The data collected for the present study includes material from collections and collecting field trips. Voucher specimens of DNA extraction or morphological study are deposited in the following collections: Facultad de Ciencias, Universidad de la República, Montevideo, Uruguay (FCE-Ar, M. Simó), Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”, Buenos Aires. Aires, Argentina (MACN-Ar, M. Ramírez), Laboratório Especial de Coleções Zoológicas, Instituto Butantan, São Paulo, Brazil (IBSP, A.D. Brescovit), Museu de Ciências e Tecnologia, Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, Brazil (MCTP, R. Teixeira), Laboratorio de Biología Reproductiva y Evolución, Cátedra de Diversidad Animal I, Universidad Nacional de Córdoba, Córdoba, Argentina (LABRE-Ar, M. Izquierdo), California Academy of Sciences, San Francisco, California, USA (CASENT, L. Esposito). Information and photographs of type material were obtained from the following institutions: The Natural History Museum, London, England (J. Beccaloni), Museo de Biología de la Universidad Central de Venezuela, Caracas, Venezuela (E. Guerrero), Muséum National d´Histoire Naturelle, Paris, France (C. Rollard), Museu Nacional, Universidade Federal do Rio de Janeiro, Brazil (A. B. Kury), Museum of Comparative Zoology, Harvard University, Cambridge, USA (G. Giribet), Museum Wiesbaden, Wiesbaden, Germany (S. Kridlo), National Museum, Dublin, Ireland (P. Viscardi), Naturmuseum Senckenberg, Frankfurt, Germany (P. Jäger), Peabody Museum of Natural History, Yale, USA (R. J. Pupedis), Royal Belgian Institute of Natural Sciences, Brussels, Belgium (F. Trus), Swedish Museum of Natural History, Stockholm, Sweden (J. Stigenberg).

Field work for this study was carried out in: Uruguay (Clemente Estable Biological Research Institute in Montevideo, Melilla in Montevideo, and Montes del Queguay Protected Area in Paysandú); Argentina (Ischigualasto Provincial Park in San Juan, Lanín National Park in Neuquén, and El Palmar National Park in Entre Ríos); Chile (Río Clarillo National Park in Santiago) and Brazil (Parque das Dunas in Salvador, Bahia, and Pró-Mata Reserve in São Francisco de Paula, Rio Grande do Sul). The collecting sites were selected based on previous records, and new ones aimed to cover the largest area and diversity of environments.

Male and female genitalia were examined under stereomicroscopes (Nikon SMZ 10 and SMZ 745), and specimens were identified by comparing them with images of type specimens, if available, or with original descriptions and taxonomic revisions. Maps were made with SimpleMappr (http://www.simplemappr.net) (Fig. 1). When coordinates were not specified in the original labels, localities were georeferenced using Google Maps, and in those cases provided between brackets in the lists of examined material.

Figure 1. 

Map showing the sampling localities of Allocosinae specimens from South America included in the phylogenetic analyses.

Specimens sequenced (N=73) included museum and fresh, field-collected material. DNA was extracted from the left legs of the specimens (two or four, depending on the size of the specimen) using the DNeasy Tissue Kit (Qiagen), following the manufacturer’s instructions. For collection specimens, which had not been kept in appropriate conditions to preserve the DNA, the QIAamp DNA Micro Kit (Qiagen) was used following the manufacturer’s instructions. The complete specimens were immersed in the lysis buffer after performing a puncture of the carapace to expose internal tissue. DNA was quantified, and the purity was determined by spectrophotometry using a ND 1000 NanoDrop (Thermo Scientific). The DNA extraction codes of all the specimens used, as well as their sex and locality data are shown in Table S1. The selection of molecular markers was made based on the information available in the literature and public genetic sequence databases (GenBank). Fragments of six genes were amplified using the PCR; the mitochondrial cytochrome oxidase c subunit 1 (cox1), 12S rRNA (12S) and NADH dehydrogenase subunit 1 (nad1); and the nuclear histone H3 (h3), histone H4 (h4) and 28S rRNA (28S). Standard spider primers and protocols were used for each gene (Wheeler et al. 2017; Planas et al. 2013). The following primer pairs were used, cox1: C1-J-1490 (Folmer et al. 1994) and C1-N-2662 or Porricosa-R1 (Laborda et al. 2022); 12S: SR-J-14233 (Simon et al. 1994) and SR-N-14503 (Croom et al. 1991); 28S: 28S “O” and 28S “C” (Hedin and Maddison 2001), nad1: TL-1-N-12718 (Hedin 1997) and M510 (Murphy et al. 2006); h3: H3F and H3R (Colgan et al. 1998); h4: H4F2er and H4F2s (Pineau et al. 2005). PCR reactions were carried out following the protocol for Taq DNA Polymerase using Standard Taq Buffer (M0273, New England Biolabs Inc.). The PCR conditions were as follows: initial denaturation at 95°C for 3 min; followed by 35 cycles at 95°C for 30 s, from 42° to 58°C for 45 s (depending on primers), and extension at 68°C for 45 s; with a final extension step at 68°C for 5 min. For the cox1 and nad1 gene fragments, successful amplification was achieved with an annealing temperature at 45°C, for 12S at 42°C, for 28S at 58°C, while for h3 and h4 at 48°C. The products obtained by PCR were visualized by 1% agarose gel electrophoresis and purified with FastAP Thermosensitive Alkaline Phosphatase and Exonuclease I enzymes (Thermo Fisher Scientific). PCR products were sequenced in both directions using the sequencing service of Macrogen, Seoul, Korea. DNA sequences were edited and aligned using the trial version of the Geneious program (Kearse et al. 2012). In addition to the sequences obtained in this study, the sequences published in Piacentini and Ramírez (2019) and deposited in the GenBank and BOLDSYSTEMS repositories were used (Table S2). These sequences come from a broad sampling of Lycosidae, so they were used as out-groups in the analysis to provide an extensive context for the subfamily Allocosinae.

Based on the molecular markers obtained in this study and those previously published in Piacentini and Ramírez (2019), Gonnet et al. (2021), and Laborda et al. (2022), a concatenated matrix was constructed, including representatives of other currently recognized Lycosidae subfamilies to test the monophyly of Allocosinae. This matrix (M1) concatenates the sequences of five molecular markers (cox1, nad1, h3, 12S, and 28S) including 154 taxa, rooted in Cupiennius spp. A second matrix (M2, 93 taxa, rooted in Aglaoctenus lagotis (Holmberg, 1876)) was built by reducing the outgroups to focus on resolving the internal relationships of Allocosinae. These matrices were analyzed using maximum likelihood (ML) and Bayesian inference (BI) (see below). A third matrix (M3) was constructed keeping only the Allocosinae specimens (86 taxa, rooted in “Arctosa” sapiranga, the taxon considered as a sister group to the rest of Allocosinae in previous analyses, see Piacentini and Ramírez 2019), removing the 28S marker (due to lack of data on terminals and low genetic variation) and adding the h4 marker sequences (see Table S1), and analyzed only under Bayesian inference. In a complementary analysis, public access sequences (GenBank and BoldSystems) from cox1 of four North American Allocosa species (A. funerea, A. noctuabunda, A. parva and A. absoluta) were added to M2 to create a fourth matrix (M4).

The alignment of cytochrome oxidase c subunit 1 (cox1), NADH dehydrogenase subunit I (nad1), histone H3 (h3), and histone H4 (h4) was trivial since no insertions or deletions were inferred. The 12S and 28S alignments were made with MAFFT (Katoh and Standley 2013) using the L-INS-i algorithm. Gappy fragments of uncertain positional homology were removed with Gblocks 0.91b (Castresana 2000). PartitionFinder2 was used to select the partition schemes and molecular evolution models that best fit the data (Lanfear et al. 2016). These analyses were based in two initial partition schemes, by genes (cox1 / 12S / nad1 / h3 / 28S) and by genes and 1st, 2nd and 3rd codon positions of the coding genes (cox1_p1 / cox1_p2 / cox1_p3 / nad1_p1 / nad1_p2 / nad1_p3 / h3_p1 / h3_p2 / h3_p3/ 12S/ 28S), hereafter called P1 and P2, respectively. The ML and BI analysis were performed based on the result indicated by PartitionFinder2.

Sequences of 73 specimens were obtained (Table S1): cox1 (71 sequences), nad1 (69 sequences), 12S (73 sequences), h3 (70 sequences) and h4 (71 sequences). Sequences of the 28S marker showed multiple overlapping peaks and were discarded from downstream analyses. The cox1 alignment was 1,278 bp long, nad1 at 615 bp, h3 at 384 bp, and h4 at 159 bp. The alignment of the 12S fragment resulted in 430 bp and 191 bp after being analyzed in Gblocks.

In this phylogenetic study, which included mitochondrial and nuclear genes, we recover the specimens considered to be representatives of the subfamily Allocosinae as a monophyletic group, and propose a new hypothesis for the phylogenetic relationships within the subfamily (Figs 2, 3, 4). Although the monophyly of the subfamily was previously reported by Piacentini and Ramírez (2019), Gonnet et al. (2021), and Laborda et al. (2022), this study included a more comprehensive taxonomic sampling, especially for South American species (13) (Fig. 1). It should be noted that the taxonomy currently does not reflect the monophyly of Allocosinae. Many species currently classified in Allocosa are not related to the type species Allocosa funerea or to other genera in Allocosinae, and are therefore incorrectly placed in the subfamily. Allocosinae also contain a member of a genus currently classified in Tricassinae (“Arctosa” sapiranga). Until these nomenclatural problems are resolved, the subfamily cannot be considered truly monophyletic.

The resulting tree placed Tricassinae + Hippasinae as a sister group to Allocosinae, consistent with findings by Gonnet et al. (2021), but differing from Piacentini and Ramírez (2019), who found Lycosinae + Pardosinae as sister group to Allocosinae. The remaining internal relationships within Lycosidae, specifically between the other subfamilies, exhibited differences from previous studies (Piacentini and Ramírez 2019; Gonnet et al. 2021). However, these variations occurred in unsupported groups, and further data will be required to confirm those relationships.

Piacentini and Ramírez (2019) included “Arctosa” sapiranga within Allocosinae, positioning it as a sister group to the rest of the subfamily. They also noted the morphological differences between “Arctosa” sapiranga and other Allocosinae species, sharing few morphological characters such as the presence of a long, triangular hyaline conductor (Piacentini, pers. obs.). Our phylogenetic analysis confirmed the position of “Arctosa” sapiranga. A similar pattern was recovered for Gen. 3 sp.1, which, while morphologically distinct from other Allocosinae species, shares specific characteristics, particularly in genital morphology with “Arctosa” sapiranga (Piacentini and Laborda, pers. obs.). Recently, Paredes-Munguia et al. (2024) reviewed the Neotropical species of Arctosa, including “Arctosa” sapiranga, and highlighted the challenge of assigning these species to a specific subfamily, as they exhibit traits of both Tricassinae and Allocosinae. However, due to a greater number of diagnostic elements, they leaned toward placement in Tricassinae. The characters shared by “Arctosa” sapiranga with Tricassinae may be plesiomorphic. This is supported by the results obtained by Piacentini and Ramírez (2019), who recovered Arctosa as a polyphyletic group. This seems to indicate that it is a “tailor’s drawer” genus lacking real synapomorphies that group the species. According to these authors’ results, the genus would be placed in Tricassinae given the position in which they recovered Arctosa cinerea (the type species of the genus). Therefore, those species recovered in other subfamilies, such as “Arctosa” sapiranga, would be incorrectly placed in the genus.

Our results supported Gen. 2 sp. 1 (aff. Allocosa panamena) as the sister group to the remaining Allocosinae, except “Arctosa” sapiranga and Gen. 3 sp.1 (Figs 2, 3). Specimens of this group were collected from a sandy streambank in a jungle environment on the Colombian Caribbean coast. This habitat and distribution align with what Dondale and Redner (1983) reported for Allocosa panamena Chamberlin, 1925. It is also morphologically similar, being probably congeneric with A. panamena but not conspecific, since it has differences in the genital structures that suggest it is an undescribed species. Gen. 2 sp. 1 (aff. Allocosa panamena) shares traits with A. funerea (and other North American Allocosa species), such as small body size and a glabrous, shiny prosoma. Phylogenetic analyses did not reflect these morphological affinities, as Allocosa funerea was not closely related to Gen. 2 sp. 1 (aff. Allocosa panamena).

The undescribed Gen. 1 sp. 1 consists of four specimens from Bahia, Brazil, specifically from sandy coastal environments, dune fields and associated vegetation. This species was included in the size dimorphism analysis by Aisenberg et al. (2023) (referred to in this study as Allocosinae sp. 6 ‘Bahia’), where it was reported that females in this species are larger than males. Another species (not included in the analysis) also found in Northern Brazil showed similar morphological characteristics of genitalia, suggesting that it is congeneric with Gen. 1 sp. 1, and would be an undescribed genus with two species.

Representatives of the genus Abaycosa were supported as monophyletic, in the same position obtained by Piacentini and Ramírez (2019) (as Pardosa flammula) and Laborda et al. (2022). The two known species of the genus A. nanica and A. paraguensis are included in the analysis. These are small species distributed in the central area of South America and are very abundant in anthropic environments Laborda et al. (2022).

The genus Allocosa was also recovered as monophyletic in the tree topologies generated by ML and BI analyses; however, it showed support only in the ML analysis under a gene-based partition scheme (P1) (Figs 2, 3). The nomenclatural status of Allocosa is tied to the inclusion of Allocosa funerea, the type species of the genus. The position of A. funerea in our analysis is congruent with the findings of Piacentini and Ramírez (2019), who positioned it as the sister group to Allocosa senex + Gnatholycosa spinipalpis. In an additional analysis (Fig. S1), a monophyletic group composed of A. funerea, A. noctuabunda, A. parva, and A. absoluta was recovered in the same position as A. funerea in other trees, as a sister group to the South American Allocosa species. Although this analysis used only the cox1 molecular marker available for A. noctuabunda, A. parva, and A. absoluta, it supports a group of North American Allocosa species. For these reasons, the boundaries of Allocosa genus are not entirely clear, as it may constitute a single genus or potentially three distinct ones. Clades A and B could indeed represent different genera than Allocosa. Based on the taxonomic history of the species they include, the names Glieschiella for clade A and Gnatholycosa for clade B are available for naming these putative genera (see World Spider Catalog 2025). Allocosa senex, A. alticeps, A. marindia, and Allocosa sp. 1 are recovered, forming clade A. The species in this group, studied by Simó et al. (2017), display distinct morphological and ecological traits. They are burrowing species adapted to sandy substrates, with specialized macrosetae on their pedipalps, structures involved in burrow digging, as well as well-developed spinnerets that they use to cover their burrows with silk (Aisenberg et al. 2010; Albín et al. 2018; Foelix et al. 2017). Allocosa senex is recovered in the topology as paraphyletic, due to the position of A. marindia (Figs 2, 3). It could be interpreted that these two species are synonyms, however, there is a large amount of independent evidence that points to morphological, ecological and behavioral differences between these two species (Aisenberg and Costa 2008; Aisenberg et al. 2010; Aisenberg and González 2011; Aisenberg et al. 2011; Bidegaray-Batista et al. 2017; Simó et al. 2017; Cavassa et al. 2022). A possible scenario is that the specimens included in the analysis and identified as A. senex belong to more than one species. A supported group includes specimens from Brazil, Uruguay, and Argentina (Figs 3, 4), with two individuals (AL27 and AL28) collected very near the type locality of the species (El Palmar National Park in Entre Ríos), which could be considered Allocosa senex s.str. The remaining specimens of A. senex could constitute different species. For example, another group of A. senex, comprising specimens from Chile and Southwestern Argentina, was recovered without support in the topology (Figs 3, 4A). These specimens were analyzed separately from A. senex in Aisenberg et al. (2023) (Chilean specimens: Allocosinae sp. 3 ‘Coquimbo’; Southwestern Argentina specimens: Allocosinae sp. 4 ‘Lanin’), in which atypical sexual size dimorphism and sex roles were reported. Clade B is also monophyletic and sister clade to clade A. It comprises specimens from Gnatholycosa spinipalpis, Allocosa yurae, and “Paratrochosina” amica. Species in this group have a broad, elevated head region and prominent chelicerae, especially in males. G. spinipalpis and A. yurae are associated with foothill habitats in Argentina and Chile (Mello-Leitão 1940; Brescovit and Taucare-Ríos 2013), whereas “P”. amica is a generalist species found in grasslands and widely distributed across Southern South America (Gonnet et al. 2021).

Species delimitation analyses identified 24 independent evolutionary lineages, about twice the number of morphology-based lineages. It should be noted that these methods delimit the population structure, so the different lineages found do not necessarily correspond to different species, and independent evidence is important to consider. Within the species Allocosa senex, five lineages were identified. As already noted, a recovered lineage, which includes specimens collected near the type locality, is considered to be Allocosa senex s.str. The remaining lineages have a wide distribution in South America. Some of these lineages, as a group comprising specimens from Chile and Southwestern Argentina (Figs 3, 4A), could be a different species to A. senex due to the great geographical distance and marked environmental differences. However, they are not morphologically distinct enough to consider these groups as distinct species. Species delimitation analyses suggest that they could be overlooked species, but further phylogeographical, ecological and behavioral studies are needed to test this hypothesis. Three lineages were identified within Allocosa marindia, two in Uruguay and one in Brazil, indicating a distribution along the Atlantic coast extending to São Paulo, Brazil. However, the morphological data do not support these as distinct species. Allocosa sp. 1 was recovered with support, including two specimens from Bahia, Brazil, which were collected in a dune environment and that have the characteristic morphological adaptations to the sandy habitats of clade A. This species was referenced as Allocosa cf. senex in Aisenberg et al. (2023), where it was reported to exhibit sexual size dimorphism reversal, suggesting possible sex role reversal. In “Paratrochosina” amica, five lineages were recognized. In this widely distributed species, specimens from distant locations formed distinct lineages; two specimens from Río Clarillo National Park in Chile were recovered as a sister group to the remaining specimens. In turn, specimens from Pró-Mata, Brazil formed a sister group to a lineage that includes specimens from Uruguay and Argentina. This latter lineage includes specimens collected near the type locality in Córdoba, Argentina, which may represent “Paratrochosina” amica s.str. Finally, two lineages were recognized in Gen. 1 sp. 1. These specimens do not exhibit notable morphological differences and were collected in similar environments at nearby locations in Brazil. However, sympatry of morphologically similar Allocosinae species has been reported in other regions (Aisenberg and Costa 2008; Laborda et al. 2022), so the presence of two lineages remains a possibility, and population-level analysis may reveal overlooked diversity.

The two approaches used to estimate divergence times, the concatenated matrix and the species tree, differed slightly, with estimates from the concatenated gene tree being slightly older. However, the confidence intervals for node age estimates are relatively broad, and in some cases, they display areas of overlap (Fig. 4 A and B). This aligns with the findings of McCormack et al. (2011), who noted that concatenated tree approach methods can significantly overestimate divergence times because they do not account for genetic divergence before speciation.

The diversification of the subfamily begins in the Oligocene and Miocene, but many clades diversified more recently in time, in the Pliocene and Pleistocene. For example, within clade A, the divergence of Allocosa alticeps from A. senex + A. marindia was estimated at 7 Ma and 3 Ma (according to the concatenated matrix and species tree, respectively). The estimate from the species tree aligns with Postiglioni et al. (2019), who placed this divergence at 2.45 Ma. Similarly, the sister species A. senex and A. marindia were estimated to have diverged at 2.5 Ma and 0.9 Ma (according to the concatenated matrix and species tree, respectively). The species tree estimates agree with Bidegaray-Batista et al. (2017) and Postiglioni et al. (2019), who indicated that this divergence occurred less than one million years ago. The diversification of clade A in the Pleistocene may be related to intense climatic fluctuations and geological changes that contributed to the modification of the landscape of the area where the species inhabit. Some of these changes include the increase in the areas of sandbanks, dunes and other sandy soil environments (Clapperton 1993; Carignano 1999; Mon and Gutierrez 2009; Nascimento et al. 2013; Turchetto-Zolet et al. 2013). Species in group A can be considered specialists in this type of environment, since they only inhabit sandy soil systems and have morphological, ecological and physiological adaptations to live there (Costa 1995; Costa et al. 2006; Aisenberg et al. 2010; Foelix et al. 2017; Simó et al. 2017; Albín et al. 2018). Therefore, it is expected that they have expanded and diversified in the subcontinent due to the expansion of their habitable environments.

Our phylogenetic framework provides insights into the diversification and taxonomy of South American Allocosinae. While some lineages, such as Abaycosa and Gen. 1, were consistently recovered with strong support across all analyses, others, particularly Allocosa and its internal structure, are ambiguously supported. Consequently, future studies will have to focus on resolving these specific groups. Future analyses will have to include a better representation of the North American fauna, both in terms of species and genes, to assess the stability of the genus Allocosa and to better define the boundaries and members of this group. Given the recent divergence times observed within the Clades A and B and the reported independent evolutionary lineages, analyzing thousands of informative molecular markers, such as single nucleotide polymorphisms (SNPs), together with ecological, morphometric, and behavioral studies will be necessary to infer species boundaries.

Here we present a phylogenetic hypothesis of the subfamily with a broad sampling of internal taxa. The main lineages that comprise Allocosinae are recognized, confirming the position of previously recognized species in the subfamily but also showing a still unknown diversity with possible new genera and species. The results presented here will be a starting point for future taxonomic contributions and an evolutionary frame of reference for ecological, genetics and behavioral studies that are already being developed in this group.