Revision, cladistic analysis and biogeography of Typhochlaena C. L. Koch, 1850, Pachistopelma Pocock, 1901 and Iridopelma Pocock, 1901 (Araneae, Theraphosidae, Aviculariinae)

Abstract Three aviculariine genera endemic to Brazil are revised. Typhochlaena C. L. Koch, 1850 is resurrected, including five species; Pachistopelma Pocock, 1901 includes two species; and Iridopelma Pocock, 1901, six species. Nine species are newly described: Typhochlaena amma sp. n., Typhochlaena costae sp. n., Typhochlaena curumim sp. n., Typhochlaena paschoali sp. n., Pachistopelma bromelicola sp. n., Iridopelma katiae sp. n., Iridopelma marcoi sp. n., Iridopelma oliveirai sp. n. and Iridopelma vanini sp. n. Three new synonymies are established: Avicularia pulchra Mello-Leitão, 1933 and Avicularia recifiensis Struchen & Brändle, 1996 are junior synonyms of Pachistopelma rufonigrum Pocock, 1901 syn. n., and Avicularia palmicola Mello-Leitão, 1945 is a junior synonym of Iridopelma hirsutum Pocock, 1901 syn. n. Pachistopelma concolor Caporiacco, 1947 is transferred to Tapinauchenius Ausserer, 1871, making the new combination Tapinauchenius concolor (Caporiacco, 1947) comb. n. Lectotypes are newly designed for Pachistopelma rufonigrum Pocock, 1901 , Iridopelma hirsutum Pocock, 1901 and Pachistopelma concolor Caporiacco, 1947. Cladistic analyses using both equal and implied weights were carried out with a matrix comprising 62 characters and 38 terminal taxa. The chosen cladogram found with X-Pee-Wee and concavity 6 suggests they are monophyletic. All species are keyed and mapped and information on species habitat and area cladograms are presented. Discussion on biogeography and conservation is provided.


Introduction
Theraphosidae is the richest mygalomorph family, comprising ca. 1/3 of the 2,693 known mygalomorph species (Platnick 2012). The New World is particularly rich in theraphosid species (Platnick 2012), comprising representatives of three subfamilies: Ischnocolinae, Theraphosinae and Aviculariinae which comprises more than a half of the 932 known theraphosid species. Despite their potential importance as top predators in ecological webs (Sergio et al. 2008), the pet trade (e. g. Yánez andFloater 2000, Bertani andFukushima 2009) and a source of important tools for pharmacological research and therapeutical leads (e. g. Escoubas and Rash 2004), theraphosids have been largely neglected in taxonomical and biological research. Only in the past 30 years have they received more attention from the scientific community in some taxonomic revisions and cladistic analyses (e. g. Raven 1985, Pérez-Miles et al. 1996, Bertani 2001, West et al. 2008, West et al. 2012, molecular analyses (e. g. Hamilton et al. 2011), biochemical and pharmaceutical (e. g. Silva et al. 2000, Bode et al. 2001, Escoubas and Rash 2004, ethological and ecological studies (e. g. Marshall and Uetz 1990, Bertani and Marques 1996, Yánez and Floater 2000 that have displayed a fascinating diversity in all these aspects. This work focuses on those aviculariines occurring mainly in Northeastern Brazil from where the type species of aviculariine genera Typhochlaena C. L. Koch, 1850, Pachistopelma Pocock, 1901and Iridopelma Pocock, 1901 were described. Within all aviculariine genera presently accepted (West et al. 2008), (Avicularia Lamarck, 1818, Iridopelma, Pachistopelma, Ephebopus Simon, 1892, Tapinauchenius Ausserer, 1871, Psalmopoeus Pocock, 1895from the New World -and Stromatopelma Karsch, 1881 and Heteroscodra Pocock, 1889 -from Africa), these three are the least understood both taxonomically and biologically.
Typhochlaena and Iridopelma share a history of synonymies and revalidations, whereas Pachistopelma has lacked important taxonomic change since its description. Typhochlaena was erected by C. L. Koch (1850) for his species Mygale seladonia C. L. Koch, 1841 from Bahia, Brazil, and Mygale caesia C. L. Koch, 1842, from Puerto Rico. Only supposed females were described and the diagnosis for the genus was very short, with references to the highly colored abdomen and well developed scopulae on the legs. Karsch (1879) described Typhochlaena magdalenae Karsch, 1879, male, from Santha-Martha, Colombia, but did not discuss the genus diagnosis. Simon (1892) synonymized Typhochlaena with Avicularia, and considered the small species with mottled abdomen to belong in a distinct group of Avicularia species: A. seladonia, A. caesia, A. magdalenae and A. glauca Simon, 1891. Simon (1892 recognized only two genera in Avicularieae: Tapinauchenius and Avicularia. However, Pocock (1901) considered that Psalmopoeus and Ephebopus should also be included in Si-mon's Avicularieae Simon, 1892 to which he also added two new genera, Pachistopelma Pocock, 1901 andIridopelma Pocock, 1901. Pachistopelma was described utilizing the male and female of a single species, P. rufonigrum Pocock, 1901from Igarassu, Brazil. Pocock (1901 considered it closely related to Avicularia due to a single spinose spur on the tibia of first leg in males, marginal sternal sigilla, absence of spines and development of scopulae, and differentiated it by the slightly procurved first ocular row in Pachistopelma. Iridopelma Pocock, 1901 was described in the same paper also using the male and female of a single species, I. hirsutum Pocock, 1901, from Pernambuco, Brazil. Pocock (1901 also related this genus with Avicularia, but differentiated it by the presence in Iridopelma of a spur on the tibia of leg II in males and leg I being longer than leg IV. Two years later, Simon (1903) synonymized Iridopelma with Avicularia, by considering the presence of a tibial spur on leg II of the male to be of little generic significance and the females lack characteristics that allow generic recognition. Mello-Leitão (1923) disagreed with Simon (1903) and considered it possible to identify females of Iridopelma by characters given by Pocock (1901), i. e. leg I longer than IV and size of the median anterior eyes. However, he received a male from the state of Bahia, Brazil with tibial spurs on legs I and II and identified it, with doubts, as Typhochlaena seladonia, thus synonymizing Iridopelma with Typhochlaena (Mello-Leitão 1923). He stated that, in the case of this specimen not being congeneric with T. seladonia, it should be considered a new Iridopelma species, named Iridopelma amphidorotherion Mello-Leitão, 1923. The generic description given by him for the genus Iridopelma should be seen as the description of that species (Mello-Leitão 1923). In the same paper (and page) it was given a new name, Typhochlaena pococki Mello-Leitão, 1923 for Iridopelma hirsutum Pocock, 1901, without further explanation. Another species, Typhochlaena zorodes Mello-Leitão, 1926 was described three years later from a male from the state of Bahia, Brazil, and distinguished from other species by the uniform abdominal coloration, cephalothorax as wide as long and anterior eyes of equal size (Mello-Leitão 1926). Roewer's (1942) Catalog listed both Iridopelma and Typhochlaena as junior-synonyms of Avicularia. Raven (1985) revalidated Iridopelma, and Smith (1993) redescribed the types of Iridopelma hirsutum and transferred from Avicularia Iridopelma zorodes (Mello-Leitão, 1926) and Iridopelma seladonium (C. L. Koch, 1841), making new combinations. Almeida- Silva et al. (2008) obtained photographs of the holotype and described the male of I. seladonium for the first time and retained the species in Iridopelma, even though the male lacks tibial spurs both on leg I and II, the diagnostic character of the genus (Pocock 1901, Raven 1985. Concerning Pachistopelma, Caporiacco (1947) described a new species, P. concolor Caporiacco, 1947, from Guyana. To date, this genus has remained unrevised.
Computer methods: A data matrix (Table 1) with 62 characters and 38 taxa was analyzed with NONA 2.0 for Windows (Goloboff 1998) and X-PEE-WEE 1.3 for Windows (Goloboff 1997). The commands h1000, h/20, amb-and mult*50 were used. Concavities 1 to 6 were used with PEE-WEE. All characters were treated as nonadditive. This is discussed fully below in cladistics.    Diagnosis. Differs from all other aviculariine genera by the domed, short distal segment of PLS (Fig. 21) and adults with sternum as long as wide, truncated behind (Fig. 20). Additionally, they are very small aviculariines having urticating hair type II on the dorsum of the abdomen and males lack both tibial spurs and spiniform process on the cymbium.
Distribution and habitat. Brazil: Northeastern, part of Central-West (state of Tocantins) and part of Southeastern (state of Espirito Santo) (Fig. 28). Specimens of Typhochlaena spp. are mostly found in Brazilian Atlantic rainforest, but one species seems to occur in drier, open environments (T. costae sp. n.). Available data for two species (T. seladonia and T. curumim sp. n.) indicates they build a silky retreat under loose tree bark (Fig. 23).

Key to species of Typhochlaena
Remarks. Almeida-Silva et al. (2008) redescribed the female of Iridopelma seladonium (C. L. Koch, 1841) after examining photographs of the recently rediscovered holotype of Mygale seladonia C. L. Koch, 1841. They presented photographs of the holotype and an illustration displaying non-spiraled spermathecae from a fresh specimen from state of Bahia, Brazil. A female exuvium examined by myself in this work showed spermathecae with two folds (Fig. 4) and a dissected female showed a single fold spermathecae (Fig. 5). Spermathecae of exuvium could have been modified by the moulting process or by preservation conditions, but the dissected specimen is well preserved. The fold is not easily seen and possibly it was overlooked by Almeida- Silva et al. (2008). The unavailability of additional specimens prevented me from investigating this question more thoroughly.
Distribution. Brazil, state of Sergipe and northeastern Bahia (Fig. 28). Natural history. One female was found under loose tree bark in Santa Luzia do Itanhy SE (September 1999) Fig. 23. Two females were found in a similar retreat in nature, in Bahia, and one specimen built a retreat with small pieces of tree bark in captivity (Almeida- Silva et al. 2008).
Etymology. The specific name is derived from the Brazilian indigenous Tupi language, meaning "child". It refers to the local children that found the type specimens high in a tree in Areia, State of Paraíba, Brazil, during an arachnological expedition.
Natural history. Three specimens were found high in a tree, under loose bark in "Mata do Pau-Ferro" reserve, September, 1999.
Etymology. The specific name is a patronym in honour of Elbano Paschoal de Figueiredo Moraes, a Brazilian environmentalist who was died early on April 2011. He was one of the founders of the NGO "GAMBA -Grupo Ambientalista da Bahia", and was well known for his efforts in preserving Brazilian Atlantic rainforest remnants in the state of Bahia, Brazil.
Distribution. Known only from Santa Teresa and Domingos Martins, in the mountain range of state of Espirito Santo, Brazil (Fig. 28). Diagnosis. Males differ from those of other species by the short and slender embolus . Females differ by the non-spiraled spermathecae, lacking lobes, diverging on their base and converging on their distal portions (Fig. 15). Additionally, males and females have cephalothorax brown and abdomen dorsally black with two series of four red spots extending laterally. Carapace, dorsum of chelicerae and most anterior dorsal region of abdomen with very long, yellow, stiff setae (Fig. 25).
Etymology. The specific name is a matronym in honour of Miriam Costa, who collected the holotype and several other new spider species during several years she worked at Instituto Butantan.
Distribution. Brazil, states of Tocantins, Maranhão and Piaui (Fig. 28). Natural history. No available data. The male was collected by pitfall trap, the female was taking during a faunal rescue in a flooded area, and the immature in fossil tree samples coming from states of Maranhão and Piauí border. Diagnosis. Male and female Pachistopelma differ from most other aviculariines, except Ephebopus, Tapinauchenius and Psalmopoeus by the straight to sligthly procurved first eye row (Fig. 37). Pachistopelma males differ from these genera by having a spinose spur on tibia I and females differ by having a dorso-ventrally flattened abdomen in combination with paired long spermathecae with a slight curvature medially and lacking constrictions or lobes. Additionally, males and females differ from these genera by both the absence of leg spines and the presence of urticating type II hair on the abdomen dorsum (except in mature females, that lack them).
Distribution and habitat. Northeastern Brazil, from the state of Rio Grande do Norte [6°22'S] to state of Bahia [13°25'S], mostly in the coastal region (Fig. 68). Both species of Pachistopelma inhabit tank bromeliads exclusively, e. g. Aechmea aquilega (Dias et al. 2000); Hohenbergia stellata, H. ridley (Dias et al. 2000); H. ramageana and A. lingulata ) (Figs 69-80) which can be found in very distinct habitats such as restinga (Figs 72-74), caatinga or even rainforest . In the last case, the spiders were in bromeliads that grow mainly on rocky outcrops exposed to direct and intense sunlight.
Color pattern ontogeny. Pachistopelma juveniles possess a metallic green general pattern, and the dorsum of abdomen having a black longitudinal central stripe and five lateral black stripes that may connect with the central stripe. Adults are brownish to pinkish or blackish, without abdominal pattern (52)(53)(54)(55)(56)(57).
Sexual dimorphism. Females and immatures have a very low cephalic region, when compared to the males, and the abdomen is dorso-ventrally flattened in the former (Fig. 34). The eye tubercle is very low in females (Fig. 36) and immatures, and the first ocular row is straight (Fig. 37). Males possess a more developed eye tubercle and the anterior ocular row is slightly procurved. Immatures and adult males have type II urticating hairs on abdomen dorsum, which becomes lost in adult females.
Distribution. Brazil, from state of Rio Grande do Norte southwards to state of Alagoas, mainly in coastal region (Fig. 68).
Natural history. All specimens with recorded field data indicate Pachistopelma rufonigrum individuals were found inside bromeliads (Bertani et al. 1994;Santos et al. 2002this work). Field observations suggest a strict dependency on bromeliads , both in restinga vegetation and in restinga associated with Brazilian Atlantic rainforest. Of these, bromeliads occurring in Atlantic rainforest areas had a low occupancy rate by Pachistopelma rufonigrum individuals than those bromeliads in xerophylous environments ). In Estação Ecologica Murici, Murici, state of Alagoas, Brazil, I observed several individuals exclusively inside tank bromeliads. These were 245-362 m a.s.l. in rocky outcrops marginated by Brazilian Atlantic forest. Bromeliads in this area form large patches in rocky outcrops where the soil is shallow and sunlight intense. The margin of these patches overlap with shaded areas contiguous with the forest, and, in agreement with Santos et al. 2004, specimens were found mainly in bromeliads more exposed to sunlight. The "bromeliad islands" in rocky outcrops are hundreds or thousands of meters apart, and no P. rufonigrum specimens were found outside of bromeliads. Thus, these populations are probably isolated. The species is abundant, ca. 30 specimens were collected, and others were observed but not collected. Individuals in all developmental stages were observed during this the period (August 2006). In agreement with Santos et al. 2004, spiders were observed diving into the phytotelma water, staying there for several minutes, when disturbed.
Color pattern ontogeny. Juveniles are almost completely metallic green, except for a pattern on dorsum of abdomen comprising a central longitudinal black stripe connected with five lateral black stripes (Fig. 40). In larger individuals carapace border and dorsum of chelicerae, coxae and trochantera are light brown. Dorsum of abdomen is light brown with a reddish area posteriorly. The black stripes remain (Fig. 41). In a next stage carapace is completely pink, as well as dorsum of coxae, trochantera and most femora. Remaining parts of legs retain the metalicgreen. The clear part of abdomen is now of a vivid red, and the black stripes remain (Fig. 42). Subadults have carapace and legs brown with sparse pinkish long hairs. Dorsum of abdomen is still reddish, but the black stripes begin to fade (Fig. 43). Adult female is completely brown with long pinkish setae on legs, carapace and chelicerae. Abdominal pattern is lacking or very inconspicuous (Fig. 44). Adult male carapace, chelicerae and legs are brown and covered by pinkish setae, except for the tarsi and a stripe on metatarsi that are black. Abdomen is a vivid orange/red, there is no vestige of any pattern (Fig. 45).
Distribution. Brazil: States of Sergipe and Northern State of Bahia, mainly in coastal regions (Fig. 68).
Sexual dimorphism. The cephalic region of female and immatures is very low in profile when compared with those of male, and abdomen is dorso-ventrally flattened in the former (Fig. 34). The eye tubercle is very low in female (Fig. 36) and immature, and first ocular row is straight (Fig. 37). Males have a more developed eye tubercle, the anterior ocular row is slightly procurved. Immatures and adult males have urticating hair type II on abdomen dorsally, which becomes lost in adult females.
Natural history. As with P. rufonigrum, all specimens examined and labeled with field data indicate they were found inside bromeliads, which agrees with field observations (Dias et al 2000;Dias andBrescovit 2003, 2004 - (Dias and Brescovit 2003), found in coastal region which is a typical enviroment for P. rufonigrum in Rio Grande do Norte state ). Bromeliad phytotelma is a source of water, food and retreat for a variety of animal species (Frank and Lounibos 2009), and in restinga regions they are a key resource for the local fauna (Santos et al. 2002;2003a, b). I failed to find Pachistopelma bromelicola sp. n. in parts of Parque Nacional de Itabaiana covered with Brazilian Atlantic rainforest. In RPPN Sapiranga, P. bromelicola sp. n. was found in restinga area, inside H. stellata bromeliads. Some of these bromeliads were very close to a house and were used in garden decoration. Inside two bromeliads we found an eggsac, protected by a retreat, in October 2007. In a region covered with Brazilian Atlantic rainforest, in RPPN Jequitiba, they were found close to a house located relatively far from forest shade. Other specimens were colected inside bromeliads in a caatinga (a xeric shrubland and thorn forest) region in Jeremoabo (R. A. Sanfilippo pers. comm.). Therefore, P. bromelicola sp. n. is distributed over contrasting environments, from rainforest to xeric caatinga and restinga. An element in common among these populations is the obligatory bromelicolous habits.
Color pattern ontogeny. The color pattern is similar to P. rufonigrum, mainly in early instars. However, the lateral black stripes of abdomen dorsum are almost always connecting with the longitudinal central stripe (Figs 52-53) whereas in P. rufonigrum they normally do not connect (Figs 41-43). Larger individuals have dark legs and dark-brown carapace, and abdominal pattern is conspicuous (Fig. 54). In subadults, abdomen is very dark, lacks a pattern, or is inconspicuous (Fig. 55). Adult female is almost completely black, except for some brown setae over carapace (Fig. 56). Adult male is also completely black, but with many long whitish setae on legs, carapace and chelicerae. Abdomen is black with long red setae (Fig. 57).

Type species. Iridopelma hirsutum Pocock, 1901 by original designation.
Diagnosis. Males of Iridopelma species differ from those of other aviculariines in the presence of tibial spurs on both leg I (Fig. 85) and II (Fig. 86). Females differ from those of all other aviculariines, except Avicularia and Typhochlaena, in that the anterior eye row is strongly procurved. Iridoplema further differs from Typhochlaena in that the most distal PLS segment is digitiform; from Avicularia in possessing spermatheca lacking an accentuated curvature medially.
Distribution and habitat. Brazil: Northeastern, from Reconcave region in state of Bahia northwards. Few records for states of Para and Tocantins (Fig. 169). Specimens of Iridopelma spp. are mostly found in Brazilian Atlantic rainforest (Figs 101-105). Some species occur in drier, open environments such as "cerrado" (Fig. 168), "caatinga" and "campo rupestre" (Figs 145-146). A single record is known from the Amazon. Natural history. Iridopelma hirsutum and I. zorodes build retreats connecting leaves with silk threads (Figs 101-105). Iridopelma katiae sp. n. make retreats inside bromelids (Figs 147-149). Habitat preference of other Iridopelma species is in need of confirmation.
Distribution. Brazil: From state of Ceará southwards to the state of Alagoas (Fig. 169). Spermathecae variation. Typical spermatheca diverges and tapers from base to apex, folds after its middle and has a strong constriction forming two distal lobes (Figs  87, 119, 121). More rarely, there is one lobe on both (Figs 88, 120) spermathecae or even three lobes (Fig. 122). Relative length and width vary, mainly in smaller specimens (Figs 119-122). However, even small spermathecae are already folded after their middle.
Natural history. Individuals of I. hirsutum are known to inhabit remnants of Brazilian Atlantic rainforest. Of eight specimens collected in Reserva Biológica de Saltinho, state of Pernambuco, and Estação Ecológica de Murici, state of Alagoas, six were found inside or close to retreats on leaves and two adult females were walking on the inner walls of an abandoned house (Fig. 106). They normally build a retreat with two leaves connected with silk threads (Figs 101, 102, 104-105) on different plants such as Piperaceae, Melastomataceae, Palmae (Fig. 101), "Guapeba" (Sapotaceae) (Fig. 105) and other not identified (Figs 102, 104). Ocasionally, they can be found with retreats in a single leaf, as in rolled Heliconiaceae leaf (Fig. 103). Retreat was above ground 0.5 to 1.0 m, but this was affected by collector's inability to search for specimens high in the vegetation. No adult males or females with eggsacs were found in the excursion period (August).
Color pattern ontogeny. Juveniles have most of the carapace, chelicerae and legs metallic green (Figs 95-96). Coxae, trochantera, basal femora dorsally and parts of the chelicerae are light brown. The abdomen is black with a light central longitudinal large spot having zigzag edges and a longitudinal central black stripe over a more reddish area. The sternum, coxae, labium and abdomen are ventrally black. As the juveniles mature, the metallic green coloration changes to brown (Fig. 97) and the central spot on dorsal abdomen becomes totally reddish (Fig. 98). In the adult females, parts of the abdominal pattern remain, the longitudinal central black stripe is broader, and the reddish area becomes inconspicuous (Fig. 99). Adult males also retain an abdominal pattern, and the reddish area is normally conspicuous (Fig. 100).
Remarks. Avicularia palmicola Mello-Leitão, 1945, holotype female from Brazil, Paraíba, Mumbaba, Aristóteles Silva, collected in coconut tree (MNRJ 2328), examined. The specimen is an aviculariine female with first eye row procurved and spermathecae having a single fold and bilobed at distal portion. These characters agree with Iridopelma and spermathecae shape is typical of I. hirsutum. Therefore, Avicularia palmicola Mello-Leitão, 1945 is transferred to Iridopelma and considered a juniorsynonym of Iridopelma hirsutum Pocock, 1901 syn. n. Diagnosis. The males resemble those of I. vanini sp. n. and I. katiae sp. n. by leg I and IV having similar length (leg IV/I length = 0.93, SD = 0.01). It differs from I. vanini sp. n. by the shorter embolus (Figs 107-109). It differs from I. katiae sp. n. by general brown pattern and lack of a reddish pattern on abdomen dorsally (Fig. 117). Females differ from all other species by two straight to slightly curved spermathecae tapering little from base to apex, not folding and with a very slight constriction forming a single distal lobe (Fig. 112).
Distribution. Brazil: states of Sergipe and Northern Bahia (Fig. 169). Spermathecae variation. Typical spermatheca is straight, or diverges little from its base; has no fold and apical portion is dilated (Figs 112, 123). Some spermathecae are shorter and taper considerably (Fig. 124), whereas others are long and curved distally (Fig. 126). The spermatheca of a specimen from Elísio Medrado, state of Bahia, is partially divided distally (Fig. 125). It is from the southermost distribution of the species and shows a different color pattern as well (Fig. 118).
Natural History. Individuals of I. zorodes inhabits remnants of Brazilian Atlantic rainforest. My own field observations in state of Sergipe (Areia Branca, Parque Nacional Serra de Itabaiana) and Santa Luzia do Itanhy (Mata do Crasto), indicate that individuals make retreats similar to those of I. hirsutum, i. e., leaves connected with silk threads. Eventually, they were found inside bromeliads. An adult male was found on a leaf in Camurujipe, São João da Mata, state of Bahia, in October.
Color pattern ontogeny. As I. hirsutum, but adult males and females lack or have an inconspicuous abdominal pattern (Figs 113-118).
Distribution. Brazil, states of Piaui, Maranhão and possibly Tocantins and Pará (Fig. 169). There is a single record for both Tocantins (Colinas de Tocantins) and Pará (Tocantins River). These specimens are smaller than those from Maranhão and Piauí, and the specimen from Pará has both male palpal bulb emboli broken, therefore its identification is only tentative.
Spermathecae variation. Typical spermatheca is long, converging, tapers strongly from base to apex, double folded and with a strong constriction forming two distal lobes (Figs 132, 151). In some specimens the apical region ends in a single lobe, that can be partially divided (Fig. 150) or in two lobes with one partially or completely divided, in this case resulting in multilobular spermathecae (Fig. 152).
Natural history. Little is known of this species. The holotype female was collected with an eggsac in November, under a fallen tree trunk in a moderately anthropized area in Parnaíba, state of Piaui. Other specimens were collected in Barreirinhas, State Diagnosis. Male resemble those of I. zorodes and I. vanini sp. n. in similar leg I and IV lengths (leg IV/I length =1.04, SD= 0.02). It differs from I. zorodes by blackish coloration with a reddish pattern on abdomen dorsum (Fig. 143). It differs from I. vanini sp. n. by shorter embolus (Figs 134-136). Female differs from those of all species in spermathecal morphology, two straight to slightly curving spermathecae tapering little from base to apex, lacking folds and with constriction forming two to multiple distal lobes (Fig. 139). Additionally, females have a general blackish coloration with a reddish pattern on abdomen dorsum (Fig. 142).
Etymology. The specific name is a matronym in honour of my wife, Kátia de Mendonça Faria, who illustrated several plates for spider taxonomy papers, this work included.
Distribution. Brazil: state of Bahia, endemic to Chapada Diamantina and its immediate surrounding regions, in Espinhaço mountain range (Fig. 169).
Natural history. Two specimens were found under rocks in Ibicoara (G. Machado, pers. comm.) and in Rio de Contas, both in state of Bahia. I found four specimens in Mucugê, Parque Nacional da Chapada Diamantina, all inside bromeliads (Vriesea atra) (Fig. 147). One of the specimens folded a bromeliad leaf and connected it with silk threads, making a retreat (Fig. 148). Inside with the mother were ca. 100 small juveniles (Fig. 149). The region where they were found is high (1400 m. a. s. l.) (Fig. 144) and vegetation is sparse, growing over a rocky substratum with shallow soil (campo rupestre) . Temperature is high during the day, falling abruptly at night. Water is largely available only during the rainy period. That's a very distinct environmental condition when compared with the habitat in rainforest of I. hirsutum and I. zorodes, and seems inhospitable, mainly for an arboreal species. In this environment, bromeliads and space under rocks are among the few available shelter for spiders.
Cladistic analysis suggests that a retreat made with tree leaves and silk is primitive for Iridopelma species, and bromelicolous habitat in I. katiae sp. n. is derived. A possibility to explain the habit change could be the existence of a primitive forest in the area which became drier more recently in geological time, resulting in grasslands over a rocky or sandy shallow soil and lack of high vegetation. Bromeliads then become the primary (and relatively scarce) available microhabitat with similar conditions and resources, e. g. leaves as substratum and high humidity, necessities for an aviculariine spider. The environmental conditions in campo rupestre areas in Chapada Diamantina could aid in understanding the specialization to endemic bromelicolous lifestyle in other spider species, such as Pachistopelma spp. Color pattern ontogeny. Juveniles have light brown legs, except for black tarsi. Carapace is black and abdomen has a broad central longitudinal red stripe over a black area. Abdomen laterals are light brown (Fig. 140). When the individual grows, it becomes darker and retains the abdominal pattern, now with a longitudinal central black stripe (Fig. 141). Adults males and females are completely black and have a conspicuous abdominal pattern remains (Figs 142-143). Diagnosis. The male differs from those of all other species by very long leg I (leg IV/I length = 0.73) (Fig. 165). The female resembles I. hirsutum by diverging spermathecae folded distally (Fig. 161), but differs by having a grayish carapace and legs and lacking an abdominal pattern (Fig. 164).

Iridopelma oliveirai
Etymology. Specific name is a patronym in honor of Judge João Carlos Sá Moreira de Oliveira, for allowing the author to have access to part of the aviculariine specimens deposited in the Instituto Butantan collections, which were fundamental for the completion of this present work.
Distribution. Brazil: known only from Central, State of Bahia (Fig. 169). Natural History. Little is known regarding this species. The vegetation in Central, state of Bahia, comprises mainly caatinga (a xerophilous vegetation). One female was found at night in a bromeliad (Bromelia laciniosa).
Color pattern ontogeny. Data for small juveniles is lacking. Large immatures specimens have a pattern similar to I. hirsutum (Figs 162-163). Adult male and female lack or have a very inconspicuous abdominal pattern (Figs 164-165). Diagnosis. The female resembles that of I. vanini sp. n. by long double folded spermathecae (Fig. 166), but differ by lacking type II urticating setae on abdomen dorsum. Male unknown.
Etymology. The specific name is a patronym in honour of Marco Antonio de Freitas, a Brazilian zoologist and geographer who collected the holotype and some other aviculariine specimens studied in this work.

Distribution.
Brazil: known only from São Desidério, state of Bahia (Fig. 169). Natural history. The only known specimen was found in cerrado/carrasco vegetation (Fig. 168), under loose tree bark about one meter above the ground (M. A. Freitas, pers. comm.).
Color pattern ontogeny. Only the adult female holotype is known. Cladistics. Searches using NONA resulted in 12 trees and strict consensus is used (Fig. 177). With X-PEE-WEE, it was found from 1 to 6 trees, depending on the concavity used (Table 2). Cladogram obtained with concavity 6 is shown in Figs 178. The main difference of cladograms obtained with the different strategies comprise the position of clade having Ephebopus, Tapinauchenius and Psalmopoeus. In NONA and X-PEE-WEE with concavities 2-6 that clade is paraphyletic in respect to other aviculariine genera (Figs 177-178); with X-PEE-WEE and concavity 1 Aviculariinae is retrieved as monophyletic. Topologies of clade Typhochlaena (Avicularia spp.1 (Pachistopelma, Avicularia spp.2, Iridopelma)) were retrieved in all employed search strategies. Resolution of the terminal clade retrieving Iridopelma as sister group of Avicularia spp.2 was obtained with X-PEE-WEE and concavities 1-5, whereas Iridopelma and Avicularia spp.2 were in a tricotomy with Pachistopelma when using NONA and X-PEE-WEE with concavity 6. Internal relationship of Typhochlaena is totally unresolved with NONA, and partially resolved with X-PEE-WEE, independently of the concavity used. Position of Iridopelma marcoi sp. n. and Iridopelma katiae sp. n. change more markedly. With concavities 1-5 Iridopelma katiae sp. n. is sister-group of Pachistopelma (Iridopelma + Avicularia spp.2). With NONA and X-PEE-WEE with concavity 6 it is part of Iridopelma clade. Iridopelma marcoi sp. n. is part of Iridopelma clade only with NONA. With X-PEE-WEE it is always polyphyletic. Its position will be fully discussed below.
The results obtained herein were compared with other published analyses for Ephebopus spp. and Aviculariinae genera relationship of West et al. (2008), which share 21 taxa and 48 characters with the present work. NONA (Fig. 177) and X-PEE-WEE with concavities 2-6 ( Fig. 178) results differed in topology with West et al. (2008) cladogram by clades having Haplopelma and Poecilotheria and that with Psalmopoeus, Tapinauchenius and Ephebopus spp. swapping positions. In West et al. (2008) these three genera were basal in the Aviculariinae and Haplopelma and Poecilotheria were sister to Theraphosinae. In the present analysis, Aviculariinae are paraphyletic, the clade Encyocratella (Haplopelma + Poecilotheria) is the sister-group of Aviculariinae and the clade with Psalmopoeus, Tapinauchenius and Ephebopus spp. is the sister group of Pelinobius + Phlogiellus. However, with X-PEE-WEE and concavity 1 Aviculariinae is retrieved as monophyletic, and, although with a slight different topology, it contains all genera proposed by West et al. (2008) to belong in Aviculariinae. Anyway, I refrain to make changes in the subfamily content as other analyses with all Avicularia species is in preparation and can give support for one or other topologies. Additional information on morphological variability of Poecilotheria, Ornithoctoninae, Selenocosmiinae and Eumenophorinae would help to resolve the conflict, but cladistic and taxonomical studies on these taxa are lacking.
In agreement with other studies (Ramírez 2003, West et al. 2008, Bertani et al. 2011) the shortest tree was obtained with X-PEE-WEE and concavity 6 (Fig. 178, Tables 3-4). It has also the highest fit (Table 2) and it is chosen as the preferred tree, on which the discussion below is done.
Typhochlaena is monophyletic and supported by characters sternum as long as wide, truncated behind (character 7, Table 4), and posterior lateral spinnerets with distal segment short, domed (character 10). The topology of Typhochlaena clade shows a tricotomy. In that T. seladonia and T. curumim sp. n. are sister-groups supported by spiraled spermathecae (character 34) with homoplasies in I. marcoi sp. n. and I. hirsutum (I. vanini sp. n. + I. oliveirai sp. n). Typhochlaena amma sp. n. and T. paschoali sp. n. are sister-groups supported by abdomen pattern in immatures (state 5 of character 29) with homoplasies with most Iridopelma species.  Pachistopelma is monophyletic and supported by 9 synapomorphies: low eye tubercle (character 0), straight anterior row of eyes (character 1), clypeus absent (character 2), flattened abdomen a quarter longer than wide in females (character 9), leg IV longer than leg I in females (character 12) and in males (character 13), leg rings on distal tibiae and metatarsi not evident (character 24), embolous straight or very slightly curved (character 40), bromelicolous lifestyle (61). All these synapomorphies are very homoplasious (3 additional steps or more), except for character 9, which is exclusive for the genus, and character 61, homoplastic with I. katiae sp. n.
Iridopelma is monophyletic only if I. marcoi sp. n. is excluded. However, some important characters for the analyses are exclusive to males (i. e., tibial spurs, cymbium protuberance) and at least one, exclusive to Iridopelma (tibial spur presence on leg II). The lack of a known male in this species undoubtedly influenced its position in the cladogram topology. Also, the absence of urticating hairs in female, that is an important synapomorphy for a large aviculariine clade, surely contributed for this result as well. On the other hand, the spermathecae shape, as well as some somatic characters indicate it is an Iridopelma species. The discovery of the male of I. marcoi sp. n. migth confirm its position.
Other Iridopelma species are in a monophyletic clade. Three synapomorphies support Iridopelma monophyly: dorsal abdominal pattern in immatures (Character 29), homoplasious with T. amma sp. n.; embolous length that changes from long to median (Character 41), extremally homoplasious, and presence of a tibial spur also on leg II, exclusive for the genus (character 49). Therefore, the main synapomorphy for Iridopelma is still the classical character proposed by Pocock (1901). However, the other proposed synapomorphy, leg I longer than leg IV (Pocock 1901) (character 12 for females and 13 for males) is extremely homoplasious. Even in Iridopelma, only I. hirsutum and I. oliveirai sp. n. males have a clearly longer leg I in relation to leg IV. All Iridopelma females have leg I and IV of similar length. Support for the node 59 (I. vanini sp. n.+ I. oliveirai sp. n.) is their habitat in deciduous forest (character 59), an homoplasious character. Node 60 I. hirsurtum (I. vanini sp. n. + I. oliveirai sp. n.) has as apomorphy the spiraled spermathecae (character 34). Node 61 is supported by character 32, presence of lobes in spermathecae, extremely homoplasious.
Biogeography. Typhochlaena, Pachistopelma and Iridopelma species are highly endemic, and there is no overlapping in distribution of species belonging to the same genus (Figs 179-181). Most areas of endemism for species of those genera are mostly concordant with river systems, as proposed for Atlantic rainforest in Northeastern and Southeastern Brazil (Pellegrino et al 2005). The northermost endemic area ranges roughly from state of Ceará/Rio Grande do Norte southwards to state of Alagoas and is limited in the south by Rio São Francisco. Iridopelma hirsutum, Pachistopelma rufonigrum and Typhochlaena curumim sp. n. distribution is restricted to this region (Figs 179-181).  Southwards, another area of endemism is recognized between Rio São Francisco and Rio Paraguaçú (Pellegrino et al. 2005). Iridopelma zorodes, Pachistopelma bromelicola sp. n. and Typhochlaena seladonia are found in this endemic area, but there are some records for P. bromelicola sp. n. south of Rio Paraguaçú (Elísio Medrado and Maracás, state of Bahia). Cladistic analysis retrieved T. seladonia/T. curumim sp. n. and P. rufonigrum/P. bromelicola sp. n. as sister species, indicating Rio São Francisco and Rio Paraguaçú as potential barriers responsible for vicariant event (Figs 179-180). However, the analysis did not retrieve Iridopelma zorodes as sister species of I. hirsutum (Fig. 181), suggesting a more complex history for Iridopelma genus in Northeastern and Center-Western Brazil. Another area of endemism was proposed for the region from Rio Paraguaçú southwards to Rio Jequitinhonha (Pellegrino et al. 2005). Typhochlaena paschoali sp. n., Avicularia diversipes and A. gamba are endemic to the region (Figs 179, 182). Fol- Figure 178. Single tree obtained with X-Pee-Wee, all characters non-additive and concavity 6. Fit = 4507.4 length = 209. lowing southwards, another area of endemism is proposed between Rio Jequitinhonha and Rio Doce (Pellegrino et al. 2005). Only Avicularia sooretama is known from this region. Interestingly, Rio Doce is not a barrier for the species, which is distributed from Rio Jequitinhonha to Rio Paraíba do Sul (Fig. 182). However, there is a single old record for the species south of Rio Doce, in Itatiaia, State of Rio de Janeiro (Bertani and Fukushima 2009). The southermost area of endemism for aviculariines is between Rio Doce and Rio Paraíba do Sul. Typhochlaena amma sp. n. is endemic to this region and cladistic analysis retrieved it as sister species of T. paschoali sp. n. (Fig. 179).
At least two hypotheses were proposed to explain the existence of endemism areas in Atlantic forest of Northeastern and Southeastern Brazil. Pellegrino et al. (2005) proposed that after a semiarid climate in Miocene/Pliocene boundary the region had higher humidity conditions in the Pliocene, and major rivers were formed, bissecting deposits from Barreiras Formation (Sugio and Nogueira 1999), thus isolating interriverine regions. The alternative hypothesis (Carnaval and Moritz 2008) provided evidence for the existence of forest refugia areas during the Quaternary that are congruent with areas of endemism known for Atlantic forest of Northeastern Brazil.
Whereas distribution of congeners did not overlap, there is strong ovelap of species of Typhochlaena with Pachistopelma, Iridopelma and Avicularia spp.2 as well as of Iridopelma with Pachistopelma (Fig. 183). Typhochlaena spp. overlap with Avicularia spp. 2 on state of Espírito Santo and Southern state of Bahia, with Iridopelma on Northern state of Bahia, Sergipe, Paraíba, Piauí, Maranhão and possibly Pernambuco, Alagoas and Tocantins. They also overlap with Pachistopelma spp. in Northern Bahia, Sergipe, Alagoas, Pernambuco and Paraíba. Pachistopelma spp. overlap with Iridopelma spp. and Typhochlaena spp. over almost all their distribution (from Rio Grande do Norte to Northern Bahia). Iridopelma possibly overlaps slightly with Avicularia spp.1 in parts of Maranhão, Pará and Tocantins, with Typhochlaena and Pachistopelma on the coast and with Typhochlaena on Northern Bahia, Sergipe, Alagoas, Pernambuco, Paraiba and Southern Ceara, Piauí, Maranhão and parts of Tocantins. Most of the overlap involves Typhochlaena -the most basal genus of the clade, and, if this genus is not considered, only a marginal overlapping of Avicularia spp.1, Avicularia spp.2 and Iridopelma will remain (except Pachistopelma, which will be discussed separately).
Habitat and evolution. The overlap of Iridopelma spp. with Pachistopelma spp. in most of the distribution of the latter deserves a more detailed discussion. Pachistopelma spp. are strictly associated with bromeliads (Bertani et al. 1994, Santos et al. 2002, 2004Dias et al 2000, Dias and Brescovit 2003, Dias and Brescovit 2004, this work), a characteristic possibly shared only with Iridopelma katiae sp. n. In all other closely related aviculariine genera with available information, the retreat is made under loose tree bark (Typhochlaena seladonia, T. curumim sp. n.), on tree trunk or on leaves of palm trees (Avicularia sp.1, A. juruensis and A taunayi), with two or more leaves connected with silk threads (Avicularia spp. 2, Iridopelma hirsutum, I. zorodes). Therefore, almost all other aviculariine species use primarily trees to make their retreats, though some use bromeliads eventually (immature Avicularia avicularia -Stradling, 1994; Iridopelma oliveirai sp. n.-type label data; immature Avicularia juruensis, immature and adult Avicularia diversipes (pers. obs.).  Pachistopelma spp. inhabit tank bromeliads -Aechmea aquilega (Dias et al. 2000); Hohenbergia stellata, H. ridley (Dias et al. 2000); H. ramageana and A. lingulata  which are mainly terrestrial in rocky outcrops, restinga and even caatinga, and are facultatively epiphytes (Siqueira Filho and Leme 2006). When terrestrials, they grow on areas of shallow, sandy or rocky soil, exposed to intense sunlight. Tank bromeliads retain rain water on their phytothemalta and are an important source of humidity to several species of animals (Frank and Lounibos 2009). Tank bromeliads are particularly abundant in drier areas of restinga which are depleted of high trees . In this xeric environment there are few available places with adequate humidity and protection against high temperatures. Therefore, bromeliads provide the few suitable microhabitats for a variety of animals.
Pachistopelma spp. seem specialized to live inside bromeliads, mainly by the dorso-ventrally flattened body of immatures and female which aids the spider in moving between the narrow inter-leaves spaces (Bertani 1994, Bertani in Dias et al. 2000. A question arising then is, why this specialization occurred with Pachistopelma spp., but not with Iridopelma spp. that live in contiguous areas? It is largely known that climate fluctuations of the Neogene and the Quaternary periods transformed large wet forested regions of Northeastern Brazil into xeric biomes (Ab'Sáber 1977, Martin et al. 1993, Suguio and Nogueira 1999, Carnaval and Moritz 2008. With the change, part of the fauna more dependent on wet and low temperature probably became extinct. Especially arboreal animals might be affected, as large trees became rare. On the other hand, bromeliads, that live normally as epiphytes or on rocks and forest borders, expanded their distribution on the ground. Thus, these plants were the few available places for the ancestor of Pachistopelma species to live. Natural selection then acted and morphological modifications took place leading to speciation and specialization to live strictly inside bromeliads. When climate changed and the region became wet again, the forest expanded and bromeliads became restricted as epiphytes or more concentrated in restinga regions in the coast, caatinga, or on rocky formations on hills. Even though the habitat along their distribution area is now suitable for an arboreal spider, the morphological, ethological end physiological specializations of Pachistopelma species did not allow them to recolonize trees. Pachistopelma populations seem to be smaller in bromeliads on shaded forest borders than in regions exposed to direct sunlight (Santos et al. 2004, pers. obs.). On the other hand, Iridopelma species expanded their distribution following forest expansion and now species of the two genera have sympatric distribution, one living inside bromeliads and the other in trees. Existence of other animal species living strictly inside bromeliads in the same region reinforces the idea. At least another spider, Nothroctenus fuxico Dias and Brescovit, 2004 is found exclusively in bromeliads, sometimes together with Pachistopelma bromelicola sp. n. specimens (Dias and Brescovit 2004). One scorpion, Tityus neglectus Mello-Leitão, 1932, is also found living inside tank bromeliads (Lourenço andEickstedt 1988, Santos et al. 2003) in same regions as Pachistopelma spp. This model basically corresponds to the ecogeographical speciation proposed by Vanzolini and Williams (1981).
At least five other aviculariine species are found in open vegetation regions -cerrado (Typhochlaena costae sp. n., Iridopelma marcoi sp. n., Iridopelma vanini sp. n.), caatinga (Iridopelma oliveirai sp. n.), sandy dunes/restinga (Iridopelma vanini sp. n.), or campo rupestre (Iridopelma katiae sp. n.). For all species with available field data, there are records of them living on trees (I. marcoi sp. n.) or in bromeliads (I. oliveirai sp. n., I. katiae sp. n.). The holotype of I. vanini sp. n. was found under a fallen tree trunk on a sandy dune region in Parnaíba, state of Piaui. Iridopelma katiae sp. n. habitat is interesting, because it is in a high region (1200-1300 m a.s.l.) in a campo rupestre area. As restinga and caatinga, the climate is severe, with rocky formations, water stress present most times of year, high temperatures during the day and low temperatures at night, and few sparse trees (Conceição et al. 2007). Again, the only available place for an arboreal spider is the frequent bromeliad islands formed by Vriesea atra. The four specimens collected in this region were inside those bromeliads, including a female with spiderlings. The other two records are for a region close to this and the spiders were found under rocks, an unusual habitat for an arboreal spider. This indicates that I. katiae sp. n. could be suffering a similar selective pressure that led to the specialization to bromeliad lifestyle in Pachistopelma spp., after a climatic and vegetational change.
Another bromelicolous spider, the salticid Psecas chapoda (Peckham & Peckham, 1894), is specialized to living inside the bromeliad Bromelia balanseae (Romero and Vasconcelos-Neto 2005). These bromeliads also occur in open vegetation formation (cerrado), which is, sometimes, bordered by forest formations. Psecas chapoda is physiologically adapted to Bromelia balanseae, and denser populations' occur far from shaded areas of forest borders (Romero and Vasconcelos-Neto 2005). A facultative mutualism involving Psecas chapoda and Bromelia balanseae has been proposed, in which the plant benefits from nutrients generated by spider debris, such as feces, exuvia, prey carcasses and silk (Romero et al. 2006, Romero et al. 2008. A similar process to that of Pachistopelma spp. could have occurred with an ancestor of this salticid spider, leading to a specialization to live inside bromeliads. Salticid association with bromeliads are much more frequent in bromelids of open environments, such as cerrado, rather than in bromeliads of forest interiors (Romero 2006). As with cerrado, campo rupestre is another key area to understanding the association between spiders and bromeliads. Alpaida quadrilorata (Simon, 1897) (Araneidae) is also associated with a bromeliadlike plant, Paepalanthus bromelioides (Eriocaulaceae) (Figueira and Vasconcellos-Neto 1991) which also has leaves in a rosette and occurs on sandy soil in campo rupestre areas of Serra do Cipó. I also observed the same association with these spider species and plant in campo rupestre of Caraça (Santa Bárbara, state of Minas Gerais). Therefore, this can be a recurrent process leading to the strict bromelicous lifestyle, and should be investigated in other animal groups having similar habits.
This complex history may be responsible for the extraordinary diversity of aviculariines in Northeastern, Southeastern and Central-western Brazil, which rivalizes with the aviculariine forms found in Northern South America and Central America in number of species, morphological variation and habitat use. Specifically, parts of Northeastern Brazil, as Reconcavo Bahiano, has one of the richest aviculariine fauna in