We present the first genome-wide molecular phylogeny of jumping spiders (Araneae: Salticidae), inferred from Anchored Hybrid Enrichment (AHE) sequence data. From 12 outgroups plus 34 salticid taxa representing all but one subfamily and most major groups recognized in previous work, we obtained 447 loci totalling 96,946 aligned nucleotide sites. Our analyses using concatenated likelihood, parsimony, and coalescent methods (ASTRAL and SVDQuartets) strongly confirm most previous results, resolving as monophyletic the Spartaeinae, Salticinae (with the hisponines sister), Salticoida, Amycoida, Saltafresia, and Simonida. The agoriines, previously difficult to place beyond subfamily, are finally placed confidently within the saltafresians as relatives of the chrysillines and hasariines. Relationships among the baviines, astioids, marpissoids, and saltafresians remain uncertain, though our analyses tentatively conclude the first three form a clade together. Deep relationships, among the seven subfamilies, appear to be largely resolved, with spartaeines, lyssomanines, and asemoneines forming a clade. In most analyses, Onomastus (representing the onomastines) is strongly supported as sister to the hisponines plus salticines. Overall, the much-improved resolution of many deep relationships despite a relatively sparse taxon sample suggests AHE is a promising technique for salticid phylogenetics.

Dionycha, jumping spiders, salticids, systematics, phylogenomics

Understanding the relationships of jumping spiders (Salticidae) long posed a challenge, given their diversity in forms and species (about 6,000 described, World Spider Catalog 2017). Recent data from a handful of sequenced genes has, however, begun to resolve many aspects of the group’s broad phylogenetic structure (Maddison and Hedin 2003, Bodner and Maddison 2012, Maddison et al. 2014). Combined with morphological information, these results have led to a comprehensive phylogenetic classification (Maddison 2015) and are beginning to enable inferences about evolutionary patterns in salticids’ structures, ecology, and behaviour. Two major gaps in knowledge remain to be filled, however, before the phylogeny can provide a high-resolution lens on salticid evolution. First, the great majority of known species are unstudied phylogenetically (and many others undiscovered taxonomically), and therefore few details are available about shallower phylogeny in most tribes and genera of the family. Second, the few genes studied do not give definitive answers in several key areas of the deeper parts of the phylogeny. Maddison et al. (2014) were unable to resolve the relationships among the seven subfamilies (as defined by Maddison 2015), except for the sister group relationship between Hisponinae and Salticinae. They were also unable to place the peculiar agoriines, and to determine the relationships among the baviines, Marpissoida, Astioida, and Saltafresia; support for the Saltafresia and Simonida was only tentative.

Our goal here is to answer remaining questions about broad salticid relationships, using data from across the genome. An efficient method to obtain data on hundreds of genes is Anchored Hybrid Enrichment (AHE; Lemmon et al. 2012; Lemmon and Lemmon 2013), a high-throughput genomics technique that uses probes designed for highly conserved DNA regions flanked by less-conserved regions. AHE has been applied for both deep and shallow relationships in spiders (Hamilton et al. 2016a,b), where it shows considerable promise for resolving phylogeny based on genome-wide data. We here apply AHE to salticids, using a combination of Spider Probe Kit versions 1 and 2 designed for spiders by Hamilton et al. (2016b, unpublished). The AHE Spider Probe Kit targets 585 phylogenetically-informative loci across the Order Araneae and delivers phylogenetic utility at both deep and shallow taxonomic depths. By providing a set of molecular markers that can be used to address evolutionary questions at multiple hierarchical levels, as well as across different research groups, the AHE Spider Probe Kit is being used to answer larger questions about spider phylogeny and evolution (Hamilton et al. 2016a,b).

Hybrid enrichment results are shown in the Supplementary Table. The 447 loci obtained in the final filtered data set represent for most taxa about 80 kb of nucleotide sequence. We were less successful at obtaining data for two taxa, with Schizocosa saltatrix having only 9377 nucleotides sequenced, and Yllenus arenarius having 36069 nucleotides. The “on target” percentage of Yllenus was low, suggesting either that its genome is unusually large, or that the sample included also some non-spider DNA. The other taxa had between 76,262 (Clubiona) and 91,238 (Hasarius adansoni) nucleotides sequenced. Alignments for each of the 477 loci are deposited, along with phylogenetic results, to Dryad (http://dx.doi.org/10.5061/dryad.n2b3h).

Fig. 1 shows the ML tree from the partitioned concatenated supermatrix. Bootstrap values are high for most clades. The unpartitioned ML, parsimony, ASTRAL and SVDQuartets gave largely concordant results, differing only where marked in Fig. 1 by -u, -p, -a, and -s respectively. In particular, unpartitioned ML places Yllenus as the sister to the rest of the Simonida (though with low bootstrap support); parsimony places Yllenus and Naphrys as sisters, and Freya as sister to Harmochirus and Habronattus; ASTRAL places Bavia as sister to the astioids and marpissoids, and Yllenus as the sister to the rest of the Simonida; SVDQuartets trades the positions of Idastrandia and Hasarius and rearranges the Simonida.

Figure 1. 

Maximum likelihood phylogeny from the partitioned concatenated matrix of 447 loci captured by Anchored Hybrid Enrichment. Numbers indicate percentage of likelihood bootstrap replicates showing the clade. Half circle indicates clades supported also in the results of Maddison et al. (2014) or, for the Amycoida, of Ruiz and Maddison (2015). Letters u, p, a, and s indicate clades that fail to appear in the analyses by unpartitioned likelihood, parsimony, ASTRAL and SVDQuartets respectively.

This first genome-wide analysis of salticids resolves the group’s phylogeny with greater confidence than previous studies, confirming and extending those results based on far fewer genes (Maddison et al. 2014; Ruiz and Maddison 2015; Maddison 2015). The results corroborate the monophyly of the Salticinae, a major clade with more than 90% of described salticid species, including most familiar species. The Spartaeinae, which includes the well-known Portia, is also supported (in our analysis: Mintonia, Cocalodes, Lapsias). Major clades corroborated within the Salticinae are the Salticoida (sensuMaddison 2015), Saltafresia, Simonida, Amycoida, and Marpissoida (here: Leikung, Tisaniba, Phidippus, Sassacus). Other clades consistent with the previous results of Maddison et al. (2014, 8 genes, salticid-wide) and Ruiz and Maddison (2015, 5 genes, within the Amycoida) are indicated with semicircles on Fig. 1.

The relationships among the subfamilies, previously poorly resolved (Maddison et al. 2014), are strongly supported in our analyses. Unsurprising is the relationship between the Hisponinae and Salticinae, which has been supported by both molecular and morphological data (Maddison 2015). The relationship among asemoneines, lyssomanines and spartaeines was anticipated (Maddison et al. 2014) but not previously well supported.

A novel result is the placement of Onomastinae as sister to Hisponinae plus Salticinae. Onomastines, like the lyssomanines and asemoneines, are long-legged translucent spiders with complex palpi and an ocular area relatively small compared to other salticids (see Wanless 1980). The distinctive features of onomastines, lyssomanines and asemoneines might have been interpreted as ancestral for the family, or as synapomorphies uniting them (Maddison 2015). Their separate placement here suggests that either their form is convergent, or that the more familiar compact brown body with an expanded ocular area evolved independently in spartaeines and hisponines+salticines. We do note, however, that despite the 100% ML bootstrap support for onomastines+hisponines+salticines, not all analyses agree on this placement. The SVDQuartets analysis places Onomastus as sister to Asemonea+Lyssomanes+Spartaeinae, as also recovered from 8 genes by Maddison et al. (2014).

Within the Salticinae, our data have succeeded in resolving the placement of one puzzling group, the agoriines, whose position was problematic to Maddison et al. (2014). Our 447 locus data clearly supports placing the agoriines within the Saltafresia, in a group with chrysillines (here represented by Heliophanus) and hasariines. Most analyses place Agorius sister to Heliophanus, though ASTRAL places it with the nearby Hasarius. Maddison et al. (2014) found Agorius and its close relative Synagelides to have unstable placement, on long branches, and varying in position drastically among the different analyses. Interestingly, their All Genes salticine analysis (their figure 18) placed agoriines with the chrysillines, a placement strongly supported in our analyses. Maddison (2015) notes the similarities of the genitalia of agoriines with the two groups indicated as close relatives here, the chrysillines and hasariines.

The relationships among the four major subgroups of Salticoida (sensuMaddison 2015) — Marpissoida, Astioida, Baviini, and Saltafresia — were not resolved well by Maddison et al. (2014: 80). Bodner and Maddison (2012) suggested the first three form a clade, but this was not corroborated by the results of Maddison et al. (2014). Our data give support to Bodner and Maddison’s conclusion, though weakly. All analyses place Bavia in a clade with the Marpissoida and Astioida (together forming the sister group to the Saltafresia), but bootstrap support is only 58% for likelihood, 67% for parsimony, and 100% for SVDQuartets. The weak support for this clade may indicate a rapid early radiation of the Salticoida, and may require considerably more data to corroborate or refute. Within the tentative clade of Baviini+Marpissoida+Astioida the detailed relationships are unresolved. Likelihood and SVDQuartets place Bavia with the astioids Myrmarachne and Orthrus but with bootstrap support less than 50% for ML, 56% for SVDQuartets; parsimony places Bavia as sister to the Marpissoida; ASTRAL places Bavia as sister to Marpissoida+Astioida.

Within the Simonida, the Harmochirina (Harmochirus, Habronattus) and Salticini (Carrhotus, Salticus) are confirmed each as monophyletic and as sister lineages, as per Maddison et al. (2014). Deeper relationships in the Simonida, among the tribes, are unclear and vary by analysis. As shown in Fig. 1, likelihood recovers (Naphrys, (Yllenus, (Freya, (harmochirines, salticines)))), with Naphrys representing the Euophryini, Yllenus the Leptorchestini, and Freya the Aelurillini. However, ASTRAL obtains (Y,(N,(F,(h,s)))), SVDQuartets (Y,(h,(N,(F,s)))), and parsimony ((Y,N),((F,h),s)). A contributing factor to this poor resolution could be the poor sequence capture for Yllenus.

Given the strength of this broad data set and its concordance with previous results, we can now be reasonably confident in our current phylogenetic classification (Maddison 2015). Our results highlight what is needed for further progress. For the deeper parts of the phylogeny, most urgent is to include the Eupoinae, not only to determine their (currently ambiguous) placement (Maddison et al. 2014), but also because their inclusion would provide a test of the supported relationships among the subfamilies. Within the Salticinae, the most basic outstanding question concerns the relative relationships among baviines, astioids, marpissoids and saltafresians. To resolve this, a much larger fraction of the genome may be needed. Of course, even once our understanding of these broad relationships stabilizes, the bulk of salticid phylogeny remains still unresolved, as not only is there no explicit phylogenetic work on most of the described species, but many species remain to be discovered.

For assistance in collecting specimens we thank especially Edyta Piascik, Mauricio Vega, and Junxia Zhang. This work was supported by an NSERC Discovery grant to W. Maddison (RGPIN 261352-2013); National Science Foundation Doctoral Dissertation Improvement Grant – DEB-1311494 to CAH; an Auburn University Cellular and Molecular Biosciences Peaks of Excellence Research Fellowship (CAH).