Biology of Leptonetidae

Leptonetids are a lineage of small, rarely encountered spiders that live in moist habitats such as leaf litter, under rocks, and especially in caves. The family includes 21 genera and 355 species placed into two subfamilies, Archoleptonetinae and Leptonetinae (World Spider Catalog, ver. 19.5, see  http://wsc.nmbe.ch, accessed 19 August 2020). The archoleptonetines (Fig. 1A) include eight species in two genera and are known from the western USA, southern Mexico, Guatemala, and Panama. Leptonetines (Fig. 1B) are more diverse (21 genera, 355 species) and have a Holarctic distribution with centres of diversity in North America, Mediterranean Europe, and east Asia. Although the archoleptonetines have few features that make them readily diagnosable by non-specialists, all leptonetines share a unique eye arrangement where the posterior median eyes are displaced from the main eye group (Ledford and Griswold 2010, fig. 24, 27, 28).

Among spiders, leptonetids are best known for their association with caves. Over 50% of described species are known only from caves and many species show a range of troglomorphic morphologies including eye reduction, depigmentation, and appendage elongation (Mammola and Isaia 2017). Most species are small (2–5 mm) and reside in delicate sheet webs from which they hang (Fig. 1C). Leptonetids are microhabitat specialists, preferring environments where moisture, temperature and humidity remain stable. Ideal habitat includes breakdown debris in caves and layered rock piles in heavily shaded areas. Observations of reproductive biology have been reported (Cokendolpher 2004; Ledford 2004; Ledford and Griswold 2010) but most aspects of their life history are unknown.

Given their habitat preferences, most species have distributions that are highly localised. Sympatry is rare and known only in a few surface-dwelling populations (Ledford 2004). Even in localities where leptonetids are known to occur, they are rarely encountered and in some regions are recognised as threatened species (US Fish and Wildlife Service 2020). Although gaps in distributional range may be partly explained by inadequate sampling, we propose that the combination of specific habitat preferences and the repeated pattern of narrow endemism for most species worldwide supports a hypothesis of dispersal-limitation for the family. The biological characteristics of limited dispersal ability and high microhabitat preference often lead to biogeographic histories that are dominated by vicariance, with sometimes rare dispersal events, as seen in many other arachnid lineages (e.g. Harrison et al. 2016; Hedin and McCormack 2017; Baker et al. 2020).

Fig. 1.

Images of live spiders in native habitats. (A) Archoleptoneta schusteri, Marin County, CA, USA; (B) Calileptoneta helferi, Mendocino County, CA, USA; (C) Tayshaneta myopica in sheet web, Travis County, TX, USA. Scale: 3 mm.

Taxonomic history

Most research on leptonetids has focused on improving understanding of α-level diversity. The western European fauna is arguably the best known due to its long history of study (Simon 1872) and the detailed descriptive efforts of Brignoli (1967a, 1967b, 1968, 1971, 1974a, 1974c, 1978, 1979a, 1979b, 1979c), Fage (1913, 1931, 1943), Kratochvíl (1935, 1938, 1978), Machado and Ribera (1986), Ribera (1978, 1988), and Ribera and Lopez (1982). Although most of these works are regionally focused, Fage (1913) treated the European fauna comprehensively and Brignoli (1970, 1979d) provided a global interpretation of leptonetid relationships and biogeography. The North American fauna has a history of monographic study starting with Gertsch (1971, 1974) who described the majority of species and Platnick (1986) who delineated three North American genera. Brignoli (1974b, 1977, 1979e) also worked on the North American fauna, providing a global perspective that sharply contrasted with Gertsch (1971, 1974). Gertsch (1971, 1974) argued that all of the North American leptonetids should be placed in the European genus Leptoneta but, unlike Brignoli, had limited understanding of other European leptonetid genera. By contrast, based on his experience working with European leptonetids, Brignoli hypothesised that the North American fauna likely included multiple genera (Brignoli 1972, 1977). Two studies (Ledford and Griswold 2010; Ledford et al. 2011) assessed the phylogeny of North American leptonetids using nucleotide sequence data and better defined the genera. During the past 10 years, the most striking advances in leptonetid taxonomy have been in Asia where over 100 new species have been described from China and South Korea (Chen et al. 2010; Lin and Li 2010; Wang and Li 2010, 2011; Seo 2015a, 2015b, 2016a, 2016b; Guo et al. 2016; He et al. 2019; Xu et al. 2019). Although most of these studies do not include phylogenetic or biogeographic analyses (but see Wang et al. 2017), they have greatly improved our understanding of the Asian fauna.

Leptonetidae have a long history of controversy surrounding their relationships to other spiders. Fage (1913) was the first to recognise that leptonetids share convergent morphology with other subterranean-adapted spiders, and Brignoli (1979d) suggested that the hypothesised relationships of leptonetids to other spiders were largely based on these convergent features. One example is that leptonetines, Telemidae, and some Ochyroceratidae have independently evolved a peculiar arrangement of the posterior spinnerets where there is a single row of aciniform gland spigots used to build sheet webs (Ledford and Griswold 2010, fig. 65–72). For this reason, many workers have placed leptonetids among the Synspermiata (formerly Haplogynae), usually as sister to Telemidae, despite the fact that much of their morphology conflicts with this position. The discovery of a cribellum in Archoleptoneta Gertsch, 1974 further compounded these conflicts and, given the traditional placement of leptonetids within Synspermiata, had implications for the interpretation of spinning organs as a whole (Ledford and Griswold 2010). Brignoli (1979d) was the first to propose that leptonetids might be more closely related to entelegynes based on the absence of a cheliceral lamina and the presence of expandable male genitalia. Ledford and Griswold (2010) reviewed leptonetid morphology and were the first to suggest the possibility of leptonetid paraphyly, proposing that at least the archoleptonetines are not part of Synspermiata, but instead are more closely related to entelegynes.

Agnarsson et al. (2013) used a supertree approach and was the first to suggest a relationship between Archoleptoneta and austrochiloids, but thought that the result was caused by long-branch attraction. Using transcriptomes, Garrison et al. (2016) recovered leptonetids outside of Synspermiata, placing Calileptoneta Platnick, 1986 as sister to Entelegynae. Several studies built upon the foundation of Garrison et al. (2016), including Shao and Li (2018) who recovered leptonetines as sister to entelegynes but did not include austrochiloids as part of their study. Fernández et al. (2018) added transcriptomes for both leptonetid subfamilies (Archoleptoneta and Calileptoneta) and in their preferred topology recovered a monophyletic Leptonetidae sister to austrochiloids (Fig. 2). Based on a combination of multigene nucleotide data and morphology, Wheeler et al. (2017) recovered a polyphyletic Leptonetidae, but did support a relationship between Archoleptoneta and austrochiloids. As part of a study on basal araneomorphs, Ramírez et al. (2021) included better representation across Leptonetidae and used ultraconserved elements (UCEs) to hypothesise a sister-group relationship between Leptonetinae and austrochiloids, rendering Leptonetidae paraphyletic.

Fig. 2.

Alternate hypotheses of leptonetid relationships from recent studies: (A) Fernández et al. (2018), fig. 1A (transcriptomes); (B) Wheeler et al. (2017), fig. 3 (morphology + single genes); and (C) Ramírez et al. (2021), fig. 2 (ultraconserved elements). Figures represent summary trees. Support values for relevant nodes are indicated.

Systematics and biogeography of Leptonetidae

In this paper, we present a phylogenomic analysis of Leptonetidae including representatives from across the geographic range of the family. Both genera of archoleptonetines and 90% of described leptonetine genera are sampled. Where possible, multiple exemplars for each genus are used, including several from type localities. We include a broad sample of austrochiloids and use Telemidae as the outgroup. Because this study is an extension of a parallel study on basal araneomorphs (Ramírez et al. 2021), we do not include a broad sampling of entelegynes in order to assess the relationships of leptonetids to other spiders. Instead, we focus on testing the monophyly of leptonetid subfamilies and explore relationships among leptonetid genera. We use our results to review the current taxonomy of Leptonetidae and identify areas in need of taxonomic revision. As the first study to produce a global phylogeny of Leptonetidae, we also gain insight into the historical biogeography of the group. Using a combination of fossils and dates for continental events, we present a chronogram and assess the biogeography of the family using dispersal–vicariance analysis. Our results provide a robust phylogenetic framework for the family that can be used as a scaffold for forthcoming taxonomic works.

Taxon sampling

Representatives for 19 of the 21 described genera in Leptonetidae were sampled from localities worldwide (Fig. 3). Our sample does not include the genera Masirana Kishida, 1942 and Rhyssoleptoneta Tong & Li, 2007. We gathered original UCE data for 28 specimens, combined with data for 17 specimens taken from previous studies (Wood et al. 2018; Ramírez et al. 2021) (  Table S1 (IS20065_AC.pdf) of the Supplementary material). Depending on availability, multiple species within a genus were used. We included a broad sample of Austrochiloidea and used Usofila pacifica Banks, 1894 (Telemidae) as the outgroup.

Fig. 3.

Distribution map of archoleptonetid and leptonetid samples included in this study.

DNA extraction

Most specimens were preserved for DNA studies (preserved in high percentage ethyl alcohol at –80°C), and genomic DNA was extracted from leg tissue using the Qiagen DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA, USA). For a handful of tissues preserved in 70–80% ethyl alcohol we used standard phenol chloroform extractions with 24-h incubation for lysis. Extraction type for each specimen is indicated in  Table S1 (IS20065_AC.pdf). Extractions were quantified using a Qubit Fluorometer (Life Technologies, Inc.) and quality was assessed by agarose gel electrophoresis. Between 11 and 500 ng of total DNA was used for UCE library preparation.

UCE data collection

UCE data were collected in multiple library preparation and sequencing experiments. Up to 500 ng of genomic DNA was used in sonication, using a Covaris M220 Focused-ultrasonicator. Library preparation followed methods previously used for arachnids, as in Starrett et al. (2017), Derkarabetian et al. (2018, 2019), and Hedin et al. (2018a, 2018b). Target enrichment was performed using the MYbaits Arachnida 1.1K kit (ver. 1, Arbor Biosciences; Faircloth 2017) following the Target Enrichment of Illumina Libraries protocol (ver. 1.5, see  http://ultraconserved.org/#protocols). Libraries were sequenced with an Illumina HiSeq 2500 with 125 bp paired-end reads (Brigham Young University DNA Sequencing Center).

Matrix filtering and assembly

Raw demultiplexed reads were processed with the PHYLUCE pipeline (ver. 1.6, see  https://phyluce.readthedocs.io/en/latest/; Faircloth 2016). Quality control and adaptor removal were conducted with the Illumiprocessor wrapper (B. C. Faircloth, see  https://github.com/faircloth-lab/illumiprocessor). Assemblies were created with Velvet (ver. 1.2.10, see  https://www.ebi.ac.uk/∼zerbino/velvet/; Zerbino and Birney 2008) and Trinity (ver. 2.11.0, see  https://github.com/trinityrnaseq/trinityrnaseq/wiki; Grabherr et al. 2011), both at default settings. Assemblies were combined for probe matching, retrieving assembly-specific UCEs and overall increasing the number of UCEs per sample relative to using only a single assembly method. Contigs were matched to probes using minimum coverage and minimum identity at liberal values of 65. UCE loci were aligned with MAFFT (ver. 7.471, see  https://mafft.cbrc.jp/alignment/software/ Katoh and Standley 2013) at default settings and trimmed with Gblocks (ver. 0.91b, see  http://molevol.cmima.csic.es/castresana/Gblocks/Gblocks_documentation.html; Castresana 2000; Talavera and Castresana 2007), with settings –b1 0.5 –b2 0.5 –b3 6 –b4 6 in the Phyluce pipeline.

We expected some paralogy at the low minimum coverage values used, but considered this as a tradeoff given the ancient divergences considered. In addition to internal checks for paralogy in Phyluce, we checked for paralogy by conducting RAxML analyses on all individual Phyluce alignments. We excluded individual loci that failed to recover the well supported clade Austrochiloidea. We did not exclude duplicate UCE loci, as found in Hedin et al. (2019) – these are distinct UCE alignments that match the same protein, but are expected to be well separated by long introns. Two data matrices were assembled for phylogenomic analyses: (1) 65% (36 of 55 terminals) occupancy matrix, exon + intron, no ‘paralogs’ matrix, and (2) the same matrix as above using very strict Gblocks settings (–b1 0.5 –b2 0.85 –b3 4 –b4 8) to further trim alignments. We visually checked to confirm that these trimmed alignments comprised mostly exon data. These two matrices were developed to test the sensitivity of results to different alignment parameters (e.g. see Portik and Wiens 2020), and are referred to as ‘relaxed’ and ‘strict’ throughout the paper.

Phylogenomic analyses

Partitioned concatenation and coalescent-based analyses were conducted. Maximum likelihood analysis of the concatenated datasets was performed using RAxML (ver. 8, see  https://cme.h-its.org/exelixis/web/software/raxml/; Stamatakis 2014) with the data partitioned by locus, the GTR+Γ substitution model, and support estimated by 500 bootstrap pseudoreplicates. Maximum likelihood analysis of the concatenated datasets was also performed using IQ-TREE (ver. 1.7-beta9, see  http://www.iqtree.org/; Nguyen et al. 2015; Minh et al. 2020a) in order to calculate gene (gCF) and site (sCF) concordance factors for the relaxed and strict data matrices (Minh et al. 2020b). These measures were used as alternative measures of support (Ane et al. 2006), especially for nodes where topology conflicted between analysis type. An SVDQuartets analysis (ver. 1.0, see  https://www.asc.ohio-state.edu/kubatko.2/software/SVDquartets/; Chifman and Kubatko 2014, 2015) was conducted on both matrices using PAUP* (ver. 4.0a, see  https://paup.phylosolutions.com/), implementing a multispecies coalescent tree model with exhaustive quartets sampling and 100 bootstrap replicates.

Molecular dating analysis

A molecular clock analysis was conducted in a Bayesian framework with the MCMCtree module in the PAML package (ver. 4.9i, see  http://abacus.gene.ucl.ac.uk/software/paml.html; Yang 2007). For the clock analysis we used the relaxed matrix RAxML topology, and treated the concatenated data as unpartitioned. The outgroup was removed from the tree and dataset, and a maximum boundary of 240 Ma was applied to the root node, based on the upper boundary of the 95% highest posterior density credible interval of the Leptonetidae + Austrochiloidea node from Fernández et al. (2018). The analysis was run with a model that assumes independent rates among branches (Yang and Rannala 2006; Rannala and Yang 2007). Estimation of the parameters (shape and scale) of the gamma distribution for the substitution rate prior (μ) was done with Baseml based on four fossil and biogeographic-based calibration points treated as fixed (shape = 1, scale = 5). We used shape = 1 and scale = 4.5 for the gamma distributed prior for σ2, or the variability in substitution rate among branches. The HKY sequence model was used and the analysis was run with birth rate, death rate, and species sampling priors of 2, 2 and 0.1 respectively. Gamma priors for κ (the transition/transversion ratio) and α (shape parameter for among site rate variation) were left as default (Yang 2007). Calibrations (see below) were treated as soft boundaries (i.e. 0.025 probability date falls beyond boundary; Yang and Rannala 2006; Inoue et al. 2010). The first 40 000 iterations were discarded as burnin, followed by 40 000 iterations sampled with 100 iterations (4 million generations). The analysis was run twice to ensure MCMC convergence, with negligible differences in posterior date estimates and 95% highest posterior density credible intervals occurring between the two runs.

Our clock analysis was calibrated based on fossils in Burmese Amber (Wunderlich 2008, 2012) and two continental events in Mediterranean Europe (Lymberakis and Poulakakis 2010; Garcia-Castellanos and Villaseñor 2011). Leptonetinae was assigned a minimum age of 98.19 Ma based on the fossil Palaeoleptoneta calcar Wunderlich, 2008 from Burmese Amber. Burmese Amber has been radiometrically dated to 98.79 ± 0.6 Ma (Selden and Ren 2017). Although this fossil cannot be confidently placed among extant genera (Magalhaes et al. 2020), it appears to share the ocular arrangement of leptonetines and we interpret it as part of the Leptonetinae (Wunderlich 2012). The Messinian Salinity Crisis, MSC (5.96–5.33 Ma) refers to the large-scale desiccation of the Mediterranean basin leading to reconnection of Mediterranean islands and part of north Africa to mainland Europe (Garcia-Castellanos and Villaseñor 2011). We set the nodes uniting Sulcia violacea (Greece) + S. cretica (Crete) and Paraleptoneta spinimana (Italy) + S. bellesi (Tunisia) to a minimum age of 5.33 Ma because these areas were most recently connected during the MSC. Greece and Turkey were formerly connected in a contiguous landmass called Ägäis. The breakup of Ägäis resulted in the formation of Aegean Islands 12–9 Ma (Dermitzakis and Papanikolaou 1981; Papadopoulou et al. 2010; Lymberakis and Poulakakis 2010). The node uniting Cataleptoneta sengleti (Greece) + C. aesculpii (Turkey) was set to a minimum age of 12 Ma based on their present distributions. We acknowledge that using biogeographic events as calibrations for divergence dating to assess biogeographic history is potentially circular in its logic. However, given the scarcity of fossils for Leptonetidae, we feel that using these recent and localised events, and the benefit they provide for estimating the age of leptonetid lineages across the Holarctic, outweighs the negative aspects.

Biogeographic reconstructions

Biogeographic analysis was performed with the Reconstruct Ancestral State in Phylogenies package (ver. 4.2, see  http://mnh.scu.edu.cn/soft/blog/RASP/; Yu et al. 2015) using the dispersal–extinction–cladogenesis model (DEC) (Ree and Smith 2008) implemented in the C++ version of Lagrange (S. A. Smith, see  http://mnh.scu.edu.cn/soft/blog/RASP/). We use the APE package (Paradis et al. 2004) in R (ver. 4.0, R Foundation for Statistical Computing, Vienna, Austria, see  https://www.R-project.org/) to convert the relaxed RAxML tree topology into a relative-rate scaled ultrametric tree (‘chronopl’ using an assigned lambda value of 0.1). Analysis settings allowed for two unit areas in ancestral distributions and equal probabilities of dispersal events between all areas. Terminal taxa were assigned to six distribution ranges: (A) eastern North America, (B) western North America, (C) South America, (D) Mediterranean Europe, (E) Australia or New Zealand, and (F) East Asia. Four of the assigned regions (eastern North America, western North America, East Asia, Mediterranean Europe) correspond to infraregions identified in meta-analyses of Holarctic biogeography (Sanmartín et al. 2001; Donoghue and Smith 2004).

Data

Voucher data, input DNA values, assembled contig numbers, and UCE locus numbers are provided in  Table S1 (IS20065_AC.pdf). In total, 408 loci were included in the final matrices. The relaxed data matrix included 97 094 basepairs (mean locus length of 238) and 39 114 parsimony informative sites; the strict data matrix included 49 656 basepairs (mean locus length of 122) and 17 638 parsimony informative sites. Raw reads from our 30 original samples have been submitted to the SRA (PRJNA694694); aligned matrices and .TRE files are available at Dryad (doi:10.25338/B8SS5F).

Phylogenomic analyses

Maximum likelihood analysis using RAxML and IQ-TREE for both data matrices resulted in identical topologies with the exception of the RAxML strict matrix (Fig. 4 and S1–S3 of the Supplementary material). All ML analyses recovered a sister-group relationship between Leptonetinae and Austrochiloidea that is highly supported with a bootstrap value of 100%. However, genealogical and site concordance factor analysis for this node resulted in relatively low values of gCF 33% and sCF 40.5% for the relaxed data matrix ( Fig. S2 (IS20065_AC.pdf)) and gCF 21.5% and sCF 40.3% for the strict data matrix ( Fig. S3 (IS20065_AC.pdf)). We present these values as contrasting measures of support, noting that for both data matrices a majority of gene trees (over 60%) support alternative resolutions of this node despite the 100% bootstrap support.

Fig. 4.

Concatenated RAxML results for the relaxed data matrix. Unless otherwise indicated, bootstrap support for all nodes is 100%. Taxa indicated in bold represent the type species for that genus.

SVDQuartets results are presented for both data matrices in  Fig. S4 (IS20065_AC.pdf)– S7 (IS20065_AC.pdf) of the Supplementary material. As suggested by the low gCF and sCF values, the SVD trees and associated 50% majority rule consensus trees show two conflicting resolutions for relationships within Leptonetidae. For both data matrices, SVD optimal trees recover leptonetid monophyly; archoleptonetines are sister to leptonetines and each subfamily is monophyletic. By contrast, the SVD 50% majority rule consensus trees show leptonetid paraphyly identical to our ML results, although support values are low. Despite the ambiguous resolution between the leptonetid subfamilies and Austrochiloidea, all three major lineages (Archoleptonetinae, Leptonetinae, Austrochiloidea) are strongly supported as monophyletic in all analyses. We therefore elevate the subfamily Archoleptonetinae to Archoleptonetidae (new rank) and restrict Leptonetidae to include only members of the subfamily Leptonetinae. This taxonomic structure is used throughout the remainder of the paper.

Relationships among the three species representing Archoleptonetidae are consistent and well supported across all analyses. We recover four main lineages within Leptonetidae which we refer to as the Paraleptoneta, Leptoneta, Protoleptoneta, and Calileptoneta groups. Although relationships among these groups vary by analysis, the Calileptoneta group is consistently sister to the Paraleptoneta + Leptoneta + Protoleptoneta groups. RAxML analyses of the relaxed data matrix results in a sister group relationship between the Paraleptoneta and Leptoneta groups (Fig. 4). However, RAxML analysis of our strict data matrix results in a sister group relationship between the Protoleptoneta and Leptoneta groups ( Fig. S1 (IS20065_AC.pdf)). This node has low bootstrap support in both cases, with 91% support for the relaxed data matrix and 58% for the strict data matrix. Further, gCF and sCF values show that only 6–11% of the gene trees support a relationship between the Paraleptoneta and Leptoneta groups which we interpret as additional evidence of topological uncertainty ( Fig. S2 (IS20065_AC.pdf)– S3 (IS20065_AC.pdf)). SVDQuartets analysis for both data matrices result in a sister group relationship between the Protoleptoneta and Leptoneta groups ( Fig. S4 (IS20065_AC.pdf)– S7 (IS20065_AC.pdf)); however, 50% majority rule consensus trees show lower support for this node.

The European genera Leptoneta Simon, 1872, Leptonetela Kratochvíl, 1978, Paraleptoneta Fage, 1913, Protoleptoneta Deltshev, 1972, and Sulcia Kratochvíl, 1938 are all supported as monophyletic. However, the type species Cataleptoneta edentula Denis, 1955 does not group with C. aesculapii (Brignoli, 1968) or C. sengleti (Brignoli, 1974). The type species Barusia maheni (Kratochvíl & Miller, 1939) is sister to B. insulana (Kratochvíl & Miller, 1939) but B. laconica (Brignoli, 1974) is sister to Cataleptoneta semipinnata Wang & Li, 2010 and does not group with the types of Barusia or Cataleptoneta. Teloleptoneta Ribera, 1988 is weakly supported as sister to Protoleptoneta. One undetermined specimen from Croatia (AR4354) is sister to Appaleptoneta Platnick, 1986 + Leptonetela Kratochvíl, 1978. Leptonetela thracia Gasparo, 2005 is supported as sister to L. jiulong Lin & Li, 2010. All North American genera are monophyletic, including Appaleptoneta Platnick, 1986, Calileptoneta Platnick, 1986, Chisoneta Ledford & Griswold, 2011, Neoleptoneta Brignoli, 1972, Ozarkia Ledford & Griswold, 2011, and Tayshaneta Ledford & Griswold, 2011. Montanineta Ledford & Griswold, 2011 is weakly supported as sister to Protoleptoneta + Teloleptoneta. The Asian genera Falcileptoneta Komatsu, 1970 and Longileptoneta Seo, 2015 are both monophyletic. A single female specimen from Taiwan (CASENT9056030) is sister to Longileptoneta but is a juvenile and unable to be confidently assigned to this genus.

Molecular dating

Results from our dating analysis are presented in Fig. 5. We summarise posterior mean divergence times for major splits below and provide a list of key divergence times in   Table S2 (IS20065_AC.pdf) of the Supplementary material. The age of the root node is estimated as early Jurassic (189 Ma). Both Archoleptonetidae (112 Ma) and Leptonetidae (126 Ma) are of Cretaceous origin and the most recent common ancestor of Austrochiloidea + Leptonetidae dates to the mid-Jurassic (163 Ma). Most of the major divergences within Leptonetidae occur in the middle to late Cretaceous. The Paraleptoneta, Leptoneta, Protoleptoneta, and Calileptoneta groups arose 87–110 Ma and the majority of more recent divergence occurred in the mid-Paleogene.

Fig. 5.

MCMC tree chronogram generated from the relaxed data matrix. Nodes A–D represent calibration points used in our analysis: A, Palaeoleptoneta calcar from Burmese Amber (98.19 Ma); B, Breakup of Ägäis (12 Ma); C, Messinian Salinity Crisis (5.33 Ma); D, Fernández et al. (2018).

Biogeography

Results for our biogeographic analysis are presented in Fig. 6. Multiple dispersal and vicariance events are inferred and, with few exceptions, their probabilities are 1.0 (  Table S3 (IS20065_AC.pdf) of the Supplementary material). No extinction events are reconstructed on the phylogeny. Archoleptonetidae is reconstructed to have an ancestral distribution in western North America. Leptonetidae is predicted to have originated in a region spanning western North America and Mediterranean Europe. The ancestor of the Paraleptoneta, Leptoneta, and Protoleptoneta groups most likely originated in Mediterranean Europe with a low probability of dispersal (0.52) to eastern North America. The Paraleptoneta and Leptoneta groups both have reconstructed origins in Mediterranean Europe. The Protoleptoneta group most likely had an ancestral distribution spanning eastern North America and Mediterranean Europe with a low probability of dispersal (0.54) to east Asia. The Calileptoneta group is reconstructed to have originated in a region spanning western North America and Mediterranean Europe.

Fig. 6.

Inferred ancestral range reconstruction based on the DEC model, based on the relaxed RAxML topology. D, dispersal; V, vicariance. Numbers at nodes are used for reference only and do not indicate measures of statistical support (see   Table S3 (IS20065_AC.pdf)).

Systematics

Recent studies of spider phylogeny have recovered conflicting results for the relationships of Archoleptonetidae and Leptonetidae, ranging from hypotheses of monophyly (Fernández et al. 2018), to paraphyly (Ramírez et al. 2021), and polyphyly (Wheeler et al. 2017). Given the conflict observed between these studies (Fig. 2), and the results of our analyses, it is not unreasonable to question the elevation of Archoleptonetidae. Further, what are the possible causes for the discordance between these results and how does our study address their limitations?

Among the primary strengths of our study is taxon sampling. We include specimens collected over the past 16 years, many species of which are monotypic, rare, or live in difficult to access habitats such as caves. Admittedly, our sampling is limited in Asia, but efforts to access additional material from this region have been challenging. By contrast, most recent studies have relatively sparse sampling of archoleptonetids and leptonetids. Fernández et al. (2018) includes two exemplars (Archoleptoneta and Calileptoneta) and provides insight into their relationships to other spiders, but lacks sufficient depth to test the monophyly of Archoleptonetidae and Leptonetidae. Wheeler et al. (2017) has broader representation with Archoleptoneta, Calileptoneta, Leptoneta, and Neoleptoneta but does not include the ecribellate Darkoneta in order to test the monophyly of Archoleptonetidae. Ramírez et al. (2021) includes denser sampling within Archoleptonetidae and Leptonetidae but is mostly focused on the evolution of tracheael systems in basal araneomorphs. Our study expands on Ramírez et al. (2021) by adding representatives of most leptonetid genera, thereby providing a robust test of the monophyly of Archoleptonetidae and Leptonetidae while gaining insight into relationships among the genera.

In addition to differences in sampling, each of these studies used a different type of data. Wheeler et al. (2017) used nucleotide data from six genes routinely used in spider systematics. Their results are perplexing as Leptonetidae is polyphyletic with Leptoneta sister to Austrochilidae and Calileptoneta + Neoleptoneta sister to Entelegynae (Fig. 3). Analytical instability was recognised as a problem by the authors, who also argued that a polyphyletic Leptonetidae was unlikely to be based on morphological synapomorphies. Given the problems of the Wheeler et al. (2017) analysis, the question becomes focused on whether or not Archoleptonetidae and Leptonetidae are sister groups.

Fernández et al. (2018) expanded on the results of Garrison et al. (2016), which used transcriptomes to infer relationships across the spider tree of life. Given that both transcriptomes and the UCEs in our study are exonic (following results of Hedin et al. 2019), the core data should be comparable and we expected similar results. As part of their study, Fernández et al. (2018) developed several data matrices, most of which recovered a monophyletic Leptonetidae (Archoleptonetinae + Leptonetinae). However, their ML analysis of a truncated, strict orthology matrix resulted in a paraphyletic Leptonetidae with Calileptoneta sister to austrochiloids. Our study also shows conflicting results for the resolution of this node with ML analyses supporting leptonetid paraphyly (Austrochiloidea + Leptonetidae) and some SVDQuartets results showing leptonetid monophyly (Archoleptonetidae + Leptonetidae), although weakly supported. Similar to Fernández et al. (2018), our strict matrix was partly intended as a sensitivity test by restricting our data to a core set of orthologs and removing flanking introns where alignment may be uncertain. Even with nearly 50% of the data removed, the strict matrix still does not consistently resolve the node as some SVDQuartets results recover leptonetid monophyly ( Fig. S4 (IS20065_AC.pdf),  S6 (IS20065_AC.pdf)). Interestingly, despite the fact that our ML analyses show 100% bootstrap support for a sister group relationship between Austrochiloidea and Leptonetidae (Fig. 4, S1–S3), the relatively low gCF and sCF values suggest that most of the gene trees recover alternate resolutions for the node. In summary, our results and those of Fernández et al. (2018) show ambiguity in the resolution of this node as reflected in both analytical sensitivity and gCF and sCF values.

Although we are encouraged by the results in our study, we recognise that conflict in the relationships of Austrochiloidea, Archoleptonetidae and Leptonetidae persists. As a consequence, we argue that the node is best viewed as a trichotomy in need of further study. One key outcome of all recent studies is that Archoleptonetidae and Leptonetidae are not part of Synspermiata, a result predicted by morphology (Brignoli 1979d; Ledford and Griswold 2010) and corroborated by all phylogenomic analyses. Regardless of the eventual placement of Archoleptonetidae and Leptonetidae, the elevation of Archoleptonetidae is supported by their morphology (Ledford and Griswold 2010), facilitates their diagnosis, and serves a grouping function. We therefore view the elevation of Archoleptonetidae as both warranted and timely, especially as our understanding of spider phylogeny improves with increased adoption of phylogenomic methods. As newly defined, Archoleptonetidae is a small family consisting of two genera; however, many areas within their distributional range are undersampled and we expect that more species likely await discovery.

In each of our analyses, we recover four main lineages within Leptonetidae which we refer to as the Paraleptoneta, Leptoneta, Protoleptoneta, and Calileptoneta groups. The Paraleptoneta and Leptoneta groups include species limited to Mediterranean Europe, whereas both the Protoleptoneta and Calileptoneta groups include species distributed on multiple continents. The North American fauna is well represented in our study, and all genera have high statistical support corroborating the hypotheses of Ledford et al. (2011, 2012). Chisoneta, Neoleptoneta, Ozarkia, and Tayshaneta comprise a single lineage that is distributed from the south-western USA to Mexico and is sister to the remainder of the Protoleptoneta group. Representation of the Asian fauna is weaker, but Falcileptoneta and Longileptoneta are monophyletic and well supported as part of the Protoleptoneta group. Although our sample includes only a single Leptonetela species from China, L. jiulong Lin & Li, 2010 is sister to L. thracia Gasparo, 2005, supporting the synonymies of the genera Guineta, Qianleptoneta, and Sinoneta (Lin & Li, 2010). With the exception of Cataleptoneta, the European genera are supported as monophyletic. However, as discussed below, several species are likely misplaced. Although our study does not include Asian representatives of Leptoneta, several Leptoneta species are part of our analyses, including the type L. convexa Simon, 1872. Based on the features of their genitalic morphology (J. Ledford, pers. obs.) and the distribution of Leptoneta in western Mediterranean Europe, we predict that all of the currently described Asian Leptoneta are misplaced.

As the first study to assess leptonetid phylogeny using a sample of most genera, it is not surprising to discover taxonomic problems and learn that some genera need revision. In particular, the generic limits among Barusia, Cataleptoneta, and Paraleptoneta are unclear. The type species Cataleptoneta edentula (Denis 1955) does not group with C. semipinnata (Wang and Li 2010) and C. sengleti (Brignoli 1974a). Barusia has similar problems as B. laconica (Brignoli 1974a) does not group with the type species B. maheni. Taxonomic issues within Barusia and Cataleptoneta date to Kratochvíl (1978), who based his diagnoses on spination differences on the male palpal tibia and not details of palpal bulb morphology as had been done by previous workers. Barusia, Cataleptoneta, Paraleptoneta, and Sulcia are all characterised by complex spination patterns on the male palpal tibia (Le Peru 2011), making the morphological distinction between them unclear. Although recent studies on Cataleptoneta (Wang and Li 2010; Deltshev and Li 2013; Gavish-Regev et al. 2016; Demircan 2020) have contributed to our understanding of its diversity, they have not provided a broad perspective on the diagnosis, distribution, or relationships of the genus as a whole. Work in progress (C. Ribera, unpubl. data) further shows that the genetic diversity within Paraleptoneta is higher than expected despite a history of synonymy (Brignoli 1979c). Based on our results, we hypothesise that at least two additional genera remain undescribed; the first includes B. laconica and C. semipinnata and the second includes C. sengleti and C. aesculapii. However, we believe that taxonomic changes to Barusia and Cataleptoneta need the context of a more comprehensive revision and we defer any changes until the lineage can be more carefully studied.

Within Leptonetidae, there are three described monotypic genera: Montanineta Ledford & Griswold, 2010 from the eastern USA, Teloleptoneta Ribera, 1988 from Portugal, and Rhyssoleptoneta Tong & Li, 2007 from China. Until this study, none of these taxa have been included in a phylogenetic analysis and the justification used in their descriptions was based solely on the degree of difference in their genitalic morphology. Although our study does not include Rhyssoleptoneta, both Montanineta and Teloleptoneta group with Protoleptoneta (Deltshev 1972) although support within this clade is low. Unpublished single-gene analyses also show an ambiguous relationship between Protoleptoneta and Teloleptoneta (C. Ribera, unpubl. data). Close inspection of the genitalic morphology of Montanineta, Protoleptoneta, and Teloleptoneta show some similarities (J. Ledford, pers. obs.) and one solution is that all of these genera could be combined into a single genus. Although this would simplify the nomenclature of the group, it would not change our interpretation of the relationships in the lineage as a whole. Namely, given the wide geographic disjunction between these genera we view them as relictual; the closest relative of Montanineta, for example, lives on a different continent. One specimen from Croatia (AR4354) likely represents yet another monotypic genus given the geographic disjunction with its closest relatives (Appaleptoneta and Leptonetela). Lastly, although Rhyssoleptoneta was not included in our analysis examination of its genitalic morphology shows similarity in palpal structures to Appaleptoneta (J. Ledford, pers. obs.) and we predict that the genus may be part of the Calileptoneta group.

Biogeography

Given that our sampling of Archoleptonetidae is limited to representatives from California and Arizona, it is not surprising that the ancestral distribution for the group is reconstructed as western North America (B). As more species are described, we expect that the biogeography of Archoleptonetidae will become better understood.

Results of our molecular dating analysis estimate the origin of Leptonetidae at 126 Ma and its ancestral distribution is inferred to encompass an area spanning western North America and Mediterranean Europe (BD, node 107, Fig. 6). Although the Paraleptoneta and Leptoneta groups are reconstructed to have ancestral distributions restricted to Mediterranean Europe (D, node 80, Fig. 6), the Protoleptoneta and Calileptoneta groups are relatively older and have ancestral distributions that include combinations of areas including eastern and western North America, Mediterranean Europe, and east Asia (nodes 97 and 106, Fig. 6).

During the mid-Jurassic, Pangaea split into the northern and southern supercontinents Laurasia and Gondwana. The origin of Leptonetidae (mid-Cretaceous) coincides with a time when Laurasia was largely connected, although North America and Asia remained separated. Most of the major divergences within Leptonetidae occurred 100–70 Ma. Within this time, Laurasia split into two palaeocontinents: Euramerica (Europe + eastern North America) and Asiamerica (western North America + Asia). North America was divided by the Mid-Continental Seaway and western North America was connected to east Asia through Beringia. Europe was divided from Asia by the Turgai Strait and eastern North America remained connected to Europe through multiple, periodic land bridges across the Atlantic (reviewed in Sanmartín et al. 2001).

During the early Tertiary a continuous region of vegetation, the Boreotropical forest, was distributed across Laurasia. The boreotropics hypothesis (Wolfe 1975; Tiffney 1985) postulates that the biogeographical disjunctions often seen in plants between eastern North America–Europe and western North America–Asia resulted from the fragmentation of Laurasia and the recession of the Boreotropical forest. Based on the age estimates, ancestral distributions, and the biology of leptonetids, we hypothesise that they were once widespread across Laurasia. One possibility is that they were associated with the Boreotropical forest ecosystem and their phylogeny is largely the product of vicariance as Laurasia fragmented. This scenario not only aligns well with our results, but also explains the repeated pattern of narrow endemism for most leptonetid species worldwide. As Laurasia drifted apart, the climate became drier and leptonetids were restricted to areas where mesic conditions similar to those found in the Boreotropical forest persisted. This may also explain the close association of leptonetids with caves, which, due to their cool and moist environments, may have functioned as refugia. As such, we argue that leptonetids may be regarded as a largely relictual fauna, similar to Holarctic relict floras (Tiffney 1985; Lavin and Luckow 1993; Milne and Abbott 2002).

As a relictual fauna, we predict that extinction has played a significant role in shaping the present distribution of leptonetids. Although our DEC analyses recover no extinction, we view this result as improbable given the age and reconstructed ancestral distributions for most groups in our analyses (Fig. 5, 6). The underestimation of local extinction is also a known issue with DEC analysis (Ree and Smith 2008). Although dispersal is inferred in our DEC results, the probabilities associated with these events are low (  Table S3 (IS20065_AC.pdf)). We also find the pattern of sympatry within Leptonetidae of particular interest; in most cases leptonetid genera do not have overlapping distributions, even among distantly related lineages. In areas where the distributions of genera are close, such as eastern Europe, our phylogeny reveals taxonomic problems. There are few cases of sympatry within genera although sampling error may be a contributing factor given the relative rarity of leptonetids in collections. We interpret these patterns as evidence for niche conservatism in the group through deep geologic time, a pattern predicted by their biology and supported by phylogeny.

Paraleptoneta and Leptoneta groups

The origin of the Paraleptoneta and Leptoneta groups dates to 105 Ma and coincides with the separation of Europe and Asia by the Turgai Strait. The Paraleptoneta group includes Barusia, Cataleptoneta, Paraleptoneta, and Sulcia, which are primarily distributed in the eastern Mediterranean but also have records in the Levant and Tunisia (Ribera and Lopez 1982; Gavish-Regev et al. 2016). At the time of the origin of this group, the eastern Mediterranean formed a continuous plate from Turkey to the Balkans, including the Aegean and Croatian Islands. The present geographical distribution of the Paraleptoneta group includes all of southern Europe (except for the Iberian Peninsula), where Pleistocene glaciations had minimal impact (Clark et al. 2009; Batchelor et al. 2019). It is likely that the group extends further north, although currently there are no known species at higher latitudes. Our analysis suggests an eastern origin (Turkey or Greece) with subsequent colonisation to the Balkan Mountains, from Albania to Croatia and Italy, extending southward to Tunisia during the Messinian Salinity Crisis (Garcia-Castellanos and Villaseñor 2011).

The Leptoneta group includes 30 species distributed throughout the western Mediterranean and 38 species in Asia (World Spider Catalog, see  http://wsc.nmbe.ch). Based on the species included in our analysis, and the distributions of all other Leptoneta from Europe, we interpret the genus to be of Iberian origin; all Leptoneta species in Europe occur in regions that were formerly part of or connected to the peninsula. Results from our molecular dating analysis estimate the divergence of Leptoneta at 33 Ma. During this time the Iberian Peninsula was located between Eurasia and North America but separated from the northern mainland of Africa. This timing also precedes the drying of the Turgai Strait and subsequent uplift of the Himalayas. A key consequence of this timing is that Leptoneta is restricted to the western Mediterranean. In conjunction with our observations of its morphology, we view this as evidence that all currently described Asian Leptoneta are misplaced and belong to other genera.

Protoleptoneta group

The Protoleptoneta group is estimated to have diverged 87.5 Ma and includes genera from eastern North America, Mediterranean Europe, and east Asia. The ancestral distribution of the lineage has a mixed probability of origin, but most likely includes an area spanning eastern North America and Mediterranean Europe (AD, node 97, Fig. 6). Two main lineages occur within the group; the first includes four genera (Chisoneta, Neoleptoneta, Ozarkia, and Tayshaneta) distributed in the southern USA and Mexico. This group is estimated to have arisen 77 Ma in eastern North America after the closure of the Mid-Continental Seaway and the subsequent uplift of the western mountain ranges as part of the Laramide orogeny. The westernmost species in this lineage, Ozarkia apachaea (Gertsch, 1974) is known from the Chiricahua Mountains and we predict that the western mountains presented a barrier to further dispersal of the group in western North America.

The second Protoleptoneta group lineage includes five widely disjunct genera: Falcileptoneta (east Asia), Longileptoneta (east Asia), Montanineta (eastern North America), Protoleptoneta (Mediterranean Europe), and Teloleptoneta (Mediterranean Europe). Given the geographic complexity, the DEC analysis infers a mixed probability of origin for this group (node 96, Fig. 6). Two possibilities are presented: (1) Mediterranean Europe + east Asia (DF, 70.9%), and (2) eastern North America + east Asia (AF, 29%). The first scenario requires dispersal to eastern North America, perhaps by the Thulean or De Greer land bridges, both of which are recognised as important routes between Mediterranean Europe and eastern North America (Sanmartín et al. 2001; Brikiatis 2014). The second scenario requires dispersal to Mediterranean Europe which seems less likely given the persistence of the Turgai Strait until the late Tertiary (30 Ma). One possible explanation for the ambiguity is insufficient sampling in Asia. Falcileptoneta and Longileptoneta, for example, are broadly distributed in Japan, Korea, and Taiwan and together include over 50 species. Our data includes only four representatives of these genera and we predict that increased sampling will greatly inform the inferred ancestral distribution of the group. Based on the number of described species, the centre of diversity for the Protoleptoneta group appears to be east Asia. The genera Montanineta and Teloleptoneta are monotypic and Protoleptoneta includes only four species, mostly known from isolated caves. Given the age of the Protoleptoneta group and its inferred ancestral distribution, we hypothesise that Montanineta, Teloleptoneta, and Protoleptoneta are relictual and perhaps the last extant members of formerly widespread lineages.

Calileptoneta group

The origin of the Calileptoneta group is estimated at 110 Ma and is reconstructed as having an ancestral distribution spanning western North America and Mediterranean Europe (BD, node 106, Fig. 6). This timing is close to the origin of the mid-Continental Seaway, which separated North America into two subcontinents (Laramidia in the west and Appalachia in the east). Given the age of the group, a plausible scenario is that the group was formerly widespread and then divided by the formation of the mid-Continental Seaway.

Within the Calileptoneta group, one lineage corresponds to Calileptoneta, which is known only from western North America (B, node 101, Fig. 6). Although we see no relationships between Calileptoneta and taxa from east Asia, the western North America–east Asia disjunction is well established and we predict that some unsampled taxa in Asia will have close affinity with Calileptoneta. The monotypic genus Rhyssoleptoneta, for example, shares palpal homologies with Calileptoneta (J. Ledford, pers. obs.) and may prove important to understanding relationships and biogeography within this group.

The second lineage includes taxa from Mediterranean Europe, eastern North America, and east Asia. The group is reconstructed to have originated in Mediterranean Europe (D, node 105, Fig. 6) with an estimated age of 103 Ma. Given the reasonably ancient age of the group, we predict that it was formerly widespread and some lineages (AR 4354 and Appaleptoneta) are relictual. Within this group, the genus Leptonetela is the most diverse with over 100 species described from Asia (China and Vietnam) and 12 from the eastern Mediterranean (Turkey, Caucasus, and Greece). Although we only have two Leptonetela species in our dataset, both Mediterranean Europe and east Asia are represented. The estimated age for the split of L. jiulong (Guizhou, China) and L. thracia (Greece) is 19 Ma (the origin of the genus Leptonetela is older, c. 80 Ma) after the closure of the Turgai Strait (Tangelder 1988; Sanmartín et al. 2001). We interpret the occurrence of Leptonetela in Europe as a recent dispersal from South Asia, that occurred before the rise of the Tibetan plateau or the formation, further north, of the Gobi Desert, which prevented connection between Central Asia and the Eastern Mediterranean. Increased sampling will likely improve our understanding of the biogeography of Leptonetela, especially in the region from Turkey to India, which remains undersampled and may hold more species.

Justification

We have decided to raise both subfamilies Archoleptonetinae and Leptonetinae to family status on the grounds that to do so is practical, making each easier to diagnose and that it does no phylogenetic harm, i.e. each is monophyletic. Further, if our preferred tree is correct (Fig. 4), with Leptonetinae sister to the Austrochiloidea to the exclusion of the Archoleptonetinae, Leptonetidae sensu lato becomes paraphyletic and relimiting the family becomes essential.

Austrochiloidea, Archoleptonetidae and Leptonetidae

This analysis and other recent analyses (e.g. Fernández et al. 2018) suggest that Archoleptonetidae and Leptonetidae are closely related to Austrochiloidea (Austrochilidae, including Hickmania, and Gradungulidae). Whereas molecular data unite these taxa, they do not share any obvious morphological synapomorphies. Austrochiloids are readily distinguished from Archoleptonetidae and Leptonetidae. The austrochiloid families share the morphological synapomorphies of a clypeal hood, i.e. a median projection over the middle cheliceral base (Griswold et al. 2005, character 30), a distal notch in the orifice of the trichobothrial base (Griswold et al. 2005, character 9), and a peculiar genitalic morphology in which the gonopore is visible anterior to the epigastric fold (Griswold et al. 2005, character 135; Ramírez 2014 character 363). Austrochiloids are large spiders, and all have austral distributions. Members of Kaiya, smallest of the gradungulids, still have a body length of 1.2 cm and a leg span of more than 3 cm, whereas some austrochiloids are veritable giants. Hickmania troglodytes (Austrochlidae) may have a body length of 2 cm and a leg span of 17 cm, and Macrogradungula moonya (Gradungulidae) may have a body length of 2 cm and a leg span of more than 18 cm! By contrast, temperate Archoleptonetidae and Leptonetidae are small spiders (less than 5 mm body length) with long, slender legs, with a leg spread of less than 2.5 cm. All have only six eyes (or none), three claws with the STC (superior tarsal claws) simple (unlike the asymmetrical claws of Gradungulidae, raptorial claws of Trogloraptoridae or enlarged, bipectinate STC of many Dysderoidea), no cheliceral lamina, unusual iridescence, especially on the legs and carapace (shared with Ochyroceratidae and Psilodercidae, among others), autospasy at the patella–tibia joint (shared with Filistatidae, Austrochilinae (Thaida and Austrochilus) and linyphioids), patella–tibia gland plates (shared with the Telemidae; similar structures occur in various Entelegynae), no external epigynum (unlike Entelegyne and some Palpimanoidea and Synspermiata), a single posterior spiracle, a divided cribellum (unlike the entire cribellum of Hypochilidae and cribellate Austrochiloidea) or a small colulus (unlike the large coluli of Telemidae, Ochyroceratidae and Psilodercidae and some Synspermiata), clypeus with margin evenly concave (unlike the median hood of Austrochiloidea), and no peg teeth (unlike Palpimanoidea).

Ledford and Griswold (2010, p. 4) diagnosed Leptonetidae sensu latu based on a set of homoplastic putative synapomorphies, which assumed placement of Leptonetidae in the Haplogynae, an hypothesis and grouping now discarded. Features cited were unusual iridescence, especially on the legs and carapace, patellar–tibial gland plates, autospasy at the patella–tibia joint, the presence of tartipores on the ALS (shared widely but absent in Synspermiata), male palpi with a fused tegulum and subtegulum but with an expandable basal haematodocha (a peculiar morphology, which requires further study in austrochiloids), and a respiratory system consisting of a pair of short median branches with long laterals that open to a single spiracle anteriad of the ALS (also found in many Entelegynae). This diagnosis overlooks notable differences between Archoleptonetinae and Leptonetinae in eye position, spinning organs and male and female genitalia: recognising Archoleptonetidae and Leptonetidae as families will make each far easier to diagnose.

Family ARCHOLEPTONETIDAE Gertsch, 1974 (new rank)

Archoleptonetinae Gertsch, 1974: 198

Family LEPTONETIDAE Simon, 1890 (new status)

Leptonetidae Simon, 1890: 80