Prior phylogenomic studies have recovered a well-supported subclade of six marronoid families, related as follows: (Macrobunidae, (Hahniiidae, (Cybaeidae, (Cicurinidae, (Toxopidae, Dictynidae))))); Gorneau et al. 2023; see also Kulkarni et al. 2023). We used previously published and original phylogenomic data for multiple genera for each of the five outgroup families from above (Suppl. material 1), allowing a robust test of cybaeid monophyly.

Most cybaeid phylogenomic data included here were derived from our original collections, although we also included a handful of ingroup samples from prior studies (Suppl. material 1). This included data for Mastigusa. Recent Sanger analyses support this genus as a cybaeid (Castellucci et al. 2023), but placement in phylogenomic analyses has been unstable (Gorneau et al. 2023, Kulkarni et al. 2023). We were able to sample and include original data for all described North American cybaeid genera (Suppl. material 1). Several rare genera were collected from their respective type localities, including Cybaeozyga, Neocryphoeca Roth, 1970, and Willisus (see Suppl. material 1). Our original sample also includes the important genera Ethobuella Chamberlin & Ivie, 1937 and Dirksia, which are listed as cybaeids by Gorneau et al. (2023) but lack phylogenomic data. Phylogenetic analyses of published Sanger data consistently place these two genera outside of Cybaeidae (e.g., Crews et al. 2020; Kulkarni et al. 2023; Castellucci et al. 2023).

Spiders were searched for in appropriate microhabitats and collected by hand or with an aspirator. Most spiders were preserved in the field in either 80% EtOH for morphological study, or in 100% EtOH for DNA analysis. Spiders were identified to genus using available keys (Bennett et al. 2017), and to the species level using the relevant taxonomic literature (Chamberlin and Ivie 1937; Roth 1956, 1970, 1981; Bennett 1988; Heiss and Draney 2004; Bennett et al. 2020, 2022b, 2023). Images of genitalia and/or entire spiders for voucher specimens used in UCE experiments are available on ecdysis (https://ecdysis.org); specimens are currently housed in the SDSU Terrestrial Arthropods Collection (SDSU_TAC).

Genomic DNA was extracted from leg tissues using the DNeasy Kit (Qiagen GmbH, Hilden, Germany), with at least 250 ng of genomic DNA used for UCE library preparation. UCE library preparation and library sequencing were performed at RAPID Genomics, with target enrichment performed using the Spider 2Kv1 (Kulkarni et al. 2020) or RTA Spider Clade probeset (Zhang et al. 2023).

Bioinformatic analyses were conducted on the Mesxuuyan HPC at SDSU. Raw demultiplexed reads were quality-filtered and cleaned of adapter contamination with Trimmomatic (Bolger et al. 2014), using parameters PE ILLUMINACLIP:$adaptersfasta:2:30:10:2:keepBothReads LEADING:5 TRAILING:15 SLIDINGWINDOW:4:15 MINLEN:40. Cleaned reads were assembled into contigs using SPADES v3.13.0 (Prjibelski et al. 2020) with the --isolate option. PHYLUCE (Faircloth 2016) was used to map and identify UCE loci, mapping contigs against the RTA Spider Clade probeset (Zhang et al. 2023) using default (80, 80) matching values.

We also extracted UCEs from published low-coverage whole genomes and transcriptomes in-silico using PHYLUCE (https://phyluce.readthedocs.io/en/latest/tutorials/tutorial-3.html). Raw reads were downloaded from the SRA (Suppl. material 1), cleaned, and assembled using the same parameters as with the UCE data. Default parameters were used across the entire process except for coverage and identity parameters in “phyluce_probe_run_multiple_lastzs_sqlite”, set at 80, 80 to maintain consistent matching values across all data types.

Individual UCE loci were aligned, trimmed, and filtering using FUSe (https://github.com/rmonjaraz/FUSe; Monjaraz-Ruedas et al. 2024). This included aligning with the MAFFT globapair option (Katoh et al. 2013), and trimming using trimAl (Capella-Gutiérrez et al. 2009) with the -automated1 option. We also removed highly divergent sequences (60%) (--remove-div -d 0.6) and sequences shorter than 70% of the total alignment length (--remove-short -s 0.7), retaining alignments with a minimum of 4 sequences and longer than 50bp. Subsequent 50% and 80% completeness matrices were created (--get-completeness -e 0.8 and -e 0.5). Finally, we manually curated the above alignments in Geneious Prime 2023, removing any remaining obviously divergent individual sequences, and matrices with an average pairwise identity below 80%. These matrices were named 50p_filtered and 80p_filtered, respectively.

Because several outgroup taxa revealed high sequence divergences (see also Gorneau et al. 2023), with phylogenetic placement and estimated branch lengths possibly impacted by filtering and sequence trimming method, we implemented an alternative trimming workflow by combining PhyIN (Maddison 2024) and FUSe. This workflow consisted of aligning with MAFFT using the globalpair option, followed by removal of highly divergent sequences (70% divergence) (--remove-div -d 0.7), trimming of gaps using a Simple Gappiness Filter (sgp.py) (Maddison 2024) with options -gS -gB -t -1 in combination with PhyIN using options -b 10 -d 2 -p 0.5 -e. Finally, after trimming, remaining short sequences were removed from alignments using FUSe (--remove-short -s 0.7) and 50 and 80% completeness matrices were created. These matrices were named 50p_PhyIN and 80p_PhyIN, respectively.

We extracted traditionally used Sanger loci (18S, 28S, H3) from raw reads and compared these to previously published Sanger data for various cybaeid taxa (Suppl. material 1). Our primary objective here was to better understand the phylogenetic behavior of Sanger sequences for particular taxa (e.g., Ethobuella and Dirksia, see Crews et al. 2020; Kulkarni et al. 2023; Castellucci et al. 2023), given our larger taxon sample, available voucher specimens, and well-resolved phylogenomic backdrop.

Sanger loci were harvested from UCEs, low-coverage genomes and transcriptome data using custom scripts (“loci_byCatch.sh”, on Dryad at https://doi.org/10.5061/dryad.2v6wwpzz4). A reference fasta file (“reference_cyb.fasta”, on Dryad) containing target loci was created from previously published ingroup data. Cleaned fastq files were mapped against this reference using BWA (Li 2013). Resulting BAM files were used for calling consensus sequences for each sample and locus using the consensus function of SAMTOOLS v1.16 (Danecek et al. 2021); for each mapped sample we retained the longest sequence for downstream analysis. Consensus sequences were merged, aligned, and trimmed with FUSe using the MAFFT globalpair option for aligning and trimAl -automated1 option for trimming, followed by alignment inspection in Geneious Prime 2023.

For concatenated UCE and Sanger matrices, and for individual Sanger loci matrices, we conducted maximum likelihood analyses with IQ–TREE 2 (Minh et al. 2020a). Concatenated analyses included individual loci as separate partitions, with 1000 replicates of ultrafast bootstrapping and optimal model search using ModelFinder (Kalyaanamoorthy et al. 2017). For 50p_filtered and 80p_filtered UCE matrices, we also estimated gene trees from individual alignments and calculated gene (gCF) and site (sCF) concordance factors (Minh et al. 2020b) using IQTree; sCF values were estimated with the likelihood option --scfl (Mo et al. 2023).

For 50p_filtered and 80p_filtered UCE matrices only, species trees were also estimated under a multispecies coalescent model using weighted ASTRAL (wASTRAL, Zhang and Mirarab 2022). Input gene trees for wASTRAL were estimated using IQ–TREE 2 with 1000 replicates of ultrafast bootstrapping and treated as unrooted. We used the wASTRAL hybrid scheme to weight gene trees using both long terminal branches and weakly-supported nodes. Internal ASTRAL branch lengths were estimated in coalescent units, with branch support measured as local posterior probability values (Sayyari and Mirarab 2016).

Holotype and paratype specimens have been deposited at the San Diego Natural History Museum (SDNHM) and Museo Argentino de Ciencias Naturales (MACN). All other specimens referenced with San Diego State University numbers are housed in the SDSU Terrestrial Arthropods Collection.

Specimen measurements were taken using an eyepiece micrometer at 4× magnification with an Olympus SZX12 stereomicroscope fitted with 10 × ocular lenses (SDSU) or a Leitz stereomicroscope with an eyepiece micrometer on an 8× ocular (MACN). All measurements are reported in millimeters. Appendage measurements were taken from the left appendage. Tarsal claws were examined using a Leica M205C microscope at 16× magnification.

Specimens were digitally imaged using a Visionary Digital BK plus system including a Canon 5D Mark II digital camera and Infinity Optics Long Distance Microscope (SDSU) or a Leica M205 with a DFC 295 digital camera (MACN). Individual images were combined into a composite image using Helicon Focus V6.6.2 software, then edited using Adobe Photoshop. Female spermathecae were dissected from specimens using fine forceps, immersed in BioQuip specimen clearing fluid on a depression slide, then imaged directly in this fluid on slides. Other images were taken with specimens immersed in filtered 70% EtOH, using KY jelly to secure samples.

Voucher specimen information for UCE and “Sanger loci” analyses are summarized in Suppl. material 1. Raw read data from original UCE capture experiments have been submitted to the Sequence Read Archive (BioProject ID: PRJNA1183915). Numbers of assembled contigs, and loci in the 50p_PhyIn UCE matrix, are included in Suppl. material 1. Input matrices, analysis log files, and output tree files are on Dryad (https://doi.org/10.5061/dryad.2v6wwpzz4).

UCE matrices included data for 10 outgroup genera, and 29 total ingroup samples representing 18 genera (Suppl. material 1). The stricter 80p_filtered and 50p_filtered matrices ranged in concatenated length from 130,545 base pairs (233 loci) to 294,777 base pairs (554 loci), respectively. 80p_PhyIN and 50p_PhyIN matrices ranged in concatenated length from 150,541 base pairs (355 loci) to 344,448 base pairs (874 loci), respectively.

Concatenated maximum likelihood analyses of UCE matrices recover outgroup relationships as expected (Gorneau et al. 2023), and cybaeid monophyly (Fig. 2, Suppl. material 2). Different alignment filtering workflows impacted recovered concatenated branch lengths, but topologies remained largely unchanged (Fig. 2, Suppl. material 2). Bootstrap support values for cybaeid monophyly are consistently high across UCE analyses (99 or 100). Gene tree topological variance, as measured by concordance factors (Lanfear and Hahn 2024), is moderate to high for this node (gene CF = 25, site CF = 36) in comparison to tree-wide values.

Consistent relationships within cybaeids include a Cryphoeca Clade (Mastigusa, Dirksia, Ethobuella, Neocryphoeca, Cryphoeca) sister to all other genera. The deepest branches within the family separate Cryphoeca Clade members from other primary lineages. Within the former, Dirksia is strongly supported as sister to Ethobuella, but other Cryphoeca Clade intergeneric relationships are relatively weakly supported.

Sister to the Cryphoeca Clade, all remaining genera are grouped into four main subclades, informally named here the Reduced Eyes Clade (REC), Patellar Fracture Clade (PFC), Cybaeota, and core Cybaeidae (all further discussed and defined below). Concatenated bootstrap support values in this part of cybaeid phylogeny are generally high, except for the node uniting Cybaeota and core cybaeids, and Neocybaeina plus Cybaeina Chamberlin & Ivie, 1932 within the latter. The Cybaeota plus core cybaeid node also has notably low site CF values (site CF = 32, 34), indicating high topological variance at this node (Fig. 2, Suppl. material 2). Concatenated results support Allocybaeina sister to Pseudocybaeota Bennett, 2022 with high support. We note that the overall five-clade structure for the entire family is reflected in early Sanger loci phylogenetic results of Spagna et al. (2010: fig. 6) and Crews et al. (2020: fig. 1), although these earlier efforts had sparser taxon sampling.

Species trees were estimated under the wASTRAL multispecies coalescent model for 50p_ and 80p_filtered matrices. For the 50p matrices, cybaeid monophyly and five cybaeid subclades are recovered with high posterior probabilities (posterior probability = 1; Fig. 3). Within the Cryphoeca Clade, Mastigusa is sister to Neocryphoeca, but at low posterior probability (pp = 0.45). Likewise, relationships between the PFC, Cybaeota, and core Cybaeidae are unresolved (pp = 0.45), and the sister relationship between Allocybaeina and Pseudocybaeota is poorly supported (pp = 0.45). The 80p_filtered ASTRAL results are broadly similar, but here Mastigusa is sister to all other Cryphoeca Clade members, eroding the support for this subclade (pp = 0.67; Fig. 3). PFC, Cybaeota, and core cybaeid interrelationships are again a trichotomy (pp = 0.47), and Allocybaeina is sister to all other core cybaeid members, rather than sister to Pseudocybaeota. Neocybaeina plus Cybaeina is recovered in both ASTRAL analyses with moderate support.

Figure 3. 

UCE wASTRAL species tree from 50p_filtered matrix, with posterior probability values. Branch lengths in coalescent units for internal branches only, terminal branch lengths arbitrary. Primary cybaeid subclades labeled.

Concatenated analyses of 18S, 28S, and H3 data for ingroup samples separate Cryphoeca Clade taxa from others, with the four lineages within the latter mostly recovered (Yorima GenBank sequences are placed outside of REC; Fig. 4). These analyses allow curation of some previously published sequences (see Suppl. material 1), and place Mastigusa as sister to all other Cryphoeca Clade members with high support, contra the recent analyses of Castellucci et al. (2023) where Mastigusa is sister to Cryphoeca. Mastigusa as sister to all other Cryphoeca Clade genera also finds moderate support in some UCE topologies (e.g., Fig. 2).

Figure 4. 

IQTree tree from concatenated 18S, 28S, and H3 Sanger loci, for ingroup samples only. Long branch leading to legacy GenBank data for Ethobuella and Dirksia at arrow. Primary cybaeid subclades labeled. Specimen numbers correspond to those in Suppl. material 1.

A conspicuously long branch is found within the Cryphoeca Clade, separating published Ethobuella and Dirksia sequences from those recovered from UCE raw read data. These long branches in the concatenated analysis likely reflect long branches also found on H3 and 18S topologies (Suppl. material 2). We suspect that these H3 and 18S sequences are contaminants and may explain the unexpected Ethobuella and Dirksia placements found in prior studies (Crews et al. 2020: fig. 2; Kulkarni et al. 2023: fig. 15; Castellucci et al. 2023: fig. 2).

Below we discuss possible morphological synapomorphies for the five primary phylogenomic subclades recovered within Cybaeidae (summarized in Fig. 5). Two of these groupings have been treated historically as formal subfamilies by some authors, but we here treat them informally, and do not elevate all equivalent clades to the subfamily level. We note that no prior circumscribed subfamily, as originally delimited, is recovered in phylogenomic analyses, further justifying this informal treatment.

Figure 5. 

Summary tree with primary cybaeid subclades and sampled subclade genera. Hypothesized diagnostic morphological characters for each subclade (as discussed in the text) summarized.

We also note here an interesting pattern of unbalanced species diversification, with species-poor lineages sister to particularly species-rich lineages. Examples include Blabomma plus Yorima Chamberlin & Ivie, 1942, sister to Cybaeozyga (~ 50 vs 5 species, many undescribed), Cybaeus sister to Cybaeina plus relatives (200 vs 10 species), and Calymmaria sister to Willisus (31 vs 1 species) (Table 1).

Cryphoeca Clade – Lehtinen (1967) included Tuberta Simon 1884, Cryphoeca, Dirksia, Ethobuella, and Calymmaria in his “Cryphoecinae”, a subfamily of his Hahniidae, but was ambiguous regarding defining features for the group. Roth and Brame (1972) later circumscribed a more inclusive “Group Cryphoeceae”, including many taxa now known to belong elsewhere (see below), but noted the morphological similarity of Neocryphoeca and Cryphoeca (see also Roth 1970). In his unpublished dissertation, Catley (1996) discussed morphological synapomorphies for a clade including Cryphoeca, Tuberta, Ethobuella, and Calymmaria, including male palpal tibial modifications and “coxal patches” (Catley 1996: fig. 110a, b). In an unpublished MS thesis, Lo Man Hung (2013) conducted a morphological parsimony analysis of hahniid subfamilies as recognized by Lehtinen (1967), and recovered Calymmaria as sister to Dirksia, but neither were related to Neocryphoeca. Importantly, outgroup sampling in both Catley (1996) and Lo Man Hung (2013) mostly lacked a diversity of cybaeids (e.g., those genera listed below), as cryphoecines at that time were allied with hahniid spiders.

Our Cryphoeca Clade essentially follows Lehtinen (1967), including also Neocryphoeca and Mastigusa, while excluding Calymmaria and relatives (see below). Neocryphoeca, Cryphoeca, Tuberta, and Mastigusa together share an elaborate male palpal conductor (Castellucci et al. 2024), but this is quite different from the condition found in Dirksia and Ethobuella. The placement of Cryphoecina as a relative of Cryphoeca, as hypothesized by Deltshev (1997), should be confirmed; this taxon lacks the elaborate conductor of the latter. The proposed synapomorphies of Catley (1996) do not apply, as his clade also included the distantly related Calymmaria. We are thus not aware of morphological synapomorphies applicable to the entire Cryphoeca Clade.

Reduced Eyes Clade (REC) – A clade of three genera (Cybaeozyga, Blabomma, and Yorima) including spiders with absent or highly reduced anterior median eyes. This condition is also found in close relatives outside of the family (e.g., some Lathys Simon, 1885, Brommella Tullgren, 1948, Cicurina Menge, 1871) but is elsewhere uncommon in cybaeids. REC members are otherwise morphologically heterogeneous. Blabomma and Yorima have historically resided in the larger “Group Cryphoeceae” (Roth and Brame 1972), often allied with Cicurina. These genera share simple circular female spermathecae but differ in male palpal and spinneret morphology. Cybaeozyga is a traditional core cybaeid, but because this genus is elusive and poorly known, has never been comprehensively treated (see Bennett et al. 2023). Cybaeozyga possess shorter, more contiguous, anterior spinnerets (see below). UCE concatenated branch lengths separating Cybaeozyga from Blabomma and Yorima are relatively long (Fig. 2, Suppl. material 2), and further analyses with increased sampling might ultimately distinguish these as two distinct groups. Overall, this clade needs revision, with many undescribed species; this revisionary work may serve to find more convincing morphological synapomorphies. The genus Symposia Simon, 1898 from South America also includes six-eyed species (see Müller and Heimer 1988), not unlike Yorima, and may reside (surprisingly) in this otherwise north temperate clade.

Patellar Fracture Clade (PFC) – A clade including Calymmaria, Willisus, and Siskiyu gen. nov. which share slender legs, a relatively long patella-tibia I, an elongate male cymbium (homoplastic outside of the family), and a distinct fracture line near the base of leg patella (Fig. 1; Roth 1981: fig. 1; Heiss and Draney 2004: figs 3, 134). As discussed in Roth (1981), this fracture line allows legs to drop easily or “autospasize” from this suture. Both Roth (1994) and Heiss and Draney (2004) note that patellar cleavage is also sometimes found in other cybaeids, including Ethobuella, REC genera (including a new Cybaeozyga species, see below), Cybaeota, and Cybaeina. As noted by Roth (1994), some of these latter genera autospasize “uncommonly”. We hypothesize an obvious patellar fracture line as a synapomorphy for the PFC; further study in the family is needed to understand if this character circumscribes a broader clade.

Cybaeota - Bennett (1988) cites genitalic apomorphies for male Cybaeota, including a unique RTA morphology (Bennett 1988: fig. 17), in comparison to other core Cybaeidae (see below). Marusik et al. (2020) noted how male Cybaeota palps are distinct from other core Cybaeidae, perhaps more like Calymmaria (consistent with unresolved ASTRAL relationships between the PFC, Cybaeota, and core Cybaeidae, Fig. 3). Cybaeota species also possess conspicuous pairs of ventral tibial and metatarsal macrosetae on anterior legs, which we hypothesize are homoplastic with similar macrosetae found in Cybaeina and relatives (and sometimes more distantly related spiders). Bennett (1989) documented Emerit’s Glands, broadly distributed on the integument and possibly producing repugnatorial secretions, in Nearctic species of Cybaeota. These glands were also found in Cybaeota wesolowskae from the Russian Far East (Marusik et al. 2020: fig. 2H). In Bennett’s (1989) surveys of cybaeid relatives, including representatives of all other primary lineages (including Dirksia, Cryphoeca, Blabomma, Calymmaria, Cybaeina, and various Cybaeus), he failed to find these glands. We hypothesize the combined possession of genitalic, leg macrosetae, and Emerit’s Glands as morphological synapomorphies for a distinct Cybaeota lineage.

Core Cybaeidae – with morphological synapomorphies as described by Bennett (1991). In particular, the male palpal retrolateral patellar apophysis includes peg setae (e.g., Bennett et al. 2023: figs 8–13), which we hypothesize have been secondarily reduced or lost in Pseudocybaeota (see also Bennett 1991). Internal relationships recovered here show similarities to morphological hypotheses of Bennett (1991: fig. 623), including consistent recovery of the Cybaeina subclade, with members (Cybaeina, Neocybaeina, Rothaeina) possessing conspicuous ventral tibial and metatarsal leg macrosetae. Within Cybaeus, UCE data recover a member of the California clade (C. opulentus Bennett, 2021) as sister to a clade including an unidentified Cybaeus from far northern CA (CASENT9030871) plus C. mellotteei (Simon, 1886) from Japan. We hypothesize that both latter taxa are members of the Holarctic Clade, as recovered in Sanger loci analyses (Copley et al. 2009; Sugawara et al. 2024).

Recent Sanger loci analyses including putative Sincybaeus Wang & Zhang, 2022 have placed this genus within a core cybaeid clade, possibly sister to Allocybaeina (Sugawara et al. 2024); these authors also extend the known distribution of Sincybaeus to include China, Japan, and possibly Korea. The affinities of Guicybaeus from China remain unclear. The six-eyed Guicybaeus appears to have peg setae (Wang et al. 2023: fig. 3C), indicating inclusion in core Cybaeidae, but also has conspicuously long posterior spinnerets.

The authors have declared that no competing interests exist.

No ethical statement was reported.

A grant from the National Science Foundation (DEB 1937725 to MH) funded this research. RMR was supported by the NSF and a CONACYT postdoctoral grant (# 770665/800981), and MJR by a grant from Fondo para la Investigación Científica y Tecnológica (PICT-2019–2745).

Conceptualization (MH, MR, RMR), Sample Collection and Curation (MH, MR), Methodology and Data Analysis (MH, MR, RMR), Writing – Original Draft (MH, MR, RMR), Writing – Review and Editing (MH, MR, RMR, Resources (MH, MR, RMR).

Marshal Hedin https://orcid.org/0000-0001-7683-9572

Martin J. Ramirez https://orcid.org/0000-0002-0358-0130

Rodrigo Monjaraz-Ruedas https://orcid.org/0000-0002-6462-3739

All of the data that support the findings of this study are available in the main text or Supplementary Information and Dryad (https://doi.org/10.5061/dryad.2v6wwpzz4) – Input matrices, analysis log files, and output tree files from phylogenetic analyses. Also, custom scripts for harvesting Sanger loci, including the reference fasta file.