Two new Brazilian species of Loxosceles Heinecken & Lowe, 1832 with remarks on amazonica and rufescens groups (Araneae, Sicariidae)

Abstract The genus Loxosceles Heinecken & Lowe, 1832 has 91 representatives in the New World. Despite medical relevancy, the taxonomy of the genus is poorly understood. South American Loxosceles were divided into four groups of species: laeta, spadicea, gaucho and amazonica; this last one has a single species, Loxosceles amazonica Gertsch, 1967. More recently, the natural occurrence of L. amazonica in the New World has been questioned, due to the strong morphological resemblance and close phylogenetic relationship with Old World species, mainly with Loxosceles rufescens (Dufour, 1820). Herein, L. amazonica is rediagnosed and its morphological variation and natural distribution discussed. Two new species closely related to it from northeastern Brazil are also described, Loxosceles willianilsoni sp. n., from the state of Rio Grande do Norte, and Loxosceles muriciensis sp. n., from the state of Alagoas. The relationships of these new species with L. amazonica and L. rufescens are discussed.


Introduction
Loxosceles Heinecken & Lowe, 1832 is a speciose spider genus with a core distribution in the New World (World Spider Catalog 2016). Several species are known also from Africa, Middle East, Mediterranean Europe and two species from China were recently described (World Spider Catalog 2016). Many species were reported as causing bites of importance to human health and several studies on their venom have been published (Gertsch 1967, Tambourgi et al. 2000, Isbister and Fan 2011. Despite this, the taxonomy of the genus is poorly understood. The most comprehensive works were done by Gertsch (1958Gertsch ( , 1967 and Gertsch and Ennik (1983) who revised New World species. After these revisions, other species were sporadically described and more recently the African, Middle East and Asian species received more attention (Binford et al. 2008, Duncan et al. 2010, Lotz 2012, Planas and Ribera 2015, Wang 1994. The South American Loxosceles were revised by Gertsch (1967), who created four groups of species: laeta with 26 species, spadicea with three species, gaucho with six species and amazonica with a single species. Loxosceles amazonica Gertsch, 1967 has been recorded from localities in the Amazon in Brazil, and Peru to northeastern Brazil. More recently, the natural distribution in the New World has been questioned, due to the strong morphological resemblance to the Old World species, mainly with Loxosceles rufescens (Dufour, 1820) (Binford et al. 2008;Duncan et al. 2010). Molecular analyses has also retrieved L. amazonica to be closely related to the Old World species (Binford et al. 2008;Duncan et al. 2010), therefore, L. amazonica origin and its relationship is still up for debate.
Herein, we describe two new species closely related to L. amazonica from northeastern Brazil. The relationship of these new species with L. amazonica and L. rufescens is discussed.

Materials and methods
The general format of the description follows Gertsch (1967). All measurements are in millimeters. Measurements of the legs and palp were taken from the dorsal aspect of the left side (unless appendages were lost or obviously regenerated) with a Mitutoyo ® digital caliper, which had an error of 0.005 mm, rounded up to two significant decimals. Structures from the left side of the specimens were chosen for descriptions. When using structures from the right side, the figures were mirrored to show them as coming from the left side and allowing easy comparison. The copulatory organs of females were dissected and submitted to digestion by a commercial protein remover for contact lenses (with pancreatin) during some minutes in order to observe the internal structure; when necessary, they were also cleared with clove oil. A Leica LAS Montage and LAS 3D module mounted on a Leica M205C dissecting microscope were used for image capture and measurements of other spider structures.
The examined specimens are deposited at MNRJ, Museu Nacional, Rio de Janeiro, and AMNH, American Museum of Natural History, New York. Geographical coordinates are denoted as primary sources between round brackets, secondary sources (Google Earth) between square brackets. The coordinates from the secondary source were obtained from the center of the municipality cited in the specimen label and are in DMS (Degrees, Minutes and Seconds) format rounded off to minutes. Maps were made with SimpleMappr, an online tool used to produce maps (Shorthouse 2010). Gertsch, 1967 Figs 1−51, 78−79 Loxosceles amazonica Gertsch, 1967: 143, pl. 4, figs 7−10, pl. 5    Specimens of L. amazonica were also found in Martins, state of Rio Grande do Norte, "a brejo de altitude" region, i.e. an area covered by humid forest surrounded by arid caatinga (Pereira Filho and Montingelli 2011), usually over mountains and hillsides with an elevation of more than 500 m (Ruiz-Esparza 2009) and that receives more than 1,200 mm of orographic rains (Prado 2003, in Ruiz-Esparza 2009). We found specimens of L. amazonica in ravines near the town (Fig. 48), in a trail on the top on the hill (Fig. 49) and under old house debris close to more humid and higher areas (about 700 m a.s.l.) (Fig. 50), as well as under rocks and tree bark near Casa de Pedra cave, in a lower region with caatinga vegetation (about 300 m a.s.l.) (Fig. 51). No specimens were found inside Casa de Pedra cave.

Loxosceles amazonica
Spermatheca variation (see Fig. 79). Specimens vary in number and size of globular lobes on spermatheca apex and seminal receptacles proportions. Specimens from Martins and Macaíba in the State of Rio Grande do Norte (Figs 26 and 31, respectively), São Raimundo Nonato, state of Piauí (Fig. 27) and Santa Quitéria, state of Ceará ( Fig. 30) have three to six lobes in each spermatheca, more or less similar in size. The seminal receptacles of specimens of these areas are slightly short and trapezoid. On the other hand, specimens of ESEC Seridó and FLONA de Açu, both in the state of Rio Grande do Norte (Figs 28 and 29, respectively) have four to five lobes, usually one of them larger than the others. The seminal receptacles are slightly longer, with a triangular shape.
It is not clear how these genitalic traits vary along the distribution of L. amazonica or if these variations reflect a higher diversity in amazonica lineage. Variation in the morphology of palps and spermatheca of other Loxosceles species has already been noted, such as in L. rufescens (Brignoli 1969 1.11,9.88,10.95,1.85,32.08;III: 6.40,1.09,6.23,7.64,1.30,22.66;IV: 7.12,1.05,7.08,8.38,1.52,26.15. Palp: femur 1.46 long, 0.31 wide; patella 0.49 long, 0.33 wide; tibia 0.88 long, 0.48 wide; cymbium 0.43 long, 0.42 wide. Labium 0.71 long, 0.38 wide. Sternum 1.78 long, 1.50 wide. Femur I 2.4 times as long, tibia I 2.7 times as long and leg I 8.7 as long as carapace. Palpal femur four times longer than wide, tibia 1.8 times longer than wide, cymbium oval (Fig. 54). Bulb suboval and approximately same size as cymbium. Embolus straight, with a strong curvature on apex, approximately 1.3 times longer than bulb length in retrolateral view, without carina (Fig. 53). Cephalic region of carapace covered by many long setae (Fig. 52). Entire pars cephalica as well as carapace border dark brown (Fig. 52). Legs and palps light brown, covered by short greyish setae on the femora and patellae (Fig. 64) Femur I 1.8 times as long, tibia I 2.0 times as long and leg I 6.5 as long as carapace. Palpal femur 4.7 times longer than wide, tibia 3.5 longer than wide, tarsus not incrassate. Spermathecae with enlarged seminal receptacles; without transversal plate; and presence of dark sclerotized lateral bands almost reaching the apex (Fig. 57). Palps pale brown, except by darker tibiae and metatarsi. Endites pale brown.
Etymology. This species is named after the biology student Willianilson Pessoa, in honor of his friendship and support during expeditions in Rio Grande do Norte. This name is masculine in gender.
Natural history. Specimens were found inside Casa de Pedra cave walking on walls, in webs inside wall cracks or under loose stones on the cave ground. This calcarian cave is very large regarding regional patterns and has turistic use (Ferreira et al. 2010). Apparently, specimens of L. willianilsoni sp. n. are found only inside the cave.
Etymology. The specific name refers to the type locality, Estação Ecológica de Murici, state of Alagoas, Brazil and is neutral in gender.
Natural history. The few specimens of L muriciensis sp. n. were found inside a burnt tree in an Atlantic rainforest conservation unit in the state of Alagoas. The ESEC Murici is one of the last and largest remnants of the northeastern Atlantic rainforest and it is inserted in a biodiversity hotspot known as the "Pernambuco Endemism Center" (Nemésio and Santos Junior 2014).  Gertsch (1967) and Silveira (2015). 79 Expanded map showing the records of the illustrated specimens of L. amazonica, L. willianilsoni sp. n. and L. muriciensis sp. n.

Discussion
In his revision of the South American Loxosceles species, Gertsch (1967) proposed four species groups for the thirty species he recognized. The only group with a single species is amazonica with the species L. amazonica described in the same paper (Gertsch 1967). This author approximated L. amazonica to the gaucho group due to the carapace marked with dark lateral bands and some incrassated segments of male palps. On the other hand, the presence of spermathecae with free receptacles with rounded lobes, not closely tied by a transverse band, resembles laeta species (Gertsch 1967). Despite L. amazonica having characteristics of both South American groups gaucho and laeta, in some genitalic features it closely resembles species of the rufescens group from the Paleartic fauna (Gertsch 1967). Due to these special characteristics, L. amazonica was considered to have group status by Gertsch (1967). After Gertsch's revision (1967), only scattered descriptions of new species of Loxosceles were published. A more embracing work was done by Binford et al. (2008), which proposed the first phylogenetic relationship hypothesis concerning representative Loxosceles species. In that work, besides morphological similarity, a molecular proximity was detected between L. amazonica and L. rufescens (Binford et al. 2008). The ubiquitous species L. rufescens, associated or not with the Chinese species L. lacta, was presented as the sister-group of L. amazonica in analyses with different types and combinations of  Planas and Ribera (2015). Legs differentiated by less than 0.5 mm are in bold. * = Legs 2 and 4, and legs 4 and 1 have difference in length less than 0.5 mm. AL = state of Alagoas, CE = state of Ceará, MT = state of Mato Grosso, PI = state of Piauí, RN = state of Rio Grande do Norte. was done by Duncan et al. (2010). Once again, L. amazonica was recovered in the clade including the northwestern African Loxosceles species. However, there was no agreement that L. amazonica was the sister-group of the monophyletic L. rufescens lineage nor the basal taxa of the northwestern African clade. The lack of resolution inside the northwestern African clade, the existence of African male specimens very similar morphologically to L. amazonica and the fact that the most recent common ancestor of L. amazonica and L. rufescens was found by Binford et al. (2008) to be too young to be explained by Gondwanan vicariance were considered by Duncan et al. (2010) to indicate that L. amazonica is derived from within northwest Africa Loxosceles and dispersed recently from one continent to other. They proposed that the split of the continents did not influence the distribution of the common ancestor L. amazonica and L. rufescens (Duncan et al. 2010). They considered L. amazonica as a species that can be easily introduced by human transport and suggested the trade between Brazil and Africa in 16 th century could explain the dispersal of L. amazonica from Africa to South America (Duncan et al. 2010). They also considered the absence of other species related to L. amazonica in South America as further evidence supporting an African origin of this species.
The discovery of two new species, herein described, closely related to L. amazonica in northwestern Brazil, throw a new light on this discussion. It is very unlikely that L. amazonica came from Africa about 500 years ago and in so little time speciated into two more different species. Another point that contradicts the argument that L. amazonica was introduced in South America is the large distribution of the species (Fig. 78). It is very improbable that such a reclusive spider would disperse to many natural localities throughout northwestern Brazil in such a short period of time, reaching remote localities in central western Brazil such as the type locality, an indigenous village difficult to access even nowadays. Furthermore, specimens of L. amazonica as well specimens of L. willianilsoni sp. n. and L. muriciensis sp. n. were found in natural environments 66) inside and outside four Conservation Units in three Brazilian states. Moreover, if L. amazonica is an invasive species as proposed by Duncan et al. (2010), their presence in larger cities in southeastern and southern Brazil would also be expected, as invasive species are normally introduced by means of human activities and benefited by urban environments, normally forming large populations. Even though they can be found in disturbed environments in northwestern Brazil, they are found in natural conditions and are not found in urban areas in localities more to the South.
The question on the origin of L. amazonica and L. rufescens lineages is, therefore, open to discussion. A way to test the origin and evolution of L. amazonica lineage would be to collect L. amazonica specimens from different parts of northern, northwestern and central western Brazil as well as other South American countries, and determine the genetic divergence among the different populations.
As demonstrated by Duncan et al. (2010), the amazonica group is recovered in the middle of rufescens lineage. Therefore, it makes no sense to use the group name amazonica, and L. amazonica, L. willianilsoni sp. n. and L. muriciensis sp. n. should be referred as belonging to rufescens group.