2005. The Journal of 33:456±467

TARSAL SCOPULA SIGNIFICANCE IN PHYLOGENETICS (ARANEAE, , THERAPHOSIDAE)

Jose Paulo Leite Guadanucci: Museu de Zoologia da Universidade de SaÄo Paulo, Instituto de BiocieÃncias da Universidade de SaÄo Paulo, Av. NazareÂ, 481, Ipiranga, CEP: 04263±000 SaÄo Paulo, SPÐBrazil. E-mail: [email protected]

ABSTRACT. Tarsal scopula condition and carapace length were studied for eighteen Ischnocolinae species. For cladistic analysis a matrix of 20 terminals and 30 characters of representatives of Ischnocol- inae, Theraphosinae, Aviculariinae, and Trichopelmatinae were analyzed using Nona 2.0 computer software. The matrix was analyzed in four different ways: 1. each tarsal scopula (legs I±IV) coded as separate characters; 2. one character with six ordered states; 3. one character with six independent states; 4. without tarsal scopula character. The ®rst two matrices result in one tree with the same indices (L ϭ 72; CI ϭ 0.54; RI ϭ 0.74) and topology: Part of Ischnocolinae is monophyletic (H. rondoni(S. longibulbi(I. algericusϩCatumiri))) and the other representatives (Oligoxystre and 1) form a distinct monophyletic group with Theraphosinae, Harpactirinae and Aviculariinae. There are no homoplasies in tarsal scopula evolution in the second cladogram. The other two cladograms show less resolution for the Ischnocolinae than the two ®rst cladorams. The tarsal scopula condition appears to have no relation to size (t ϭϪ0.80433; P ϭ 0.438247) and should be used in phylogenetic analysis of Ischnocolinae because it provides information on the character variability within the subfamily. Keywords: Phylogeny, South America, cladistics

The condition of the tarsal scopula has had subfamily based on a plesiomorphic character an important role in the systematics of the Is- state (divided tarsal scopula), the situation of chnocolinae Simon 1892. The scopula shows the group's systematics is very confusing. Ra- ontogenetic differentiation, being divided in ven (1985) considered Ischnocolinae a para- all juvenile Theraphosidae and becoming en- phyletic group that should have been revised tire in adults of some groups (Pocock 1897; at the generic level and grouped into mono- Gerschman de Pikelin & Schiapelli 1973; phyletic units. However, since the description PeÂrez-Miles, 1994). The condition of the tar- of the type-genus Ausserer 1871, sal scopula has been considered a good taxo- only a few genera have been revised. Gersch- nomic tool and has already been used to di- man de Pikelin & Schiapelli (1973) revised agnose genera and species groups in the subfamily as a whole but of the 42 genera Theraphosidae. Its use in phylogenetics is included in this study, only 10 are in fact Is- questionable since, within the Theraphosinae, chnocolinae representatives. The remaining 12 the presence of a divided scopula is related to were subsequently synonymyzed or trans- small sized species (PeÂrez-Miles 1994). Char- ferred to other families and subfamilies. Rud- acterized as theraphosids with a divided tarsal loff (1997) revised the genus scopula (plesiomorphic state), Ischnocolinae Karsch 1879 but did not present a diagnosis is considered a paraphyletic group. Ischnocol- of the genus and its species; the identi®cation inae is the subfamily of Theraphosidae Tho- key does not include all the species and the rell 1869 that shows the broadest geographic structures are poorly illustrated. Smith (1990) distribution, with species occurring in north- published a taxonomic revision of European ern, central and eastern Africa, Seychelles, the and African Ischnocolinae and presented de- Middle-East, the Mediterranean region, cen- scriptions and diagnoses for all genera. tral and south Americas and the Antilles For over a century, the tarsal scopula state (Smith 1990; Rudloff 1997; Vol 2001). Con- ``divided by a longitudinal band of setae'' was sidering that Ischnocolinae was proposed as a used in Theraphosidae (PeÂrez-Miles

456 GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 457

1994). Gerschman de Pikelin & Schiapelli species used for tarsal scopulae condition and (1973), following Ausserer (1871), considered carapace length: the tarsal scopula condition an important tax- ( onomic character and stated that the divided Holothele rondoni (Lucas & BuÈcherl 1972): 1 , ApiaÂcas, Mato Grosso, (MZSP 18046); 1 tarsal scopula is present in all juvenile thera- (, ApiaÂcas, Mato Grosso, Brazil (MZSP 18038); phosids. Juvenile Theraphosinae Thorell 1870 1 (, ManicoreÂ, Amazonas, Brazil (MZSP 18990). have divided tarsal scopulae that become en- longibulbi Soares & Camargo 1948: 4 (, tire in the adult stage, in Ischnocolinae the Itirapina, SaÄo Paulo, Brazil (MZSP 22756). scopulae remain divided into adulthood (Po- Genus 1: 3 (, Colinas do Sul, Serra da Mesa, cock 1897; Gerschman de Pikelin & Schia- GoiaÂs, Brazil (MZSP 18992). pelli 1973; PeÂrez-Miles 1994). Although this petropolium Guadanucci 2004: 1 (, Pe- ontogenetic differentiation was detected, the troÂpolis, Rio de Janeiro, Brazil (IBSP 8596). ( divided condition continued to be used caus- Catumiri chicaoi Guadanucci 2004: 1 , Una, Ba- hia, Brazil (IBSP 9514). ing the inclusion of juvenile Theraphosinae Catumiri uruguayense Guadanucci 2004: 1 (, Lav- within Ischnocolinae. This problem remained alleja, A guas Blancas, Uruguay (IBSP 9491). unresolved until Raven (1985) considered Is- Catumiri argentinenese (Mello-LeitaÄo 1941): 1 (, chnocolinae a paraphyletic group that presents Jujuy, Yuto, El Pantanoso, Argentina (MACN poorly developed tarsal scopulae. The tarsal 6424). scopula as a phylogenetic character was used Genus 2, sp. 1: 3 (, JaraguaÂ, GoiaÂs, Brazil (MPEG for the ®rst time by PeÂrez-Miles (1992) in a 1677) ( preliminary cladistic analysis of the subfamily Genus 2, sp. 2: 1 , Serra Norte, ParaÂ, Brazil Theraphosinae. In this paper he shows that the (MPEG 1678). Genus 2, sp. 3: 1 (, Ilha Marajo Breves, ParaÂ, Bra- entire tarsal scopula is synapomorphic for zil (MPEG 1679). some genera of this subfamily. Later, PeÂrez- Genus 2, sp. 4: 1 ( (IBSP 11083), 1 ( (IBSP Miles (1994) discussed the value of the tarsal 11086), 1 ( (IBSP 11087), Pimenta Bueno, Ron- scopula in Theraphosinae systematics and doÃnia, Brazil; 1 (, Mineiros, GoiaÂs, Brazil (IBSP concluded that the scopula condition is related 8070). to spider size, although some exceptions exist. Genus 2, sp. 5: 1 (, Linhares, EspõÂrito Santo, Brazil Considering that the role of the tarsal scopula (IBSP 8654); 1 (, Linhares, EspõÂrito Santo, Bra- ( in Theraphosidae systematic remains obscure, zil (IBSP 7987); 1 , Porto Seguro, Bahia, Brazil (IBSP 11084); 1 (, IlheÂus, Bahia, Brazil (IBSP the goal of this study is to vary the use of the 11085). character ``tarsal scopula'' in a phylogenetic Oligoxystre new species 1: 1 ( (IBSP 9488), 1 ( analysis for Ischnocolinae and discuss the re- (IBSP 9489), 1 ( (IBSP 9486), 1 ( (IBSP 9484), sults. Central, Bahia, Brazil.

METHODS Cladistic analysis.ÐThe cladistic analysis was carried out using Nona version 2.0 (Go- The material examined belongs to the fol- loboff 1993). Search strategy was mult* with lowing institutions: Instituto Butantan, SaÄo 100 replications. The data matrix included 20 Paulo (IBSP); American Museum of Natural terminal taxa and 30 characters and was con- History, New York (AMNH); Museo Argen- structed with NDE (Nexus Data Editor) ver- tino de Ciencias Naturales Bernardino Riva- sion 0.5.0 (Page 2001). The out-group was davia, Buenos Aires (MACN); Museo de la chosen based on the phylogenetic relation- Plata, La Plata (MLP); Museu de Zoologia, ships of Mygalomorphae presented by Golo- Universidade de SaÄo Paulo, SaÄo Paulo boff (1993). In order to avoid an excess of (MZSP); Zoological Museum University of missing entries, we preferred used a Bary- Copenhagen (ZMUC); Museu Paraense Emi- chelidae Simon 1889 rather than a Paratropi- lio Goeldi, BeleÂm (MPEG). didae Simon 1889 as the out group, since the Tarsal scopula condition was observed un- last family presents some incomparable char- der a stereomicroscope. Following PeÂrez- acters with Theraphosidae. Character polarity Miles (1994), a few isolated long and thin was read straight from the preferred clado- hairs in the tarsal scopula were not considered gram following Nixon & Carpenter (1993). divided. Carapace length was used to estimate Below is a list of species used in the cla- spider size. Below is a list of Ischnocolinae distic analysis: 458 THE JOURNAL OF ARACHNOLOGY

Figures 1±2.ÐRelationship hypothesis between Ischnocolinae and other Theraphosidae groups. 1. (L ϭ 72; CI ϭ 0.54; RI ϭ 0.74). Tarsal scopula coded in four characters (22±25), one for each leg. 2. (L ϭ 72; CI ϭ 0.54; RI ϭ 0.74). Tarsal scopula coded as one character with six ordered states. GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 459 Characters Table 1.ÐMatrix composed of 20 terminals and 31 characters. 0 0 0 000000 0 0 0000Ð0000000000000 0 0 1 000010 1 0 0210Ð1000110000112 0 1 2 3 4 56789101112131415161718192021222324252627282930 0 1 0 1 0 0000000 0 1 0 0 0210Ð1000000000010 0 0 0000100 0 1 0 0 0212 1 0 1 0 0010000000110 000010ÐÐ0210Ð0010000010110 01ÐÐ0Ð100011 0101 0 0010000010110 0 1 1 0 ? 0 0000101ÐÐ0Ð000010 ? 0 1103 0 1 1103 0100120010113 0 0000000010110 0 1 0 0 0 000010 0 0 1003 1 0100000010110 0 0 1 0 0 000010 0 0 1003 1 0100000010110 0 0 1 1 0 0000102Ð0 0 1 0 0 000000 0210Ð0010121220116 0 00 0210Ð0002121200116 1 0 0 0 0011000 0 1 1 0 0110Ð1000121211116 0 0 001100 0 0 0212 0 1000121201116 p40 1 0 1 0 000010 0 1 1 1 0 1100Ð0010121100115 1 0 000010sp.4 0 1 1 1 0 1100Ð0010121100115 1 0 000010sp.5 0 1 1 0 0 1100Ð0010121110115 1 0 010010sp.1 0 1 1 0 0 1100Ð0010121110115 1 0 010010sp.2 1 0 1100Ð0010121110115 sp.3 p 0sp. 1 0 1 0 000000 1 0 0222 0 0001121000114 Oligoxistre Oligoxistre Oligoxistre Oligoxistre BarychelinaeHolothele rondoni Sickius longibulbi 0Ischnothele algericus Catumiri 0 petropo- lium Catumiri chicaoi Catumiri uruguay- ense Catumiri argenti- nense Genus 1Pterinochilus muri- nus Avicularia avicularia Euathlus vulpinus 0Vitalius vellutinus Oligoxistre 1 annae Tapinauch 460 THE JOURNAL OF ARACHNOLOGY

BARYCHELIDAE: Reichlingia annae (Reichling Brazil (IBSP 8019); 1 &, Votuporanga, SaÄo Pau- 1997): 1 (,1&, New River Lagoon, Orange lo, Brazil (IBSP 8693). Walk, Belize (AMNH). Tapinauchenius sp. (Aviculariinae): 1 (, TucuruõÂ, THERAPHOSIDAE: Avicularia avicularia (Lin- ParaÂ, Brazil (IBSP 4925-A); 1 &, Rio MarupõÂ, naeus 1758) (Aviculariinae): 1 (, JacareÂ, Rio ParaÂ, Brazil (IBSP 4676). Trombetas, OriximinaÂ, ParaÂ, Brazil (MZSP Vitalius vellutinus (Mello-LeitaÄo 1923) (Theraphos- 5687); 1 &, JacareÂ, Rio Trombetas, OriximinaÂ, inae): 1 (, Porto Cabral, Rio ParanaÂ, Teodoro ParaÂ, Brazil (MZSP 5687). Samapaio, SaÄo Paulo, Brazil (MZSP 14953); 1 &, Euathlus vulpinus (Karsch 1880) (Theraphosinae): Teodoro Samapaio, Porto Cabral, Rio ParanaÂ, 5 (, Osorno, Chile (IBSP 3817-A); 4 &, Osorno, Teodoro Sampaio, SaÄo Paulo, Brazil (MZSP Chile (IBSP 3817-B). 3150). Catumiri petropolium Guadanucci 2004 (Ischno- colinae): 1 (, PetroÂpolis, Rio de Janeiro, Brazil CLADISTIC ANALYSIS (IBSP 8596); 1 (, PetroÂpolis, Rio de Janeiro, Below is a list of the characters used to con- Brazil (IBSP 8606). struct the data matrix. The matrix was ana- Catumiri chicaoi Guadanucci 2004 (Ischnocolinae): lyzed in four different ways: 1. tarsal scopula 1 (, Una, Bahia, Brazil (IBSP 9514); 1 &, Una, of each leg was coded as a separate character Bahia, Brazil (IBSP 9514). (characters 22±25) and these were treated as Catumiri uruguayense Guadanucci 2004 (Ischno- ordered; 2. the four tarsal scopula were coded colinae): 1 (, Lavalleja, AÂ guas Blancas, Uruguay as a single character with six ordered states (IBSP 9491); 1 &, Lavalleja, AÂ guas Blancas, Uruguay (IBSP 9507). (character 30); 3. the four tarsal scopula were Catumiri argentinenese (Mello-LeitaÄo 1941) (Is- coded as a single character with six indepen- chnocolinae): 1 (, Jujuy, Yuto, El Pantanoso, Ar- dent states (character 30); 4. the character(s) gentina (MACN 6424); 1 &, Catamarca, Argen- of the tarsal scopula were deactivated in the tina (MLP 14608). matrix. The optimization option was ACCT- Genus 1 (Ischnocolinae): 1 (, Fazenda Sandoval, RAN. Abbreviations: L ϭ length of character; Porto Nacional, Tocantins, Brazil (IBSP 8585); 1 CI ϭ consistency index; RI ϭ retention index. RI ;0.66 ؍ CI ;3 ؍ Fazenda Sandoval, Porto Nacional, Tocantins, 0. Male tibial spur (L ,& present; 1. absent; 2. present, formed ,0±.(0 ؍ .(Brazil (IBSP Holothele rondoni (Lucas & BuÈcherl 1972) (Is- by thick spines. The great diversity of male ( chnocolinae): 1 , ApiaÂcas, Mato Grosso, Brazil tibial spur morphology might be related to re- & (MZSP 18046); 1 , TucuruõÂ, ParaÂ, Brazil productive isolation. The tibial spur is the ®rst (IBSP). structure that touches the female and could act Ischnocolus algericus Thorell 1875 (Ischnocoli- nae): 1 (,1&, El Araish, Marocco (ZMUC 620, as a mechanism for the female to recognize a 628). conspeci®c male (Coyle 1985; Eberhard 1985; Oligoxystre new species 1 (Ischnocolinae): 1 (, Jackson & Pollard 1990). Since the structures Central, Bahia, Brazil (IBSP 9487); 1 &, Toca da that compose the tibial spur are under inde- EsperancËa, Jussara, Bahia, Brazil (IBSP 8549). pendent evolution, the ``tibial spur'' is coded Oligoxystre new species 2 (Ischnocolinae): 1 (, in three different characters. Chapada dos GuimaraÄes, Mato Grosso, Brazil 1. Apical megaspine in the tibial spur (L .Ð0. present; 1.(0 ؍ RI ;0.33 ؍ CI ;3 ؍ IBSP 9495); 1 &, Chapada dos GuimaraÄes, Mato) Grosso, Brazil (IBSP 9504). absent. ؍ Oligoxystre new species 3 (Ischnocolinae): 1 (, 2. Prolateral branch of tibial spur (L .Ð0. present; 1. absent.(0 ؍ RI ;0.33 ؍ SaÄo Domingos, GoiaÂs, Brazil (IBSP 8625); 1 &, 3; CI ؍ CI ;4 ؍ Serra da Mesa, MinacËuÂ, GoiaÂs, Brazil (IBSP 3. Metatarsus I of males (L ؍ .(9467 ( 0.25; RI 0.66).Ð0. straight; 1. dorsoven- Oligoxystre new species 4 (Ischnocolinae): 1 , trally curved. This character shows great var- TucuruõÂ, ParaÂ, Brazil (IBSP 9459); 1 &, TucuruõÂ, iation in the degree of curvature and in the ParaÂ, Brazil (IBSP 7936). & taxa in which it is present. ؍ , Oligoxystre new species 5 (Ischnocolinae): 1 Toca da EsperancËa, Central, Bahia, Brazil (IBSP 4. Flexion of metatarsus I of males (L ؍ ؍ 8553). 1; CI 1; RI 1).Ð0. ¯exes outside the murinus Pocock 1897 (Harpactiri- prolateral branch of tibial spur; 1. ¯exes be- nae): 1 (, Africa (IBSP); 1 &, Kenya (IBSP). tween the two branches of tibial spur. The way Sickius longibulbi Soares & Camargo 1948 (Is- that the metatarsus ¯exes is related to the po- chnocolinae): 1 (, ParnaõÂba, Mato Grosso do Sul, sition of the tibial spur. GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 461

Figures 3±4.ÐRelationship hypothesis between Ischnocolinae and other Theraphosidae groups. 3. (L ϭ 79; CI ϭ 0.48; RI ϭ 0.56). Tarsal scopula coded as one character with six unordered states. 4. (L ϭ 57; CI ϭ 0.57; RI ϭ 0.72). Tarsal scopula deactivated. 462 THE JOURNAL OF ARACHNOLOGY

Figure 5.ÐVariation possibilities of Ischnocolinae tarsal scopulae. States correspond to character 30. Full line corresponds to divided scopulae with setae; dashed line corresponds to entire scopulae with band of setae.

;2 ؍ Ventral depression of the palpal tibia 12. Lateral lobe of spermathecae (L .5 .Ð0. absent (Fig. 7); 1.(0.5 ؍ RI ;0.5 ؍ Ð0. CI.(1 ؍ RI ;1 ؍ CI ;1 ؍ of males (L straight or slightly curved, occupying more present (Fig. 8). ؍ RI ;0.5 ؍ CI ;2 ؍ than half of the article; 1. sigma-like, occu- 13. Maxillae (L pying half the article. Autapomorphic for I. 0.87).Ð0. many cuspules (more than 50); 1. algericus. Diagnostic for the genus Ischno- few cuspules (less than 45). (0.5 ؍ RI ;0.33 ؍ CI ;6 ؍ colus (Raven 1985). 14. Labium (L Ð ordered.Ð0. cuspules absent; 1. few cuspules.(1 ؍ RI ;1 ؍ CI ;1 ؍ (L .6 0. apical keel absent; 1. apical keel present. (less than10); 2. many cuspules (more than The presence of keels on the bulb is a syna- 15). RI ;0.4 ؍ CI ;5 ؍ pomorphy of Theraphosinae (Raven 1985; 15. Labium shape (L Ð0. much wider than long (2.5±3.(0.57 ؍ -Perez-Miles et al. 1996; Bertani 2000). How ever, small keels were observed on the bulb times wider); 1. almost as wide as long (less of some Oligoxystre species. These are not than 2 times wider); 2. longer than wide. ؍ RI ;0.6 ؍ CI ;5 ؍ considered homologous to the Theraphosinae 16. Tarsal claw (L bulb keels. 0.6).Ð0. bare, without teeth; 1. two rows of 7. Prolateral keels on the palpal bulb (L teeth; 2. median row of teeth; 3. prolateral row .Ð0. absent; 1. present. of teeth; 4. single tooth.(1 ؍ RI ;1 ؍ CI ;1 ؍ ؍ CI ;1 ؍ Synapomorphy of Theraphosinae (Raven 17. Tarsal claw with teeth (L Ð0. present on all legs; 1. present.(1 ؍ Perez-Miles et al. 1996; Bertani 2000). 1; RI ;1985 .(Ð on anterior legs (I±II.(1 ؍ RI ;1 ؍ CI ;1 ؍ Subtegulum (L .8 ؍ CI ;3 ؍ narrow, not extending over the tegulum; 1. 18. Posterior sternal sigilla (L .0 -Ð0. marginal; 1. submar.(0.33 ؍ wide, extending over the tegulum. Synapo- 0.33; RI morphy of Theraphosinae (Raven 1985; Per- ginal. ؍ RI ;1 ؍ CI ;1 ؍ ez-Miles et al. 1996). 19. Metatarsus I (L ;Ð0. more than ¾ of the article scopulate.(1 ؍ Ventral region of the cymbium (L .9 .Ð0. as wide as long; 1. less than half of the article scopulate.(0.6 ؍ RI ;0.33 ؍ CI ;3 RI ;0.25 ؍ CI ;4 ؍ longer than wide. This character is very 20. Metatarsus IV (L .1 -Ð0. less than half of the article scop.(0.62 ؍ common among Ischnocolinae although it does not represent a synapomorphy for this ulate; 1. more than half of the article scopu- group. late. ؍ RI ;0.66 ؍ CI ;3 ؍ Legs spines (L .21 ;1 ؍ Size of the lobes of cymbium (L .10 Ð0. similar; 1. different. 0) ordered.Ð0. many spines, especially on.(1 ؍ RI ;1 ؍ CI ,tibia and metatarsus; 1. few reduced spines ؍ Lobular state of spermathecae (L .11 ;Ð0. unilobular on the apical region of tibia and metatarsus.(0.71 ؍ RI ;0.33 ؍ CI ;3 (Fig. 7); 1. multilobular (Fig. 8). 2. spines absent. The presence of several GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 463 spines was considered plesiomorphic for III entire; 5. tarsal scopula I±III entire and Theraphosoidina (Raven 1985). Representa- scopula IV entire with a longitudinal band of tives of Aviculariinae Simon 1874 show a re- setae; 6. all tarsal scopula entire. Although duced number of leg spines. Bertani (2002) state 1 was not observed in any of the speci- demonstrated that state 2 is a synapomorphy mens examined in this study, it was included for Aviculariinae sensu stricto. in the matrix since it was observed in an on- RI togenetic series. The exuvia of a specimen of ;0.5 ؍ CI ;2 ؍ Tarsal scopula I (L .22 -Ð0. divided by a longitudinal band Grammostola actaeon (Pocock 1903) was ob.(0.85 ؍ of setae; 1. entire. served and it demonstrated that from state 0 -to 2, two steps must be counted since the tar ;0.5 ؍ CI ;4 ؍ Tarsal scopula II (L .23 ordered.Ð0. divided by a longi- sal scopula II does not turn into undivided (0.86 ؍ RI tudinal band of setae; 1. entire with a longi- with a band of setae unless the scopula I is tudinal band of setae; 2. entire. In the present undivided. study a third state was identi®ed which differs from state 0 in having type A setae (Rovner RESULTS 1978; PeÂrez-Miles 1994) mixed with type B The ®rst cladogram (Fig. 1) refers to the setae (Rovner 1978; PeÂrez-Miles 1994). In tarsal scopula character separated into four in- this state the tarsal scopula is not divided but dividual characters (characters 22±25). It re- there are lined setae forming a longitudinal sulted in a single tree (L ϭ 72; CI ϭ 0.54; RI band. ϭ 0.74). It shows that part of Ischnocolinae, -RI represented by the taxa (Holothele ron ;1 ؍ CI ;1 ؍ Tarsal scopula III (L .24 -Ð0. divided by a longitudinal band of doni(Sickius longibulbi(Ischnocolus algeri.(1 ؍ setae; 1. entire. cusϩCatumiri))), is monophyletic. The re- RI maining Ischnocolinae form a distinct group ;1 ؍ CI ;2 ؍ Tarsal scopula IV (L .25 -ordered.Ð0. divided by a longitudinal with Harpactirinae Pocock 1897, Theraphosi (1 ؍ band of setae; 1, entire with a longitudinal nae and Aviculariinae. This hypothesis sug- band of setae; 2, entire. gests that part of Ischnocolinae is the sister- group to the polytomy presented by Raven ؍ RI ;0.66 ؍ CI ;3 ؍ Clypeus (L .26 0.87).Ð0. absent; 1. present, narrower than (1985) solving in part the cladogram present- the diameter of the anterior median eyes; 2. ed in that study (Fig. 6). present, wider than the diameter of the ante- The second cladogram (Fig. 2) refers to the rior median eyes. tarsal scopula coded as a single character with six ordered states (character 30). It resulted in ؍ CI ;1 ؍ type III (L .27 -Ð0. absent; 1. present. Synapo- a single tree with the same topology and in.(1 ؍ RI ;1 morphy of Theraphosinae (Raven 1985; Per- dices of the ®rst cladogram but with different ez-Miles et al. 1996). optimizations in the following nodes: Catu- 28. Apical article of posterior lateral miri petropolium;((Genus 1((Tapinauchenius -Ð0. sp.(((Avicularia avicu.(1 ؍ RI ;1 ؍ CI ;1 ؍ (L domed or rounded; 1, digitiform. This char- laria(Euathlus vulpinusϩVitalius vellutin- acter is widely used to separate Barychelidae us)(Oligoxystre spp.)))); ((Tapinauchenius from Theraphosidae, the latter presenting the sp.((Pterinochilus murinus(Avicularia avicu- distal article of the PLS digitiform. However, laria(Euathlus vulpinusϩVitalius vellutin- among Barychelidae this character shows us)(Oligoxystre spp.))); (Pterinochilus muri- great variation making it impossible to place nus(Avicularia avicularia(Euathlus some representatives within either family. vulpinusϩVitalius vellutinus)(Oligoxystre spp.)) and ((Pterinochilus murinus(Avicularia ;1 ؍ Anterior maxillary projection (L .29 -Ð0. poorly developed; 1. avicularia(Euathlus vulpinusϩVitalius vellu.(1 ؍ RI ;1 ؍ CI developed. tinus)). Moreover it showed no homoplasies .for the tarsal scopula character ؍ CI ;9 ؍ Tarsal scopula I-IV (L .30 Fig. 5).Ð0. all scopula di- The third cladogram (Fig. 3), which is a) (0.92 ؍ RI ;0.66 vided; 1. only tarsal scopula I entire; 2. only consensus of 16 trees, refers to the tarsal scop- tarsal scopula I entire and scopula II entire ula coded as a single character with six in- with a longitudinal band of setae; 3. only tar- dependent states. The only monophyletic sal scopula I and II entire; 4. tarsal scopula I± groups in this tree are (Euathlus vulpi- 464 THE JOURNAL OF ARACHNOLOGY

Figure 6.ÐRelationship hypothesis among Theraphosidae subfamilies (Raven 1985). nusϩVitalius vellutinus); (Tapinauchenius I is present in large species (H. rondoni) and sp.ϩAvicularia avicularia); Catumiri spp. and entire scopula I in small (Genus 2 spp.). Oligoxystre spp. The fourth cladogram (Fig. 4), which is a DISCUSSION consensus of two trees, refers to the tarsal Anterior-posterior gradation.ÐSome scopula character deactivated in the matrix. characters (e.g. number of spines, develop- The only Ischnocolinae representatives that ment of teeth on the paired tarsal claws) show formed a monophyletic group are (Oligoxystre an anterior-posterior gradation that was de- spp.ϩIschnocolus algericus). scribed by Raven (1985). Concerning the tar- The tarsal scopula I condition did not show sal scopula, many species of Ischnocolinae any relation to spider size (t ϭϪ0.80433; P ϭ present different states on legs I±IV, where 0.438247) (graphic 1). Divided tarsal scopula there is a tendency towards the anterior legs

Figures 7±8.Ð7. Unilobular spermathecae (Catumiri argentinense). 8. Multilobular spermathecae, arrow showing the lateral lobe (Oligoxystre sp4). GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 465 presenting the apomorphic state. If the ple- role of this character in theraphosid evolution siomorphic state (divided scopula) is present remains obscure. PeÂrez-Miles (1994) ex- on leg I, then legs II±IV always show the plained that the scopula of tarsus IV was the same state. If the apomorphic state (entire only one used in order to avoid ambiguity, scopula) is present on leg I, legs II±IV may since scopula division width increases on the or may not have the scopulae entire. The op- hind legs (Raven 1985). It can be supposed posite happens if we observe the state of the that the tarsal scopula condition within Ther- tarsal scopula on leg IV: if scopula IV is di- aphosinae is either all legs with scopula di- vided, it can be entire or divided on the an- vided or entire. Species of the genera Hapal- terior legs. If scopula IV is entire, all the an- opus Ausserer 1875 and Homoeomma terior legs will present the apomorphic entire Ausserer 1871 have all tarsal scopula divided state. Moreover, if all the tarsal scopulae are (pers. obs.). divided on all legs, then leg IV will present Ischnocolinae is a very problematic group the widest band of setae dividing the scopula. that lacks synapomorphies (Raven 1985) and Character dependency.ÐConsidering the its genera and species are mostly recognized ontogenetic differentiation (Gerschman de Pi- by sexual characters (e.g. spermatheca and kelin & Schiapelli 1973), the anterior-poste- bulb morphology; presence, absence and mor- rior gradation (Raven 1985) and the variabil- phology of structures of the tibial apophysis). ity of states (described above) of the tarsal So far, Ischnocolinae is considered a paraphy- scopula in the Theraphosidae it is reasonable letic subfamily and the results presented in to conclude that there is a dependency of this this paper show that at least part of this group character between the legs. As such, a poste- is supported by having more than half of the rior tarsal scopula will not become entire un- metatarsus IV occupied by scopula. The main less the anterior one is entire during the on- dif®culty to infer phylogenetic hypotheses for togenesis and an anterior tarsal scopula will Ischnocolinae is the reduced number of char- not be divided if the posterior one is entire. acters that can be de®ned, since these An ontogenetic series composed of the exu- have a very homogeneous morphology. Dif- viae of a specimen of G. actaeon showing this ferent from Theraphosinae, Ischnocolinae sequenced transformation and all the combi- shows a great diversity of tarsal scopula nations found within the Ischnocolinae (Fig. states. If the divided scopula is related to 5) con®rms this character dependency. small sized species, the large ones would be Tarsal scopula as a character.ÐPeÂrez- more likely to present the scopula I entire, Miles (1992) used the tarsal scopula as a char- which does not happen in Ischnocolinae (Fig. acter on a preliminary cladistic analysis for 9). Since the tarsal scopula condition is not Theraphosinae. In this analysis the plesiom- related to spider size in Ischnocolinae (t orphic state was ``at least one of the tarsal ϭϪ0.80433; P ϭ 0.438247), this character scopula divided''. It means that if a certain might have an important role in ischnocoline species has the tarsal scopula IV divided, it phylogenetics. This importance is evident would be coded as plesiomorphic and the state when cladogram 4 is analyzed: monophyletic of scopula I would be ignored. It would be groups like (S. longibulbiϩPterinochilus mu- interesting to study the condition of the ante- rinus) are based on characters that are very rior tarsal scopulae in the species that have the variable (clypeus and curvature of metatarsus scopula IV divided. I of males); the monophyletic group ((E. vul- The use of the tarsal scopula in phyloge- pinusϩV. vellutinus)(R. annae(Tapinauchenius netics was discussed by PeÂrez-Miles (1994). sp.ϩA. avicularia)) has R. annae (Trichopel- According to him, results presented in that pa- matinae) as the sister-group of Aviculariinae. per questioned the use of this character in cla- This does not agree with the basal position of distic analysis since a close relation between Trichopelmatinae within Theraphosidae pro- body size and scopula condition (small sized posed by Raven (1994). Furthermore, the po- species tend to possess divided scopula), in sition of R. annae within this group does not Theraphosinae and Harpactirinae adults, could agree with the monophyly of Theraphosinae suggest a functional adaptation or a develop- ϩ Aviculariinae proposed by Lucas (et al. mental effect. However it is admitted that the 1991) and PeÂrez-Miles (1992). 466 THE JOURNAL OF ARACHNOLOGY

Figure 9.ÐCarapace length and tarsal scopula I condition for some Ischnocolinae species.

It was possible to show all the variability ACKNOWLEDGMENTS of the tarsal scopula condition with the six I would like to thank Dr Ricardo Pinto-da- state character. There was no mask on the var- Rocha and Cristina Anne Rheims for their iation of a character using the synthetic code, valuable suggestions on the manuscript and contrary to Pogue & Mickevich (1990). The the following curators for loan of specimens: difference between cladograms 1 and 2 is that The material examined belongs to the follow- the second one provided no homoplasies for ing institutions (abbreviations and curators in the tarsal scopula character. Another differ- parentheses): A.D. Brescovit (IBSP); N.I. ence between the two optimizations is that the Platnick (AMNH); C. Sciocia (MACN); L.A. ®rst cladogram has more synapomorphies on Pererira (MLP); R. Pinto-da-Rocha (MZSP); the nodes mentioned above. Since the char- N. Scharff (ZMUC); A.B. Bonaldo (MPEG). acter dependency is admitted, these synapom- I also thank Dr Robert Raven for valuable ophies (characters 22, 23, 24) might be false. comments and suggestions on the manuscript. The use of the tarsal scopula condition as six Thanks also to the colleagues of the Labora- unordered states of only one character admits tory of Arachnology (LAL) of the Instituto de that the transformation from divided to entire BiocieÃncias da Universidade de SaÄo Paulo for scopula is independent in all pairs of legs. If suggestions on the study. this is true, we could ®nd a spider with divid- ed anterior scopula and entire posterior scop- LITERATURE CITED ula, or even scopula I and II entire, scopula Ausserer, A. 1871. BeitraÈge zur Kenntniss der Ar- III divided and scopula IV entire. The depen- achniden-familie der Territelariae Thorell (My- dency of tarsal scopula condition between the galidae Autor.). Verhandlungen der Zoologisch- legs means that the more entire the scopula Botanischen Gesellschaft in Wien 21:177±224. the more apomorphic conditions (states) it Bertani, R. 2000. Male palpal bulbs and homolo- will show. From these results it is possible to gous features in Theraphosinae (Araneae, Ther- conclude that since there is relation between aphosidae). Journal of Arachnology 28:29±42. Bertani, R. 2002. Morfologia e evolucËaÄo das cerdas tarsal scopula condition and spider size, this urticantes em Theraphosidae (Araneae). Tese de character should be used in cladistic analysis. Doutorado, Instituto de BiocieÃncias, Universida- Additional Ischnocolinae taxa must be includ- de de SaÄo Paulo. ed in this analysis in order to provide a better Coyle, F.A. 1985. Observations on the mating be- knowledge of this character. havior of the tiny mygalomorph spider, Microh- GUADANUCCIÐTARSAL SCOPULA SIGNIFICANCE 467

exura montivago Crosby & Bishop (Araneae, Di- tematic signi®cance. Journal of Arachnology 22: pluridae). Bulletin of British Arachnological 46±53. Society 6:328±330. PeÂrez-Miles, F., S.M. Lucas, P.I. da SilvaJr., & R. Eberhard, W.G. 1985. Sexual selection and Bertani. 1996. Systematic revision and cladistic genitalia. Harvard University Press, Cambridge, analysis of Theraphosinae (Araneae: Theraphos- Massachusetts. 244pp. idae). Mygalomorph 1:33±68. Gerschman de Pikelin, B.S. & R.D. Schiapelli. Pocock, R.I. 1897. On the spiders of the suborder Mygalomorphae from the Ethiopian Region, 1973. La subfamilia ``Ischnocolinae'' (Araneae: contained in the collection of the British Muse- Theraphosidae) Revista del Museo Argentino de um. Proceedings of the Zoological Society of CieÃncias Naturales ``Bernardino Rivadavia'', En- London 1897:724±774. tomologia 4:43±77. Pogue, M.G. & M.F. Mickevich. 1990. Character Goloboff, P.A. 1993. A reanalysis of mygalomorph de®nitions and character state delineation: the spider families (Araneae). American Museum beÃte noire of phylogenetic inference. Cladistics Novitates 3056:1±32. 6:319±361. Guadanucci, J.P.L. 2004. Description of Catumiri n. Raven, R.J. 1985. The spider infraorder Mygalo- gen. and three new species (Theraphosidae: Is- morphae (Araneae): cladistics and systematics. chnocolinae). Zootaxa 671:1±14. Bulletin of the American Museum of Natural Jackson, R.R. & S.D. Pollard. 1990. Intraspeci®c History 182:1±180. interactions and the function of courtship in my- Raven, R.J. 1994. Mygalomorph spiders of the Bar- galomorph spiders: a study of Porrhothele anti- ychelidae in Australia and the western Paci®c. podiana (Araneae: Hexathelidae) and a literature Memoirs of the Queensland Museum 35:291± 706. review. New Zealand Journal of Zoology 17: Rovner, J.S. 1978. Adhesive hairs in spiders: be- 499±526. havioral functions and hydraulically mediated Lucas, S., P.I. da SilvaJr. & R. Bertani. 1991. The movement. Symposium of the Zoological Soci- genus Ephebopus Simon, 1892. Description of ety of London 42:99±108. the male of Ephebopus murinus (Walckenaer Rudloff, J.P. 1997. Revision der Gattung Holothele 1837). Spixiana 14:245±248. Karsch, 1879 nebst Aufstellung einer neuen Gat- Nixon, K.C. & J.M. Carpenter. 1993. On outgroups. tung Stichoplastoris gen. nov. (Araneae, Thera- Cladistics 9:413±426. phosidae) und Wiedereinsetzung einiger weiterer Page, R.D.M. 2001. Nexus Data Editor for Win- Gattungen der Mygalomorphae. Arachnologisch- dows, version 0.5.0. Computer program available es Magazin 5(2):1±19. at http://taxonomy.zoology.gla.ac.uk/rod/NDE/ Smith, A.M. 1990. Baboon Spiders. of nde.html Africa and the Middle East. Fitzgerald Publish- PeÂrez-Miles, F. 1992. AnaÂlisis cladistico preliminar ing, London. Vol, F. 2001. Description de Pseudoligoxystre bo- de la subfamilia Theraphosinae (Araneae; Ther- livianus sp. et gen. n. (Araneae: Theraphosidae: aphosidae). Boletin de la Sociedad Zoologica Ischnocolinae), de Bolivie. Arachnides 50:3±10. Uruguay (2Њ epoca) 7:11±12. PeÂrez-Miles, F. 1994. Tarsal scopula division in Manuscript received 17 September 2004, revised 31 Theraphosinae (Araneae, Theraphosidae): its sys- March 2005.