 16S rRNA gene from the venom of two scorpion species: Paruroctonus becki (family Vaejovidae) and Anuroctonus phaiodactylus (family Anuroctonidae). We aimed to uncover cryptic microbial diversity, such as SG1, that may have been overlooked by previous short-read sequencing platforms. Our results revealed SG1 was indeed present in A. phaiodactylus but its presence was undetected in P. becki. We thus extend the host range of SG1 beyond Vaejovidae to include Anuroctonidae, refining the evolutionary boundaries of this unusual Mollicutes-scorpion association.

Given their ability to have survived for millions of years and withstand extreme environmental conditions, scorpions serve as intriguing models to study long-term evolutionary relationships with bacteria. Previous analyses of the scorpion microbiome have shown that microbial diversity is influenced by numerous factors including nutritional status, host species, geographical region, and tissue-specific microenvironments (Bolaños et al., 2016, 2019; Elmnasri et al., 2018; Shimwell et al., 2023; Murdoch et al., 2026).

While many of these studies used gut or telson tissue, Murdoch et al. (2026) was the only study that focused on the microbiota of the venom secretion itself. Animal venoms are composed of a complex mixture of compounds with unique biological functions (Utkin, 2021). Venom serves primary ecological roles, such as in defense and host-prey interactions (Schendel et al., 2019). Using Illumina 16S rRNA gene sequencing, Murdoch et al. (2026) characterized the venom microbiome of Anuroctonus phaiodactylus (Wood, 1863) and Paruroctonus becki (Gertsch and Allred, 1965). While these species were found to share a core microbiome, their overall microbial communities remained distinct. Given these findings, investigating the bacterial communities within the venom is important to understanding the microbial dynamics that persist within the complex and selective environment of the venom.

Many prior studies of scorpion microbiomes revealed interesting taxa, including those of the class Mollicutes, phylum Mycoplasmatota (Bolaños et al., 2016, 2019; Elmnasri et al., 2018; Shimwell et al., 2023). Of particular interest is the novel Mollicutes lineage SG1, or Scorpion Group 1. SG1 is a highly divergent and little-studied lineage of symbiotic bacteria that has only been documented in scorpions (Bolaños et al., 2016, 2019). Thus far, SG1 has only been found within members of family Vaejovidae in Mexico, with high frequency in Vaejovis smithi and some Mesomexovis species (Bolaños et al., 2016, 2019; García-Santibáñez et al., 2022). Whole genome sequencing of SG1 from V. smithi revealed several genomic features that distinguish SG1 from other Mollicutes, such as having a high density of genes for glycerol and polyamine transporters and containing genes for the Arginine Deiminase pathway. These differences are thought to allow SG1 to survive in the toxic environment of the venom gland (García-Santibáñez et al., 2022). SG1 has so far not been detected in other major scorpion families, including Caraboctonidae, Euscorphiidae, Diplocentridae and Buthidae (Bolaños et al., 2019). Whether SG1 could be found in families beyond these is currently unknown.

In the present study, we sequenced venom DNA from A. phaiodactylus and P. becki using Oxford Nanopore Technology (Nanopore) to detect SG1 16S rRNA genes. These species occur farther north than previously sampled scorpion hosts and are among the few that extend into the Great Basin Desert, a colder desert system not represented in earlier SG1 studies. In P. becki, colonization of much of the Great Basin appears to have followed climatic warming after the Last Glacial Maximum (Graham et al., 2013). A. phaiodactylus may show a similar biogeographic history, raising the possibility that SG1 has persisted through recent host range expansion into newly available northern desert habitat. However, based on its previously reported association with vaejovid scorpions, we hypothesized that SG1 would be detected in P. becki, thus extending its known distribution to another vaejovid host, but not in A. phaiodactylus, which belongs to the more distantly related family Anuroctonidae.

Scorpion and venom collection

The scorpion specimens and venom used in this study were used in our previous study (Murdoch et al., 2026), which outlines the scorpion and venom collection protocols in more detail. Briefly, scorpions were collected at night using handheld ultraviolet lights (Stahnke, 1972). Included in this study were 16 A. phaiodactylus (3 males, 13 females) and 14 P. becki (sex undetermined) specimens from four distinct localities in the Mojave and Great Basin deserts (Permit Number - GW-210220004-21022-001, California Department of Fish and Wildlife). Scorpions were sterilized with ultraviolet light and/or ethanol prior to venom collection. Strategies to elicit venom ejection included agitation and electrical stimulation (Murdoch et al., 2026). Venom volumes ranged from about 3 µl to 20 µl per sample. The venom was spun down and DNA extracted immediately or after storage for about two weeks at -20 °C. P. becki samples from Alico, CA (n=5) were processed together; all other P. becki samples (from Tonopah, NV; n=9) were processed together with all of the A. phaiodactylus samples (various locations; n=16). Supplementary Table 1 provides additional sample information including collection date, geographical location, sex, etc.

DNA extraction from scorpion venom and prescreen for 16S rRNA genes

DNA was extracted using the methods from Shimwell et al., 2023 and Murdoch et al., 2026, using a Master Pure DNA Kit (Biosearch Technologies). To assess contamination potential, samples lacking venom were processed through all extraction steps as negative controls. All venom samples and negative controls were prescreened using PCR for 16S rRNA genes as outlined in Murdoch et al., 2026. None of the negative controls produced 16S PCR products and thus were not subsequently sequenced.

Nanopore 16S library preparation and sequencing

Libraries were prepared according to the Oxford Nanopore Technologies (ONT) instructions using the 16S Barcoding Kit V14 (SQK-16S114.24) with minor changes. Because venom DNA was limited, 20 ng rather than the suggested 30 ng of DNA was used as a template for 16S PCR using LongAmp Hot Start Taq 2X Master Mix (NEB, M0533S), 10 ul of 16S indexed primers (ONT) and PCR water added to a final volume of 50 ul. PCR cycles were as follows: 1 cycle - 95 °C, 1 minute; 3 cycles - 95 °C, 20 seconds, 50 °C, 30 seconds, 65 °C, 2 minutes; 38 cycles - 95 °C - 20 seconds, 55 °C, 30 seconds, 65 °C, 2 minutes; 1 cycle - 65 °C, 5 minutes, 4 °C hold. DNA concentrations of PCR samples were quantified with Qubit 4 Fluorometer (Invitrogen) and normalized equimolar amounts were pooled prior to cleaning with AMPure XP beads (Beckman Coulter). Rapid adapters were added to a 100 fmol library, followed by preparation of the sequencing mixture and loading onto a primed Flongle Flow Cell (R10.4.1; FLO-MIN114). Sequencing occurred for about 24 hours on a MinION Mk1C. MinKNOW software was used for high accuracy base calling.

SG1 sequence analysis

Raw reads were processed in Geneious Prime v2025.2.2. Using the BBDuk plugin with default parameters, we trimmed both ends of the sequences to a threshold of Q ≥ 10 and removed short, uninformative reads. This preprocessing step filtered out low-quality bases and potential artifacts before downstream mapping. We used the MF134725.1 SG1 16S rRNA gene sequence from Vaejovis smithi (NCBI GenBank) as a reference to search for similar sequences within our A. phaiodactylus and P. becki samples in Geneious Prime under custom sensitivity and no fine-tuning (Bolaños et al., 2019). We searched a total of 19,890 reads for A. phaiodactylus (16 independent samples plus 6 samples that were run in duplicate) and 768 reads for P. becki reads (14 independent samples). Supplementary Table 1 shows the total reads and SG1-mapped reads per sample. Specimens with greater than 5 reads that mapped to the reference sequence were processed for consensus sequence generation. The reference sequence was removed from these assemblies and low-quality 5’ and 3’ ends were manually trimmed. Multiple sequence alignment was conducted using MUSCLE with default parameters in Geneious Prime. Alignments were subsequently used to form consensus sequences under custom sensitivity, and then low-quality 5’ and 3’ ends were manually trimmed.

SG1 phylogenetic analysis

Twenty-six SG1 sequences previously found in Vaejovidae scorpions (Bolaños et al., 2016, 2019; García-Santibáñez et al., 2022) and the 9 consensus sequences from this study, deposited in NCBI GenBank accession numbers PZ384969–PZ384977, were used to construct a phylogenetic tree. The sequences were aligned using MUSCLE in MEGA version 11 (Tamura et al., 2021; Edgar, 2004) with default parameters and manually trimmed to the conserved region. Under the Bayesian Information Criterion on MEGA version 11, HKY was determined as the appropriate best-fit model. A phylogenetic tree was created with Bayesian Inference in MrBayes 3.2.7a (Ronquist et al., 2012). MrBayes was run for 775,000 MCMC generations with a 25% burn-in utilizing the HKY best-fit model. Two runs were completed with 4 chains each. Using TreeAnnotator v10.5.0, part of the BEAST 1.10 package (Suchard et al., 2018), run 1 maximum clade credibility target tree with a burnin of 155 trees was generated. Run 1 maximum clade credibility target tree was then used as a user target tree for run 2 to generate the final tree file used for the visualization of the phylogenetic tree in FigTree v1.4.4 (Suchard et al., 2018). Convergence was confirmed via Tracer v1.7.2 (Rambaut et al., 2018).

Haplotype network analysis

We used PopArt v.1.7.2 (Leigh and Bryant, 2015) to construct a haplotype network from the final aligned SG1 dataset to visualize relationships among scorpion-associated SG1 sequences. Because relationships among closely related haplotypes are often not strictly bifurcating, we used the median-joining approach, which favors short mutational connections and integrates alternative minimum-spanning relationships into a single network. This approach is well-suited for comparing SG1 haplotypes recovered from individuals of the same scorpion species (A. phaiodactylus). In the network, each node represents a distinct haplotype, node size is proportional to the number of samples represented, and branch hash marks indicate the number of mutational differences between connected haplotypes.

Detection and frequency of SG1 in P. becki and A. phaiodactylus venom

Using Nanopore 16S rRNA sequencing we were unable to detect SG1 sequences in 14 P. becki venom samples (Figure 1A). While 768 reads were analyzed across the P. becki samples, and knowing P. becki to be a vaejovid, the lack of SG1 is likely due to the low read count limiting the detection of rare transcripts, rather than a true biological absence. In contrast, we detected SG1 in 6 out of 16 (38%) A. phaiodactylus samples (Figure 1A). From 19,890 total reads, 137 mapped to SG1; a frequency of 1 in every 145 reads. Four venom samples yielded only one or two SG1 reads, each with a unique haplotype, but were not further analyzed because the low read depth was deemed insufficient to distinguish between a sequencing error and true haplotype (Figure 1B). Using Geneious Prime, SG1 mapped to 88 reads forming 5 haplotype consensus sequences from one sample, and 44 reads forming 4 haplotype consensus sequences from another (Figure 1B). The read lengths of the SG1 sequences from these nine haplotypes ranged from 959 to 1551 base pairs and averaged 1469 base pairs, thus covering the majority of the 16S rRNA gene. All sequences from A. phaiodactylus compared to SG1 in Vaejovids had percent identities above 97% indicating similarity at the genus level. There was no significant difference between SG1 and either scorpion sex (Fisher’s exact test p=0.518) or geographical location (Freeman-Halton extension of Fisher’s exact test p=0.119).

16S rRNA analyses shows A. phaiodactylus SG1 is monophyletic

We constructed a 16S rRNA phylogenetic tree using the sequences newly generated here from A. phaiodactylus and previously published sequences from vaejovid scorpions to determine the evolutionary relationship of these SG1 sequences (Figure 2A). The phylogeny depicts two major clades, most notably, showing that all A.phaiodactylus SG1 sequences form a distinct clade that is unique and highly divergent from the previously characterized Vaejovidae SG1 lineages. While the majority of vaejovid SG1 sequences grouped together in a separate clade, the haplotypes from A. phaiodactylus venom remained distinct from the M. aff. oaxaca sequence (MF124732.1) (Bolaños et al., 2019). This relationship was strongly supported by a posterior probability of 1.0. However, most nodes within the A. phaiodactylus clade were poorly supported. This tree structure was confirmed using a Maximum Likelihood method as well (Supplementary Figure 1).

Using an alternative approach to estimate evolutionary divergence, we generated a haplotype network with the A. phaiodactylus and vaejovid SG1 sequences. The network comprised six vaejovid (Figure 2B, blue) and two A. phaiodactylus (Anuroctonidae) haplotypes (Figure 2B, red) organized in a linear topology. Both groups yielded a large central haplotype and one to four smaller peripheral haplotypes. One of the vaejovid haplotypes (Figure 2B, arrow) clustered more so with Anuroctonidae, likely due to the number of mutations (seen as tick marks) that distance this haplotype from the rest of the vaejovid haplotypes. This network confirms the phylogenetic data with the SG1 haplotypes from Anuroctonidae being largely divergent from those of Vaejovidae.

Our findings show for the first time that a unique SG1 lineage is found in the Anuroctonidae family. Long-read 16S rRNA Nanopore sequencing identified 9 A. phaiodactylus haplotypes that were distinct from SG1 sequences found in vaejovid scorpions. This finding contradicts the prevailing belief that SG1 is unique to the scorpion family of Vaejovidae (Bolaños et al., 2016, 2019; García-Santibáñez et al., 2022). The detection of SG1 in A. phaiodactylus expands the known host range of this symbiont and suggests an alternative evolutionary history than previously known.

Although it can be difficult to definitively rule out contamination in microbial studies like this, several lines of evidence suggest that the SG1 sequences detected were not the result of contamination. Importantly, negative controls lacking venom were processed through all DNA extraction steps; these, along with no-template PCR controls in the prescreen, failed to produce 16S rRNA products and were excluded from downstream sequencing. While index-only controls were not included during library preparation, results were consistent across duplicate sequencing runs. Of six samples sequenced in duplicate, one repeatedly showed SG1 sequences, while SG1 was consistently undetected in the other five. Furthermore, despite all specimens being processed in one of two batches, sharing identical timelines for collection through to DNA extraction, the unique haplotypes recovered across samples, even with low SG1 reads, indicate that these sequences likely represent reliable biological variation rather than systemic or cross-sample artifacts/contamination. This is consistent with our previous work on these same specimens (Murdoch et al., 2026), where beta diversity metrics demonstrated a statistically significant difference between the microbial composition of the venom compared to exterior controls. Because those prior results confirmed that the sequences identified in the venom were not due to environmental contamination, exterior controls were not subsequently sequenced using Nanopore in this study.

In our aforementioned prior study that used the same samples from A.phaiodactylus and P. becki, we tested the venom for 16S rRNA sequencesusing Illumina short-read sequencing. Despite analyzing approximately 2.5 million reads, however, we were unable to detect SG1 (Murdoch et al., 2026). We suspect the inability to detect SG1 using the Illumina platform might be explained by primer bias (Johnson et al., 2019). Alignment of our Anuroctonidae SG1 haplotypes with Illumina sequencing for 16S rRNA identified a seven base pair difference in the reverse primer compared to the SG1 region (Supplementary Figure 2). These differences were likely sufficient to prevent primer binding and subsequent amplification, thus, showing that short-read sequencing may not always be effective for finding SG1 or other Mollicutes lineages.

In contrast, Nanopore sequencing with its long-read capability was sufficient to find SG1 in our A. phaiodactylus samples (Figure 1). Almost 40% of our A. phaiodactylus samples had SG1 sequences, with an estimated frequency of 1 in every 145 reads. Although four of six specimens yielded only one or two sequences, these are unlikely to be the result of low-level contamination, as these haplotypes were distinct from those in the other two individuals, suggesting they may represent genuine, low-abundance variants. In comparison, Bolaños et al. (2016) reported SG1 in 94% of Vaejovis smithi samples and at a 1 in 55 OTU frequency, although the frequency per read was unknown. The reduced frequency of SG1 in A. phaiodactylus compared to V. smithi samples suggests that SG1 may not be a major component of A. phaiodactylus venom. Alternatively, SG1 may be more frequent than our current estimate and difficult to detect. Moreover, the lack of detection of SG1 in P. becki could be due to the lower sequencing depth rather than a true absence of the symbiont (Figure 1). The limited read number may be attributed to the small volume of venom released per individual (3 ul to 20 ul per sample), whether from P. becki or A. phaiodactylus, and the resulting low DNA yields despite normalization prior to sequencing. In our study using 16S rRNA Illumina sequencing of venom, after bioinformatic processing about one third of the P. becki samples had insufficient sequencing reads for inclusion in the final analysis (Murdoch et al., 2026). Increasing our sample size and testing tissues beyond the venom, that could provide higher DNA yields, may shed more light on the frequency of SG1 in these scorpion species, in other scorpion species, and perhaps in other organisms.

Given that the haplotypes identified in this study are highly divergent from those recovered in vaejovids, our discovery of SG1 in Anuroctonidae could mean that SG1 has had a long-term, stable association with this host lineage. This is supported by our phylogenetic analysis in which SG1 forms a distinct clade with a high posterior probability (Figure 2). Although our sequences show phylogenetic proximity to the M. aff. oaxaca SG1 sequence, the amount of mutational differences suggests an ancient divergence rather than a recent host-switching event. This conclusion is further reinforced by the mutational steps and genetic distance shown in the haplotype network (Figure 2).

The placement of the M. aff. oaxaca SG1 sequence complicates a simple scenario of SG1 cospeciation in scorpions. This sequence does not fall within the main vaejovid SG1 cluster, but was instead recovered closer to the A. phaiodactylus SG1 haplotypes in both the phylogeny and haplotype network (Figure 2). As a result, the sampled SG1 sequences from Vaejovidae are not monophyletic with respect to the A. phaiodactylus SG1 lineage. This pattern suggests that the SG1 history is likely more complex than a strict host-symbiont cospeciation model. Bolaños et al. (2019) reported SG1 from vaejovid scorpions and noted an apparent correspondence between host and symbiont phylogenies, but also emphasized that limited sampling prevented a formal test of cospeciation. Our results remain consistent with an ancient association between SG1 and scorpions, but the paraphyletic placement of the M. aff. oaxaca sequence raises the possibility that additional processes are involved, including ancient SG1 diversity, incomplete sorting of symbiont lineages, or host switching. Broader sampling across vaejovids, anuroctonids, and other scorpion families will be needed to distinguish among these alternatives. Moreover, whole genome sequencing of M. aff. oaxaca SG1 and A. phaiodactylus SG1 would be interesting to compare any potential functional adaptations or metabolic differences that may allow SG1 to survive across taxonomically distant scorpion hosts.

In summary, our results show that SG1 in scorpions is more deeply structured and less taxonomically restricted than previously recognized. The discovery of a distinct SG1 lineage in A. phaiodactylus causes one to revisit the finding that this symbiont is confined to vaejovid hosts and points instead to a more complex evolutionary history within scorpions. Equally important, this lineage in A. phaiodactylus, remained hidden until examined with long-read sequencing, underscoring how biological structure can go unnoticed dependent upon the methodology utilized. More broadly, our findings suggest that additional SG1 diversity may still be hiding in unsampled or overlooked scorpion lineages, waiting to be revealed.