Ninetinae are small to tiny ground-dwelling spiders that are largely restricted to arid environments (Huber and Brescovit 2003; BA Huber, unpublished data). With only 31 described extant species, the subfamily is by far the smallest of the five currently recognized subfamilies in Pholcidae. Ninetinae seem to be diverse in the New World (ten named genera + about four unnamed genera; BA Huber, unpublished data) where they represent the most southern (Argentina) and most northern (Canada) autochthonous pholcid records worldwide. Only two genera (Ninetis Simon, 1890 and one unnamed; BA Huber, unpublished data) are known from the Old World.

Their short legs make them superficially strikingly different from ‘typical’ long-legged pholcids. This distinctness was recognized as early as 1893, when Eugène Simon classified the only ninetine species available to him in a separate subfamily “Ninetidinae”, as opposed to all other pholcids classified in Pholcinae (Simon 1893). Subsequent morphological and molecular phylogenies have partly supported this view (Huber 2000, Dimitrov et al. 2013) but never convincingly with strong support.

Our present analyses include 15 species representing eight of the eleven described genera, originating from both the New World and the Old World (Figure 2). A sister-group relationship between Ninetinae and all other pholcids is not supported by our analyses. Instead, all four analyses put Ninetinae as sister to Artema Walckenaer, 1837, and this clade is in turn sister to all other pholcids. For reasons discussed below (under Arteminae), we consider this relationship between Artema and Ninetinae dubious. The conclusion here is that Ninetinae are ‘basal’, either with Artema or without, but in any case the external relationships of Ninetinae remain unsatisfactorily resolved and need further study.

Figure 2. 

Ninetinae and Arteminaea Artema sp. n. “Om14” (Oman) b Gen. n. (Ninetinae) sp. n. “Om6” (Oman) c Chisosa diluta (USA) d Gen. n. (Arteminae) sp. n. “Ind82” (Sulawesi).

The monophyly of the subfamily receives high to full support in all analyses but the composition is slightly different from previous concepts: the North American Chisosa Huber, 2000, originally thought to be a representative of Ninetinae (Huber 2000), is moved to Arteminae. This move is also supported by male genitalic characters (massive palpal femur; procursus with dorsal apophysis and ventral pocket) and by somatic characters (exposed tarsal organ; reduction of epiandrous spigots; Huber 2000). Another genus that was previously (Huber and El Hennawy 2007, Huber 2011b) thought to be a member of Ninetinae is Nita Huber & El Hennawy, 2007. As already suggested in a previous analysis (Dimitrov et al. 2013), Nita is not a member of Ninetinae but of Arteminae.

The internal relationships of Ninetinae suggested by the molecular data are difficult to evaluate: they are mostly neither supported nor contradicted by morphological data. Two details are remarkable because they suggest that South America may not only be the most diverse region as far as Ninetinae are concerned but also the ancestral region of the subfamily. First, the analyses fully support a monophyletic North and Central American/Caribbean clade (Pholcophora Banks, 1896; Papiamenta Huber, 2000; and unidentified taxa from Cuba and Puerto Rico; “clade 2e” in Huber 2011b) that is either nested among South American ancestors or is sister to the South American Ibotyporanga Mello-Leitão, 1944 (with reasonable support in the 4+ genes tree only). Based on its geographic distribution, we predict that the Mexican Tolteca is also a member of this clade. Second, the two Old World genera (Ninetis and an undescribed genus from Oman) are also sister taxa (with low to modest support) and in all analyses (except for the 4+ genes analysis where Ninetis is missing) nested among South American taxa.

All our analyses exclude the name-giving genus Artema from the clade containing all other Arteminae and invariably place Artema as sister to Ninetinae (Figure 2), formally precluding the use of the name Arteminae for this clade. We do not propose a new subfamily name for this clade but treat it as ‘other Arteminae’ because we consider the position of Artema dubious. Artema shares with ‘other Arteminae’ a unique pair of structures on the procursus: a ventral pocket and a dorsal apophysis. These structures are associated with asymmetric palp insertion in both species studied with respect to this detail [Physocyclus globosus (Taczanowski, 1874), Artema nephilit Aharon et al., 2017: Huber and Eberhard 1997, Aharon et al. 2017]. The structures are present in all Arteminae, even in taxa that were previously thought to be representatives of other subfamilies, such as Chisosa and Nita (previously in Ninetinae; see above), and Wugigarra Huber, 2001 (previously in Modisiminae; see below) (Huber 2000, 2001, Huber and El Hennawy 2007). By contrast, these structures are apparently absent in all other Pholcidae. Curiously and unexplainable to us, previous molecular analyses have supported a position of Artema among ‘other Arteminae’ (Astrin et al. 2007: fig. 1, Dimitrov et al. 2013).

Some of the 99 currently known species of Arteminae are relatively large spiders with long, strong legs and high globose abdomens. The genus Artema, in particular, includes probably the largest pholcids in terms of body mass (Aharon et al. 2017). However, tiny species that were previously assigned to Ninetinae partly because of their size (Chisosa, Nita) are now included in Arteminae, and their ‘basal’ position in the cladogram suggests that ancestral Arteminae may in fact have been tiny. Just like Ninetinae, Arteminae often occur in rather dry regions, sometimes even in deserts like the Australian Trichocyclus Simon, 1908. They have a wide distribution, but are apparently absent from Sub-Saharan Africa and from South America (except for “Geneve59”, a tiny undescribed species representing a new undescribed genus on Curaçao and Aruba).

The monophyly of ‘other Arteminae’ is supported in all our analysis, even though with low support (possibly because of the dubious position of Artema, see above). Similar to our previous analysis (i.e. except for the position of Artema; Dimitrov et al. 2013), ‘other Arteminae’ is sister to Modisiminae, with variable support (reasonable support only in the RogueNaRok tree; in other trees, bootstrap support is low but SH values range from 82 to 99). This sister group relationship is weakly supported by morphology: ‘other Arteminae’ and Modisiminae lack epiandrous spigots. However, epiandrous spigots have been lost several times convergently in Pholcidae (Huber 2000, BA Huber, unpubl. data).

Internal relationships in ‘other Arteminae’ are partly resolved with reasonable support. The data suggest a large Indomalayan-Australasian clade, including the genera Trichocyclus and Wugigarra (Australia), Holocneminus Berland, 1942 (SE Asia and Pacific; excluding the misplaced and highly isolated H. huangdi Tong & Li, 2009), and a new undescribed genus (without any described species; ranging from Eastern Indonesia to the Pacific). Sister to this clade is either the New World genus Physocyclus Simon, 1893 alone or Physocyclus together with the Middle-Eastern monotypic Nita. However, support values for any of these options are low and morphological data do not favour (nor contradict) any of them. Finally, the ‘basal’ branches, i.e., those leading to the taxa outside the Indomalayan-Australasian clade and Physocyclus (and Nita in the case of the IQ-TREE analysis) lead to a group of North American and Caribbean taxa (the North American genus Chisosa being sister to a tiny undescribed species representing a new undescribed genus on Curaçao and Aruba: “Geneve59”), and to the SE-Asian Holocneminus huangdi, an isolated species that appears misplaced also by morphological criteria (A Valdez-Mondragón, pers. comm., Nov. 2015).

Modisiminae are the typical pholcids of the humid Neotropics, where they occupy a wide variety of microhabitats from leaf litter to high among the vegetation. This ecological variability is paralleled by a wide range of body forms, from tiny ground-dwelling forms (e.g., Gertsch 1982, Huber and Rheims 2011) to some of the largest pholcids with leg spans of over 15 cm (e.g., Huber and Astrin 2009, Huber 2015, 2018). With currently 480 species in 24 genera, Modisiminae is one of the two large subfamilies of Pholcidae, with several species-rich genera (e.g., Anopsicus Chamberlin & Ivie, 1938; Psilochorus Simon, 1893; Modisimus Simon, 1893; Mesabolivar González-Sponga, 1998; Carapoia González-Sponga, 1998) and many undescribed species.

All previous analyses have supported this group (Huber 2000, 2001, Bruvo Mađarić et al. 2005, Dimitrov et al. 2013), even though with minor differences in composition. The equivalent ‘New World clade’ in Huber (2001) still included the Australian Wugigarra, a genus that has since been moved to Arteminae (Dimitrov et al. 2013). As a result, Modisiminae is now considered to be restricted to the New World.

Our analyses all recover Modisiminae, but with very low support values. This is possibly due to the mysterious Andean genus Priscula Simon, 1893 (Figure 3) that is either included in Modisiminae (IQ-TREE) or not (RAxML). The position of Priscula has always been considered problematic. Simon (1893) created a separate taxon “Prisculeae” for this genus; Brignoli (1981) synonymized it with Physocyclus; the first morphological cladistic analysis (Huber 2000) supported the position of Priscula near Physocyclus but this result was explicitly doubted (Huber 2000: 129). In the molecular analysis of Dimitrov et al. (2013)Priscula was excluded because the positions of the included species varied dramatically among different types of analyses. Morphologically, Priscula differs from (other) Modisiminae by the presence of ALS piriform gland spigots and by the absence of a retrolateral apophysis on the male palpal coxa (Huber 2000), i.e., it has retained plesiomorphic characters. A sister-group relationship between Priscula and other Modisiminae appears thus plausible from a morphological point of view.

Figure 3. 

‘Basal’ Modisiminae a Gen. n., sp. n. “Br16-50” (Brazil) b Priscula andinensis? (Venezuela) c Gen. n., sp. n. “Br16-196” (Brazil) d Tupigea sp. n. “Br14-47” (Brazil) e Pisaboa silvae (Brazil) f Psilochorus imitatus (USA) g Modisimus incertus (Cuba).

Despite the low support values, we thus consider Modisiminae (including Priscula or not) a likely monophyletic group. Several morphological characters support Modisiminae (incl. Priscula): an exposed tarsal organ; the reduction of epiandrous spigots (shared with ‘other Arteminae’; see above); and a large distance between ALE and PME (Huber 2000). As indicated above (section Arteminae) our data weakly support a sister-group relationship between ‘other Arteminae’ and Modisiminae.

Within Modisiminae, many support values are extremely low, and the suggested relationships are thus unreliable (Figure 3). In addition, taxon sampling is very uneven, with some genera well represented (e.g., Carapoia, Mesabolivar, Modisimus), and others poorly represented or entirely missing (see below). However, several results are consistent among analyses and noteworthy for various reasons: they suggest groups that appear feasible in terms of biogeography; they suggest interesting evolutionary scenarios; and they suggest formal taxonomic changes, some of which have been suggested before based on morphology.

Apart from Priscula, the ‘basal’ branches within Modisiminae lead to small South American unnamed taxa (Figure 3). In particular, the two species “Br16-44” and “MACN270” are both tiny, with body lengths of 0.9 and 1.3 mm, respectively. Other ‘basal’ branches lead to an unnamed Amazonian genus (“Br16-178” and “Br16-50”; body lengths: 1.5–1.8 mm) and the Atlantic Forest genus Tupigea Huber, 2000 (body lengths: 1.3–1.9 mm; Huber 2000, Huber and Rheims 2011). This suggests a similar evolutionary scenario as proposed for ‘other Arteminae’ above, i.e., that ancestral Modisiminae may have been small ground-dwelling species. Priscula is once again the disturbing factor in this scenario: all known representatives of Priscula are medium-size to large spiders (Huber 2000), possibly surpassed (as far as body mass is concerned) by Artema only. In both Arteminae and Modisiminae, the emerging picture is one of medium-sized forms missing or disappearing early, large forms experiencing little subsequent changes in body shape and poor subsequent speciation (Artema: currently eight species; Priscula: currently 17 species), and small forms diversifying dramatically in size, shape, and numbers (‘other Arteminae’: currently 91 species; Modisiminae without Priscula: currently 463 species).

The next branch (Figure 3; Chibchea Huber, 2000 to Pisaboa Huber, 2000) includes several South American genera, some of them diverse but poorly represented in our analyses (e.g., Chibchea). The close relationship between Pisaboa and Waunana Huber, 2000 was already suggested in the original descriptions of these genera (Huber 2000), even though based on highly homoplastic characters (vertical hairs on male leg tibiae in high density; shape of apophysis on male palpal femur). A close relationship of these two genera with Chibchea either receives very low support (IQ-TREE, RAxML) or is not recovered (RogueNaRok); it is neither supported nor contradicted by morphology. Clearly, this clade needs a much denser sampling and the addition of missing taxa that are possibly related (e.g., Pomboa).

The next clade (Figure 3) includes all North and Central American and Caribbean taxa, suggesting that the ancestor of this clade arrived in the region from South America. This scenario was explicitly rejected by Dimitrov et al. (2013) based on the supposed age of the group (~120–170 Ma). However, our upcoming analysis has not been able to confirm this age (Eberle et al. 2018; we were not able to calculate convincing absolute ages from the data). The clade is recovered in most analyses (it is paraphyletic in the 4+ genes tree) but always with low support (only SH values are reasonable to high). The only geographic outlier in this clade is South American ‘Psilochorus’. North American (‘true’) Psilochorus and South American ‘Psilochorus’ each receive high to full support but are never resolved as sister taxa. Whether South American ‘Psilochorus’ are ancestral within this large clade or represent a case of back-colonization is currently impossible to say; the internal nodes in this clade have partly too low support to favour a particular scenario. The inclusion of the Central American Ixchela Huber, 2000 in this clade fits the geographic pattern and contradicts a previous speculation (in Huber 2000) that Ixchela might be close to the South American genus Aymaria Huber, 2000. In much the same way, the only Central American representative of Coryssocnemis Simon, 1893 included in our analyses is placed in this group, far away from ‘true’ South American Coryssocnemis (the polyphyly of Coryssocnemis has long been suspected: Gertsch 1971, Brignoli 1981, Huber 1998, 2000). The Cuban endemic genus Platnicknia Özdikmen & Demir, 2009 is deeply nested within the large genus Modisimus. It is resolved as sister to a distinctive group of Hispaniolan leaf-dwelling representatives of Modisimus (the “leaf-dwelling species group” in Huber et al. 2010) and synonymized below. Finally, the large genus Anopsicus (63 described species) is poorly represented in our analyses. The three species included are all undescribed, do not group together, and are nested among Modisimus. Since neither the type species of Anopsicus is included nor is a potential close relative (or at least another species from Yucatán), the monophyly and position of Anopsicus both remain dubious.

Sister to the previous North and Central American and Caribbean clade is another large, entirely South American clade (Figure 3, bottom). The sister-group relationship is very poorly supported, but the monophyly of the South American clade has modest (4+ genes) to reasonable (RogueNaRok) support. It is divided into three subclades with reasonable to full support plus the genus Aymaria that is represented by a single species and whose position within this clade is not convincingly resolved. The first subclade included is here informally called the ‘Mesabolivar clade’ (Figure 4); the second subclade is largely Venezuelan and thus called ‘Venezuelan clade’ (Figure 5); the third subclade is the genus Carapoia (Figure 5).

Figure 4. 

Mesabolivar clade a Otavaloa lisei (Brazil) b Mesabolivar maraba (Brazil) c Litoporus sp. n. “Br16-153” (Brazil) d Mesabolivar cyaneotaeniatus (Brazil) e Mesabolivar kathrinae (Brazil) f Mesabolivar saci (Brazil).

Figure 5. 

Venezuelan clade + Aymaria + Carapoia a Mecolaesthus yawaperi (Brazil) b Aymaria sp. n. “Br16-188” (Brazil) c Carapoia rubra (Brazil) d Carapoia kaxinawa (Brazil) e Carapoia pulchra (Brazil) f Carapoia agilis (Brazil).

Within the ‘Mesabolivar clade’ (Figure 4), our analyses suggest two specific relationships that are likely to have drastic taxonomic consequences. First, Litoporus Simon, 1893 is nested among ‘true’ northern South American Mesabolivar. This has been suggested before (Dimitrov et al. 2013), but that previous analysis included a single species of Litoporus whose generic identity was uncertain (Huber et al. 2013). The present analyses include several unambiguous (Amazonian) representatives of Litoporus. Our data support the monophyly of Litoporus (full support) but also its position within Mesabolivar (reasonable to high support). Second, Mesabolivar is composed of two sub-clades: ‘true’ northern South American Mesabolivar, and southern South American (largely Atlantic Forest) ‘Mesabolivar’. The southern sub-clade includes the monotypic genus Teuia Huber, 2000 (synonymized with Mesabolivar in Huber 2018; the type species of Teuia is not included but a putatively closely related species: M. sepitus). Potential formal taxonomic changes are discussed in the Taxonomy section below. The close relationship between Otavaloa Huber, 2000 and Mesabolivar is neither supported nor contradicted by morphological data.

The ‘Venezuelan clade’ (Figure 5) receives high to full support and is composed of several genera that are either known from Venezuela only (Systenita Simon, 1893, Stenosfemuraia González-Sponga, 1998), from Venezuela and Trinidad and Tobago (Coryssocnemis), or from Venezuela plus neighboring countries (Mecolaesthus Simon, 1893). A close relationship among these genera had been suspected before based on morphology (Huber 2000), and molecular data have always supported this (Bruvo-Mađarić et al. 2005: 28S data and combined analysis; Dimitrov et al. 2013). Our data suggest that Coryssocnemis may be nested within Mecolaesthus, but our taxon sampling is weak, the topology is unstable (Systenita is either nested within Mecolaesthus or not), and several internal nodes in the clade have low support. Formally, Coryssocnemis still includes several obviously misplaced species: several Central American species (see above), and several Atlantic Forest (Brazilian) species whose identity is probably impossible to resolve (poor descriptions, lost types; see Huber 2000, 2018).

The third subclade in the South American clade is Carapoia (Figure 5). Unlike Mesabolivar it is monophyletic and apparently less problematic, but just as Mesabolivar, the genus has become very difficult to diagnose, mainly because of ‘untypical’ species added to the genus based in large part on the present molecular data (Huber 2018). Both for Mesabolivar and Carapoia our analyses suggest several species groups that are also supported by morphological data. For a detailed discussion of these groups, see Huber (2018).

Smeringopinae is a relatively homogeneous subfamily (with respect to body shapes, colour, webs, and microhabitats), and in this sense similar to Ninetinae and Arteminae but very unlike Modisiminae and Pholcinae. Most of the 125 known species of Smeringopinae are medium-size to large, have long legs, elongated to cylindrical abdomens, and all have eight eyes. Another similarity to Ninetinae and Arteminae is that Smeringopinae are often found in rather arid regions. The most obvious exception is the largely humid tropical genus Smeringopina Kraus, 1957, which is also the genus with the smallest and largest representatives in the subfamily (with body lengths ranging from 2.5–10 mm) and with the widest range of microhabitats used (leaf litter to large sheltered spaces) (Huber 2013). The original distribution of the subfamily is Africa, the Mediterranean, and the Middle East. Three species have attained much wider distributions, resulting from human-mediated dispersal (Huber 2011b).

As in previous molecular analyses (Bruvo-Mađarić et al. 2005, Astrin et al. 2007, Dimitrov et al. 2013), Smeringopinae is sister to Pholcinae (Figure 1) with reasonable to high support. This relationship is also supported by morphology: the two taxa share tarsus IV comb-hairs spread over the entire length of the tarsus (Huber and Fleckenstein 2008).

The monophyly of Smeringopinae receives reasonable to high support in all our analyses. Previous molecular analyses have partly supported Smeringopinae, but also suggested rather obscure relationships [e.g., the position of Holocnemus pluchei (Scopoli, 1763) among Ninetinae in Astrin et al. 2007]. Holocnemus pluchei was included in preliminary analyses of the present data but its position was drastically unstable, so we decided to exclude it from the final analyses. Smeringopinae monophyly is rather weakly supported by morphology, i.e., by the presence of a large thoracic pit on the carapace (rather than a narrow furrow or an evenly domed carapace; cf. Huber 2011b).

Within Smeringopinae, our data strongly support a basal split between a northern clade (Mediterranean, northern Africa, Middle East, Central Asia) and a southern clade (Sub-Sahara) (Figure 6). This basal split was also recovered in a morphological cladistic analysis (Huber 2012). Within the northern clade, Hoplopholcus Kulczynski, 1908 is sister to all other genera and not close to Stygopholcus Kratochvil, 1932 as repeatedly claimed by Brignoli (1971, 1976, 1979) but contested by Senglet (1971, 2001). The genera Hoplopholcus, Stygopholcus, and Crossopriza Simon, 1893 all receive full support, but the small Mediterranean genus Holocnemus Simon, 1873 (only three described species) continues to be problematic even after the exclusion of H. pluchei. The two species of Holocnemus included in our analyses never group together, and no morphological synapomorphy is known to suggest their sister-group relationship (in fact, Holocnemus has never been revised).

Figure 6. 

Smeringopinae a Hoplopholcus sp. n. “Mar66” (Turkey) b Stygopholcus absoloni? (Bosnia and Herzegovina) c Crossopriza sp. n. “Om11” (Oman) d Smeringopus pallidus (Philippines) e Smeringopina pulchra (Ghana) f Smeringopina ankasa (Ghana).

The southern (Sub-Saharan) clade includes Smeringopus Simon, 1890 and Smeringopina, and is also supported by a unique number of epiandrous spigots (two) (Huber 2012). The paraphyly of Smeringopus has been suggested before (Dimitrov et al. 2013), and our larger data set supports this view, but with low support values. Two of the species groups of Smeringopus proposed in Huber (2012) appear closer to Smeringopina than to other Smeringopus: the chogoria group and the rubrotinctus group. Morphological data do not support this view but they neither strongly contradict it: the two species groups lack the distinctive arrangement of pores on the pore plates (in groups or ‘islands’) and the retrolateral furrow on the male palpal femur present in all other species of Smeringopus (Huber 2012). Remarkably, Smeringopus and Smeringopina are largely separated geographically, with Smeringopus being most diverse in southern and eastern Africa, and Smeringopina in western and central Africa (Huber 2012, 2013). The chogoria and rubrotinctus groups are geographically restricted to an area where Central Africa (the Guineo-Congolian center of endemism) meets East Africa (Huber 2012). Other than that, our sampling in Smeringopus is not dense enough to test the species groups proposed in Huber (2012). Remarkably, though, the isolated ‘basal’ position of S. ngangao Huber, 2012 is supported by the present analyses.

Our analyses include 30 of the 44 described species of Smeringopina (68%), and all species groups proposed in Huber (2013) except two monotypic ‘groups’ (S. fon Huber, 2013; S. ngungu Huber, 2013). Even though for some species only one gene (CO1) was sequenced, our analyses support several species groups and deeper relationships proposed previously (Huber 2013), based on cladistic analysis of morphological characters. Morphology placed the West African guineensis group as sister to all other Smeringopina; all our analyses support both the monophyly of the guineensis group and its sister-group relationship with all other congeners. The next two branches are composed of representatives of the lekoni group, which is thus here considered paraphyletic rather than monophyletic. The ankasa and cornigera groups are both supported, as is their sister group relationship to each other. The attuleh group is supported, but not as sister to the ankasa + cornigera groups but as sister to the following group. The last clade is composed of representatives of the simplex and beninensis groups, but the clear dichotomy in the molecular trees is not equivalent to these groups. Instead, the simplex group includes all ‘basal’ representatives originally assigned to the beninensis group; the beninensis group includes only those species that have a light transversal element ventrally on the abdomen (character 9 in Huber 2013, which is thus less homoplastic than previously thought).

Pholcinae resemble Modisiminae in several respects. Their highest diversity is in the humid tropics and subtropics, and a large variety of body forms reflect adaptations to different microhabitats. With currently 922 species in 26 genera, Pholcinae is also similar to Modisiminae in diversity. In contrast to Modisiminae, Pholcinae is largely restricted to the Old World, with the notable exception of the New World endemic genus Metagonia Simon, 1893 and a few possibly relict species in Pholcus and Micropholcus (Huber 2011a, Huber et al. 2014). While only a single species of Modisiminae has followed humans around the globe [Modisimus culicinus (Simon, 1893)] and one further species has spread widely in Europe and neighboring regions [Psilochorus simoni (Berland, 1911)], several synanthropic species in Pholcinae have attained worldwide distributions or extended their ranges to another continent [most notably Pholcus phalangioides (Fuesslin, 1775); Spermophora senoculata (Dugès, 1836); Micropholcus fauroti (Simon, 1887); Pholcus manueli Gertsch, 1937].

The sister-group relationship between Pholcinae and Smeringopinae is well established (see above). The same is true for the monophyly of Pholcinae. All our analyses support this subfamily (reasonable to high support), and morphological data have also supported this group (presence of male lateral proximal cheliceral apophyses, Huber 1995, 2000; tarsus IV comb hairs in a single row, Huber and Fleckenstein 2008).

Even though Pholcinae are well represented in our analyses (317 of 597 species, i.e., 53%) internal relationships in this subfamily continue to be problematic. Several ‘basal’ nodes are poorly supported (Figure 1); in part the topology is highly sensitive to different algorithms of analysis; and some details appear dubious from the perspective of morphology. However, many details are strongly supported by morphology, including some deep nodes (e.g., the Pholcus group of genera); and some nodes, even though weakly supported or in conflict with morphology, provide reasonable and testable predictions for further research (e.g., the polyphyly of Spermophora Hentz, 1841; the close relationship of certain Sri Lankan taxa with African rather than Southeast Asian taxa; the monophyly of African Pholcus).

The subfamily is here divided into three operational groups, more for the sake of convenience than as a reflection of the support values they receive. Actually, support is low for all of them, but much of this division is consistent among different analyses and may well reflect real major groups. ‘Group 1’ (Figs 7, 8) is entirely composed of small six-eyed taxa, and is roughly equivalent to what was originally subsumed under the name Spermophora. ‘Group 2’ (Figure 9 part) is also entirely composed of six-eyed taxa and is remarkable because it places the exclusively New World genus Metagonia close to African and Madagascan taxa. ‘Group 3’ (Figure 9 part, Figs 10–12) includes the fully supported Pholcus group of genera as proposed previously (Huber 2011a) and its sister genus Quamtana Huber, 2003, a sister-group relationship that has also been proposed before (Huber 2003c). In the tree shown here (and in the RogueNaRok tree), the ‘Spermophora’ dieke group has an isolated position outside of the three operational groups. In the other trees, it is part of ‘group 1’.

Figure 7. 

Pholcinae ‘group 1’ (Spermophora and relatives) a Gen. n., sp. n. “Ind206” (Halmahera); b‚ Spermophora sp. n. “Ind27” (Sumatra) c Aetana baganihan (Philippines) d Spermophora senoculata (Turkey) e Savarna tessellata (Thailand) f Wanniyala agrabopath (Sri Lanka).

Figure 8. 

Belisana and Hantu. For Belisana, the background colours signify microhabitat: red = ground; green = leaf. D, domed web; R, highly regular ‘curtain’ web. Photos a Hantu niah (Sarawak) b Hantu kapit (Sarawak) c Belisana sandakan (Sumatra) d Belisana sabah (Sabah) e domed web of Belisana sp. n. “Mal77” (Malaysia) f regular ‘curtain’ web of Belisana bohorok (Sarawak).

Figure 9. 

Pholcinae ‘group 2’ (Zatavua and relatives, Metagonia), and Quamtana (marked: non-South African species). Photos a Metagonia taruma (Brazil) b Metagonia sp. n. “Br07-1” (Brazil) c Metagonia bifida? (Brazil) d Quamtana sp. n. (cf. mabusai) (Germany).

Figure 10. 

Calapnita-Panjange clade a Kintaqa satun (Malaysia) b Meraha narathiwat (Thailand) c Calapnita vermiformis (Philippines) d Apokayana kapit (Sarawak) e Uthina sp. n. “Ind121” (Indonesia) f Panjange casaroro (Philippines).

Figure 11. 

Micropholcus-Leptopholcus clade a Micropholcus sp. n. “Br15-152” (Brazil) b Canticus sepaku (East Kalimantan) c Micromerys baiteta (West Papua) d Leptopholcus borneensis (Singapore).

Figure 12. 

Pholcus a P. creticus (Crete) b P. camba (Sulawesi) c P. mulu (Sarawak) d P. baka (Gabon) e P. sp. n. “SL43” (Sri Lanka).

This group includes some genera named long ago, like Spermophora, Belisana, and Paramicromerys Millot, 1946. Most other genera were described relatively recently and resulted either from splitting of Spermophora (e.g., Spermophorides Wunderlich, 1992; Buitinga Huber, 2003; Savarna Huber, 2005; Khorata Huber, 2005) or from the discovery and description of new species (Aetana Huber, 2005; Wanniyala Huber & Benjamin, 2005; Hantu Huber, 2016).

A Southeast Asian clade that is consistently resolved with high to full support but variably placed either inside ‘group 1’ (IQ-TREE, RogueNaRok) or outside of the three operational groups as an isolated fourth group (4+ genes, RAxML) is composed of Aetana, Southeast Asian ‘Spermophora’, and an undescribed new genus from Indonesia (“Ind206”). Morphological data have suggested a close relationship of Aetana with Savarna, Khorata, and Hantu (Huber et al. 2015). The positions of those three genera in our molecular trees are all unstable and problematic (see below). Thus, we consider it premature to conclude that the morphological data were misleading, and suggest that the positions of Savarna, Khorata, and Hantu need further analysis. A similar problem occurs with Southeast Asian ‘Spermophora’. The monophyly of the five species included receives reasonable to high support, but this group does not seem to be close to the type species S. senoculata. However, the position of S. senoculata varies strongly among analyses, and the idea that Southeast Asian taxa are in fact congeneric with S. senoculata (Huber 2005a) should not yet be discarded based on the present molecular data.

In Aetana, our analyses include 16 of 18 (89%) described species plus two undescribed species. The monophyly of the genus is highly to fully supported even though morphological support appeared weak (Huber et al. 2015). All four species groups proposed after cladistic analysis of morphological characters (Huber et al. 2015) are supported, but with different relationships among each other. Most of these relationships among species groups receive low support, but the kiukoki group is resolved as sister of the omayan group (with modest support) and this is in conflict with the results from morphology (Huber et al. 2015). The two unnamed subgroups within the kinabalu group and within the omayan group, respectively, proposed in Huber et al. (2015) are all recovered (with modest to full support).

The next clade within ‘group 1’ (Figure 7) includes three taxa whose position varies strongly among different analyses (see above): the type species of Spermophora, S. senoculata, and the Southeast Asian genera Khorata and Savarna. Spermophora senoculata is alternatively resolved as sister to the African ‘Spermophora’ akwamu group (RAxML) or to the African ‘Spermophora’ kyambura Huber & Warui, 2012 (4+ genes, RogueNaRok). Its sister group is essentially unknown. As indicated above, a close relationship with Southeast Asian ‘Spermophora’, even though never recovered by our analyses, should not be definitely discarded. Khorata and Savarna are sister taxa in some analyses (low support; IQ-TREE, RogueNaRok), but wide apart in others. The former result is considered more plausible for two reasons: (1) morphology supports a close relationship between Khorata and Savarna (Huber et al. 2015); (2) the alternative topology (4+ genes, RAxML) places the Southeast Asian Savarna as sister to an East African clade.

The large Asian genus Belisana (Figure 8) is well represented in our analyses (30 species) but seems to suffer from rogue taxa, paralogs, and/or other unidentified problems. Only the RogueNaRok tree resolves a monophyletic Belisana. In other analysis, either Hantu (RAxML) or Hantu and ‘Spermophora’ kyambura are nested within Belisana (IQ-TREE). A close relationship between Belisana and Hantu (that is also suggested in the RogueNaRok tree) is strongly contradicted by morphology: several characters support a close relationship between Hantu, Khorata, and Savarna (Huber et al. 2015). We have no explanation for the position of Hantu in our trees. Intriguingly, H. niah Huber, 2016 (but not H. kapit Huber, 2016) was placed in a clade together with Khorata and Savarna in preliminary analyses of the present data. On the other hand, the African ‘Spermophora’ kyambura might indeed be close to Belisana. In fact, had it been collected in Southeast Asia, it would probably have been assigned to Belisana. It was tentatively assigned to Spermophora because African ‘Spermophora’ were polyphyletic anyway and because the closest known record of Belisana was from India, more than 5000 km east. However, the position of ‘Spermophora’ kyambura varies among analyses and should be considered unresolved.

Our sample of Belisana includes numerous representatives from different microhabitats (litter and leaves) and with different types of webs (‘usual’ pholcid domed sheets and highly regular ‘curtain’ webs; Figs 8e–f; see also Deeleman-Reinhold 1986a, Huber 2005b). The present data suggest multiple microhabitat shifts within Belisana, but note that many nodes within the genus have very low support values. These low values also impede a proper interpretation of the fact that the two species with a ‘usual’ domed web (marked with D in Figure 8) included in the analyses (B. “Mal77”, B. tambligan Huber, 2005) are not ‘basal’ but nested among species with highly regular ‘curtain’ webs (marked with R in Figure 8) [confirmed for B. bohorok Huber, 2005; B. leuser Huber, 2005; B. “Bor85”; B. junkoae (Irie, 1997); B. sabah Huber, 2005; BA Huber, unpubl. data].

Except for the Sri Lankan genus Wanniyala, all remaining taxa of Pholcinae ‘group 1’ (Figure 7) are African, Madagascan, and Mediterranean. They are grouped together but with very low support. South African and Madagascan ‘Spermophora’ were not available for sequencing and are thus not included in our analyses; we predict they are members of this clade. As mentioned above, some analyses (RAxML, 4+ genes) placed the East Asian genus Savarna within this clade; we consider this topology dubious.

A close relationship between the West African ‘Spermophora’ tonkoui group and Wanniyala is suggested in all our analyses, even though with low support (only SH values are consistently at 96–97). This relationship is also supported by morphology: the two taxa share a hinged procursus with a membranous process arising from the proximal part (see Huber and Benjamin 2005: fig. 7, Huber 2003b: fig. 293, Huber and Kwapong 2013: fig. 101).

The following clade (Figure 7) places the Central African ‘Spermophora’ awalai group as sister to the Macaronesian and Mediterranean genus Spermophorides, both together sister to the Madagascan genus Paramicromerys, and all together sister to an undescribed Madagascan genus (“CAS13”). Support for these relationships is modest, and the clade is different in composition in the 4+ genes tree (Spermophorides is missing from this analysis).

The last clade in Pholcinae ‘group 1’ is highly to fully supported in all analyses and includes the East African genus Buitinga and East African ‘Spermophora’, each with full support in all analyses. The sister group relationship between these two taxa makes sense geographically but is not evident from morphology.

This operational group (Figure 9 part, i.e., without the ‘Spermophora’ dieke group and Quamtana) is similar to ‘group 1’ in that it is composed entirely of six-eyed species. It is weakly supported, indicating that the exact placement of its two clades among other Pholcinae remains dubious. The two clades, however, both receive high to full support in all analyses. The first clade unites the African genera Anansus Huber, 2007 and Nyikoa Huber, 2007 with the Madagascan genus Zatavua Huber, 2003. In a cladistic analysis of morphological characters (Huber 2007), the group was also recovered (even though as paraphyletic) when using successive character weighting (but not when using equal character weights). The character supporting this close relationship was the proximal cheliceral apophyses that point backwards in Anansus just as in Nyikoa and Zatavua (Huber 2007). The idea that these genera might be ‘basal’ in Pholcinae, i.e., sister to all other Pholcinae (Huber 2003a, 2007) is not supported by our analyses, but considering the low support values at deeper nodes within the subfamily it is neither strongly contradicted.

The second clade is the New World genus Metagonia. The genus is species-rich (currently 85 species) and ranges from Argentina to Mexico. The monophyly of the genus has never been seriously contested, and its position among the otherwise almost exclusively Old World Pholcinae has been strongly supported before, both using morphology (Huber 2000) and molecules (Bruvo-Mađarić et al. 2005, Dimitrov et al. 2013). Our analyses include 30 species of Metagonia, and provide for the first time a test of the operational species groups proposed in Huber (2000: 54–55). Even though those groups were not based on formal cladistic analysis (such an analysis has not yet been performed for Metagonia) but on overall and specific similarities, all of them appear mostly or entirely congruent with the present analyses (the only exception is the single aberrant M. globulosa). They are listed here in the sequence in which they appear in Figure 9, with newly proposed informal names. (1) taruma group (group “3” in Huber 2000); a South American group that is here resolved as monophyletic and not as a paraphyletic ‘basal’ group as speculated before (Huber 2000); (2) petropolis group (no species was known in 2000); a group of litter-dwelling species restricted to the Brazilian Atlantic Forest (Huber et al. 2005b); (3) bifida group (group “1” in Huber 2000); this South American group includes the type species M. bifida Simon, 1893; all species included share a sclerotized epigynum and (except for the ‘basal’ undescribed species “G062”) a distinctively bifid abdomen; (4) potiguar group (not recognized in Huber 2000); this group includes cave dwelling species in Brazil (M. potiguar Ferreira et al. 2011) and Jamaica (M. jamaica Gertsch, 1986) and the aberrant litter-dwelling M. globulosa Huber, 2000 (which was misplaced in group “5” in Huber 2000); rica group (group “4” in Huber 2000); a mainly North and Central American group, possibly ranging into South America, but not including all Caribbean and Central American species as speculated previously (Huber 2000; see previous group); (5) furcata group (group “5” in Huber 2000, together with M. globulosa); includes only M. furcata Huber, 2000 and the undescribed species “Br09-55”; as suspected previously (Huber 2000), it is close to (sister of) the next group; (6) delicata group (group “2” in Huber 2000); this group is composed of very small species and ranges from Mexico to northern Argentina.

A sister-group relationship between the African genus Quamtana and the Pholcus group of genera (Figure 9) is recovered in all our analyses. Support values are low, but a morphological cladistic analysis has partly suggested the same relationship (based on a distinct sclerite connecting the genital bulb to the palpal tarsus; Huber 2003c). The monophyly of Quamtana is highly supported in all analyses (except for the 4+ genes analysis). It was also supported by morphological data when using character weighting (but not in the equal weights analysis; Huber 2003c).

Within Quamtana (Figure 9), our data suggest that there is no simple geographic pattern with respect to South African species (the large majority) versus species from other parts of Africa (marked in Figure 9). By contrast, three species groups with reasonable to full support include species from both South Africa and other regions: the South African Q. filmeri Huber, 2003 is sister to the Madagascan undescribed species “CAS5”; the South African Q. vidal Huber, 2003 and Q. umzinto Huber, 2003 are placed in a group with species from East and Central Africa (Q. kabale Huber, 2003, “Cam117”); and the group including the South African Q. embuleni Huber, 2003 and Q. bonamanzi Huber, 2003 also includes species from East and Central Africa (Q. kitahurira Huber, 2003, Q. oku Huber, 2003). We suspect that Quamtana was once widely distributed throughout Africa but largely replaced by more modern taxa in humid regions and extinguished in northern Africa. The Paris amber fossil Quamtana huberi Penney, 2007 supports this view, but its generic assignment is uncertain (Penney 2007).

All remaining clades together (Figs 10–12) represent the Pholcus group of genera sensuHuber (2011a). This clade was first proposed in Huber & Fleckenstein (2009) based on the distinctive simplified shape of the tarsus IV comb-hairs, and later supported in a cladistic morphological analysis by an additional character (female epigynal ‘knob’) (Huber 2011a). All our analyses fully support this clade. The previous morphological analysis (Huber 2011a) identified two major problems within this clade: (1) relationships among genera were basically unresolved, resulting in large polytomies; and (2) several species groups assigned to Pholcus appeared more closely related to other genera. The present analysis strongly supports the polyphyly of Pholcus in its previous composition, and it provides for the first time a reasonable framework to redefine generic limits in this large group (currently 501 species).

The first major clade within the Pholcus group of genera (Figure 10) is composed of three Southeast Asian genera (Calapnita, Panjange, Uthina Simon, 1893) as well as several Southeast Asian and Sri Lankan species groups that were originally tentatively assigned to Pholcus (Huber 2011a; Huber et al. 2016a, 2016b, Huber and Dimitrov 2014). We informally call it the ‘Calapnita-Panjange clade’ because many species in this group are leaf-dwellers, and representatives of Calapnita and Panjange are particularly strongly adapted to life on green leaves. Remarkably, even some of the species collected in the leaf litter (under large dead leaves on the ground) look like leaf-dwellers rather than litter dwellers (i.e., they have long abdomens, long legs, light colouration; e.g., Kintaqa satun (Huber, 2011) and K. schwendingeri (Huber, 2011); and Malaysian representatives of Tissahamia, previously the Pholcus ethagala group). Ancestral character state reconstruction suggests that the ancestor of the entire clade was leaf-dwelling (Eberle et al. 2018).

The present analyses reject the monophyly of Calapnita (Figure 10). A recent cladistic analysis of morphological data (Huber 2017) resolved Calapnita as monophyletic but with low support (< 50 using Jackknifing). On the other hand, support for the two subgroups, previously called phyllicola group and vermiformis group, is full in all analyses. The two species groups have been identified long ago (Deeleman-Reinhold 1986b), and have been supported by cladistic analysis (Huber 2017). Our analyses strongly suggest that the vermiformis group is closer to species previously in Pholcus than to the phyllicola group (see below). The phyllicola group is thus elevated to genus rank (Nipisa; see Taxonomy section below).

Within Nipisa (Figure 10), the internal relationships proposed previously (Huber 2017) are mostly supported even though data gaps are severe in this genus (several species with only two genes): (1) N. lehi (Huber, 2017) [but not N. kubah (Huber, 2017)] is a ‘basal’ species, i.e., sister to all other species (reasonable to high support); (2) a clade including the species with egg-sacs that have all eggs aligned in a single row (weak support, possibly because N. kubah is included, which is contradicted by morphology and egg-sac shape); (3) a clade including N. semengoh (Huber, 2011) and its sister group, characterized by the position of the tarsal organ on a turret, a serrate embolus, and the shape of the pore-plates (full support).

The relationships within ‘true’ Calapnita (previously vermiformis group) proposed in Huber (2017) are only partly supported: (1) a clade with a continuous connection between epigynal plate and ‘knob’ (all species in the present analysis except C. bario Huber, 2017 and C. saluang Huber, 2011; high support); (2) within the previous clade, a clade characterized by a prolateral process at the tip of the procursus (in the present analysis: C. nunezae Huber, 2017 and C. dinagat Huber, 2017; full support).

The present analyses also reject the monophyly of Panjange (Figure 10). They split the genus into two unrelated lineages, one of which is equivalent to what was previously called the nigrifrons group; the other is equivalent to the previous vermiformis + cavicola groups (Deeleman-Reinhold and Platnick 1986, Huber and Nuñeza 2015). Our analyses place each group with reasonable to full support in clades together with species previously assigned to Pholcus. A morphological cladistic analysis has recently supported the monophyly of Panjange based on the presence of parallel ridges ventrally on the procursus and on the reduction of the bulbal uncus (Huber and Nuñeza 2015). However, the monophyly was lost when using specific weighting parameters (implied weighting with K = 1 and K = 2), and some morphological characters do in fact support the split of Panjange: (1) the loss of distal cheliceral apophyses in ‘true’ Panjange and its closest relatives according to the present analyses; (2) the loss of an uncus in ‘true’ Panjange and its closest relatives according to the present analyses. The nigrifrons group is thus elevated to genus rank (Apokayana; see Taxonomy section below).

Apokayana is recovered with full support. This is remarkable considering the fact that in the morphological analysis its equivalent (the Panjange nigrifrons group) was supported by a single homoplastic character only (Huber and Nuñeza 2015). Within the genus, our analyses identify two subgroups with full support each. These groups do not correspond to the relationships suggested in Huber and Nuñeza (2015). In that analysis, each node was based on a single character, some of them not particularly convincing. We thus tend to prefer the present grouping even though our matrix is particularly incomplete in this genus (we did not manage to get 28S and CO1 sequences for any of the six species included).

The monophyly of ‘true’ Panjange (vermiformis + cavicola groups) is supported by several morphological characters (Huber and Nuñeza 2015), and receives high support in our present analyses (except for the 4+ genes analysis). The cavicola group (including also the two undescribed species “Ind103” and “Ind109”) was recovered as paraphyletic in Huber and Nuñeza (2015) but is here resolved as monophyletic. By contrast, the lanthana group which was supported by two morphological characters, one of them considered particularly strong (the unique direction of the embolus, pointing in the opposite direction of the appendix) is resolved as monophyletic only in the RogueNaRok tree; in the IQ-TREE and RAxML trees it is paraphyletic with respect to the cavicola group (actually, these trees suggest a basal trichotomy). Within the lanthana group, three species (P. malagos Huber, 2015; P. casaroro Huber, 2015; P. camiguin Huber, 2015) share asymmetric male pedipalps, a character that is extremely rare in spiders (Huber et al. 2007, Huber and Nuñeza 2015). This group is not recovered in any of the present analyses, where it consistently includes the symmetric P. lanthana Deeleman-Reinhold & Deeleman, 1983 (requiring a regain of symmetry or two origins of asymmetry). Only the sister group relationship between P. dinagat Huber, 2015 and P. marilog Huber, 2015 is strongly supported by both morphology and molecules. In conclusion, alternative topologies within the lanthana group are supported by seemingly strong molecular and morphological data, respectively.

Ten species groups previously assigned to Pholcus (in Huber 2011a, Huber et al. 2016a, 2016b) are representatives of the ‘Calapnita-Panjange clade’ (Figure 10). Of these, nine are entirely Southeast Asian; only the ethagala group (now Tissahamia) has representatives in Southeast Asia and Sri Lanka. For some of these species groups, our data provide strong evidence about the sister-group or close relatives. All of these groups are here transferred from Pholcus to new genera (see Taxonomy section below).

A sister-group relationship between Kelabita (previously the Pholcus andulau group) and Apokayana (previously the Panjange nigrifrons group) is fully supported in all our analyses (except for the 4+ genes tree where Kelabita is not represented). Both genera are restricted to Borneo and share habitus, colouration, web structure, and microhabitat (Huber and Leh Moi Ung 2016, Huber et al. 2016a).

The western Indonesian ‘Pholcus kerinci group’ (Huber 2011a) and the Philippine ‘Pholcus domingo group’ (Huber et al. 2016b) are both fully supported, as is their sister-group relationship (Figure 10). They are joined in the new genus Teranga (see Taxonomy section below). Together with Tissahamia (previously the Pholcus ethagala group) and with ‘true’ Panjange they form a clade that is recovered in all analyses with reasonable support. This clade was supported in almost exactly the same composition by morphological cladistic analysis (Huber 2011a) except that the ‘Panjange’ nigrifronsgroup was also included (the ‘Pholcus domingo group’ was not yet known in 2011). The clade is supported by the loss of the bulbal uncus and by the loss of distal male cheliceral apophyses [in Huber 2011a, the latter character supports a more inclusive taxon (including Leptopholcus Simon, 1893) that is strongly rejected by the present molecular data]. Tissahamia consists of two subgroups, a Sri Lankan subgroup and a Southeast Asian (Malaysian Peninsula, Sumatra) subgroup. The subgroups are consistently recovered in all our analyses with modest to reasonable support, but the monophyly of the entire group is only recovered in the IQ-TREE analysis (low support). Morphological analysis recovered the group, but with varying support depending on weighting regime (Huber 2011a).

Three genera composed of species that were previously assigned to Pholcus are consistently placed in a highly supported clade together with ‘true’ Calapnita (Figure 10): Paiwana (previously not assigned to a group), Muruta (previously the Pholcus tambunan group; Huber et al. 2016b), and Meraha (previously the Pholcus krabi group; Huber et al. 2016b). We know of no convincing morphological synapomorphy for this group but note two interesting similarities: representatives of ‘true’ Calapnita and of Meraha share the loss of piriform gland spigots on the anterior lateral spinnerets (Huber 2011a, 2017, Huber et al. 2016b); representatives of ‘true’ Calapnita and of Muruta and Paiwana share the distinctive shape of the epigynum (roughly triangular, with ‘knob’ directed towards anterior; Huber et al. 2016b, Huber 2017; Huber and Dimitrov 2014).

For two further genera composed of species previously assigned to Pholcus the present analysis supports the monophyly but gives not clear indication about their closest relatives within the ‘Calapnita-Panjange clade’ (Figure 10): Pribumia (previously the Pholcus minang group; Huber 2011a) and Kintaqa (previously the Pholcus buatong group; Huber et al. 2016a). All analyses except the IQ-TREE analysis place Kintaqa as sister to Uthina, but with low support. We know of no potential morphological synapomorphy that links these two groups. Pribumia is in our analysis represented by four species. Of these, P. diopsis (Simon, 1901) is never placed within the group; together with P. atrigularis (Simon, 1901) it is detected as a rogue taxon and excluded in the RogueNaRok tree. External relationships of Pribumia remain dubious. The hypothesis that the genus might be close to Tissahamia (previously the ‘Pholcus ethagala group’; Huber 2011a) is supported by numerous distinctive morphological similarities but it is not supported by the present data. However, note that in our analysis Pribumia suffers seriously from missing data (we were not able to sequence 28S for any of the four species).

The second major clade within the Pholcus group of genera (Figure 11) is composed of four ‘old’ genera (Micropholcus; Leptopholcus; Micromerys Bradley, 1877; Pehrforsskalia Deeleman-Reinhold & van Harten, 2001) and Cantikus (previously the Pholcus halabala group; Huber 2011a, Huber et al. 2016a). Except for one clade of Neotropical Micropholcus, all representatives are Old World taxa. We informally call it the ‘MicropholcusLeptopholcus clade’. This clade receives full support in all our analyses, and major internal relationships are also well resolved. Three subclades are fully supported each: Micropholcus; Cantikus; and a subclade including Leptopholcus, Micromerys, and Pehrforsskalia. All analyses put Micropholcus as sister to Cantikus, but with modest support.

Micropholcus is ecologically diverse, including ground-dwelling as well as rock- and leaf-dwelling species, and together with Pholcus it is also the only genus with autochthonous species in both the New and Old World. Our analysis rejects the previous idea that Micropholcus is ‘basal’ in the Pholcus group of genera (i.e., in a basal trichotomy, with Sihala occupying the second branch and all other taxa the third branch; Huber 2011a). Within Micropholcus, our analyses all support a monophyletic New World clade, but with low support values (reasonable support in the 4+ genes analysis). Within the New World clade, a Caribbean clade is fully supported. A remarkable sister-group relationship that is highly supported by the present data is between the Moroccan ‘Pholcus’ agadir (now transferred to Micropholcus; see Taxonomy section below) and the undescribed Philippine species “Phi114”. Both have very limited distributions; only one further species of Micropholcus (other than the pantropical M. fauroti) is known from between Morocco and the Philippines: M. jacominae Deeleman-Reinhold & van Harten, 2001 from Yemen. We suspect that Micropholcus in the Old World has a relict distribution, just as it has been hypothesized for South American Micropholcus (Huber et al. 2005a, 2014).

Cantikus was recently revised (as ‘Pholcus’ halabala group; Huber et al. 2016a) and divided into a ‘core group’ that was supported by numerous morphological and behavioral similarities, and a group of species that were assigned to the group tentatively. This tentative assignment was based mainly on preliminary results from the present molecular analysis; a putative morphological synapomorphy for the entire genus Cantikus was and is not known. The present analyses fully support both the entire genus and the core group; the genus includes C. quinquenotatus (Thorell, 1878), making the quinquenotatus group proposed in Huber (2011a) obsolete; and it highly to fully supports the sister group relationship between the two rock-dwelling species C. kuhapimuk Huber, 2016 and C. khaolek Huber, 2016.

The clade including Leptopholcus, Micromerys, and Pehrforsskalia (Figure 11) was only partly supported in a previous cladistic analysis of morphological data (Huber 2011a): while Leptopholcus and Micromerys were consistently seen as sister taxa (with a mono- or paraphyletic Leptopholcus), the position of Pehrforsskalia varied widely. The characters supporting a close relationship among the three genera are the distal position of the lateral apophyses on the male chelicerae, and the absence of frontal cheliceral apophyses (Huber 2011a). The present analyses fully support this clade. Within the clade, ‘basal’ relationships are unresolved, essentially resulting in a tetrachotomy: (1) ‘Leptopholcus’ podophthalmus (Simon, 1893) is not clearly included in ‘true’ Leptopholcus. (2) The Australasian Micromerys receives full support in all analyses. (3) The African Pehrforsskalia is only represented by its type species. (4) ‘True’ Leptopholcus receives reasonable to full support and includes both African and Asian representatives but not the Asian L. podophthalmus (and its putative close relative L. tanikawai Irie, 1999 that is not included in our analyses). Within Leptopholcus, our data provide little resolution, but an Asian clade (represented by L. borneensis Deeleman-Reinhold, 1986 and L. kandy Huber, 2011) receives reasonable support. Among these four clades, Pehrforsskalia is the only one that does not share the distinctively serrated tip of the male palpal trochanter apophysis (Huber 2011b), suggesting that it may be sister to the other three clades.

The third and last major clade within the Pholcus group of genera is ‘true’ Pholcus (Figure 12). Support for this group is very low in the IQ-TREE analysis, which reflects the fact that one of the two basal subclades (including the phungiformes and bidentatus groups and P. mentawir Huber, 2011) is closer to the Micropholcus-Leptopholcus clade than to ‘true’ Pholcus in some analyses (RAxML, RogueNaRok). By contrast, the 4+ genes analysis recovers the monophyly of ‘true’ Pholcus with reasonable support, suggesting that the poor support or non-monophyly of ‘true’ Pholcus in some analyses may result from the many missing data in our full matrix.

Even after removing the eleven species groups that are here placed in the CalapnitaPanjange clade and in the MicropholcusLeptopholcus clade, Pholcus continues to be the most species rich genus in Pholcidae. It now contains 321 species, most of which are distributed in tropical and subtropical Old World regions. The only exception is the kingi group with ten species in the southeastern USA (Huber 2011a). Most species of Pholcus resemble the synanthropic type species P. phalangioides in being relatively large, long-legged, brown, and in having a cylindrical abdomen; most or all of these species build their webs in large sheltered spaces. However, the genus is ecologically diverse and includes small litter dwellers with relatively short legs, rock- and ground dwellers with oval abdomens, and pale leaf-dwellers with worm-shaped abdomens.

In a first effort to structure the known diversity of Pholcus, the genus was divided into 29 operational species groups (Huber 2011a), including 25 species groups in the ‘core group’, i.e., in ‘true’ Pholcus. Even though the aim was to identify monophyla, some groups were explicitly proposed as ‘waste baskets groups’ (e.g., the bamboutos group) or as “probably not monophyletic” (e.g., the circularis group). The present analysis clarifies a number of relationships, it supports several of the species groups and rejects others, and it confirms the non-monophyly of some groups as suspected. However, we acknowledge that internal relationships in Pholcus remain highly uncertain and need considerably more work. Our data seem to suffer from two main problems that result in variable topologies among different types of analyses: (1) Even though Pholcus is in our analyses represented by more species than any other genus (59), our sample is still highly incomplete, including only 18% of the described species and entirely missing seven of the previously suggested species groups (alticeps, nenjukovi, ponticus, zham, yichengicus, taishan, and nagasakiensis groups). (2) The percentage of missing sequences is high in Pholcus, partly due to the fact that we identified paralogs for 28S and 18S that we excluded, partly due to other unidentified problems.

Of the 25 operational species groups within ‘true’ Pholcus proposed previously (Huber 2011a), ten are supported by the present data: phungiformes group, bidentatus group, calligaster group, Macaronesian group, gracillimus group (excl. P. mentawir), bicornutus group, chappuisi group, lamperti group, debilis group (incl. P. nkoetye Huber, 2011 and P. kribi Huber, 2011), and guineensis group. Four groups are represented by single species (taarab group, ancoralis group, phalangioides group, kingi group). Seven groups are missing in the analyses (see above). For the remaining four groups, the present analyses reject the monophyly: (1) The bamboutos group is polyphyletic as expected and the six species in our analyses split into four parts; of these P. kribi is moved to the debilis group; P. bamboutos Huber, 2011 is close to the guineensis group; the affinities of the other four species are unclear. (2) The circularis group is represented by three species; of these, P. nkoetye is moved to the debilis group; P. leruthi Lessert, 1935 and P. rawiriae Huber, 2014 are sister species and close to the guineensis group. (3/4) The opilionoides and crypticolens groups are both rejected, but together with the North American kingi group they form a monophylum with reasonable to high support (except for the 4+ genes analysis) but with unknown affinities with other groups.

The present analysis identifies two major clades within ‘true’ Pholcus that are remarkable even though support values are low to modest. (1) A clade combining the ancoralis, gracillimus, and bicornutus groups is composed of large dark Southeast Asian and Australasian species; a close relationship between the ancoralis group and the bicornutus group has been suspected before, based on male ocular area modifications (Huber 2011a: 314). (2) A large clade including all Subsaharan African taxa. This clade has low bootstrap support but SH values range from 81 to 96, so we consider this a first tentative indication that tropical African Pholcus might form a large monophylum. The two species that disrupt this picture were both identified as rogue taxa: P. taarab Huber, 2011 (which is not included in the clade but is African), and P. phalangioides (which is included but is most probably not originally African). On the other hand, the inclusion of the Sri Lankan genus Sihala Huber, 2011 in this clade is plausible, even though weakly supported. Our data highly support the inclusion of Sihala in ‘true’ Pholcus, but neither morphology nor molecules seem to give an indication about its sister taxon.