Zoological Journal of the Linnean Linnean Society Society,, 2017,2016. 179 With, 767–864. 63 figures With 63 figures Phylogeny of Sicariidae spiders (Araneae: Haplogynae), with a monograph on Neotropical Sicarius IVAN L. F. MAGALHAES1,2*, ANTONIO D. BRESCOVIT3 and ADALBERTO J. SANTOS2 1Division� Aracnolog�ıa, Museo Argentino de Ciencias Naturales ‘Bernardino Rivadavia’, Av. Angel� Gallardo 470, C1405DJR, Buenos Aires, Argentina 2Departamento de Zoologia, Instituto de Ciencias^ Biologicas,� Universidade Federal de Minas Gerais, Av. Antonio^ Carlos, 6627, Belo Horizonte, Minas Gerais, CEP 31270-901, Brazil 767 3Laboratorio� Especial de Coleçoes~ Zoologicas,� Instituto Butantan, Av. Vital Brazil, 1500, Sao~ Paulo, SP, CEP 05503-900, Brazil Received 2 September 2015; revised 29 March 2016; accepted for publication 30 March 2016 Sicariids are an infamous spider family containing two genera: the poorly known Sicarius Walckenaer and the medically important Loxosceles Heineken & Lowe. We present the first broad survey of the morphology of the family from a phylogenetic perspective in order to resolve its relationships. We scored morphological, behavioural and venom feature data for 38 taxa, including New and Old World species of both Sicarius and Loxosceles and three outgroups. Our results point to the monophyly of Sicariidae and its two genera as currently delimited, with the identification of novel synapomorphies for all of them. We present evidence of a group of ‘true’ Loxosceles composed of all members of the genus except those from the spinulosa species group. Sicarius have a very interesting phylogenetic structure, with species from the Americas and Africa forming reciprocally monophyletic groups. Thus, we resurrect Hexophthalma Karsch to accommodate African Sicariinae species. We discuss the evolution of venom proteins, spinning organs and cryptic appearance in Sicariidae. We delimit Sicarius to include only the 20 Neotropical species, and present a taxonomic revision of the genus. We redescribe nine previously known species, elevate one subspecies to species, and describe six new species from Argentina, Bolivia, Brazil and Peru. Sicarius levii sp. nov., from Argentina, is probably the largest known haplogyne spider in both body length and mass. Six specific names are considered junior synonyms of older names. All these species are illustrated and have their distributions mapped. We present an interactive key for identifying Sicarius species. © 20172016 The Linnean Society of London, Zoological Journal of the Linnean Society, 20162017 doi: 10.1111/zoj.12442 ADDITIONAL KEYWORDS: crypsis – evolution – morphology – new species – revision – Scytodoidea – Sphingomyelinase D – systematics – taxonomy – venom. which are moderately diverse (~130 species) and INTRODUCTION more diversified in Africa and the Americas (World Sicariidae Keyserling stands as one of the most infa- Spider Catalog, 2015). mous spider families, as it contains the medically Sicariids are currently allocated in Scytodoidea, an important genus Loxosceles Heineken & Lowe – the ancient group of true spiders with simple genitalia brown, recluse or violin spiders. The limits of the (Garrison et al., 2016). The monophyly of Sicariidae family have changed since the classical work of has been supported by morphological data (Platnick Simon (1893), and now it is thought to contain only et al., 1991; Ramırez, 2000; Labarque & Ramırez, two genera, the other one being Sicarius Walcke- 2012), although in these analyses only a few exem- naer, known as six-eyed sand spiders. They are hap- plars have been included. The only comprehensive logyne, medium- to large-sized, sedentary spiders, phylogenetic hypothesis for the family is that of Bin- ford et al. (2008), who presented DNA sequence data of a broad sampling of species of both Sicarius and *Corresponding author. E-mail: [email protected] Loxosceles. Their analysis recovered the monophyly © 20172016 The Linnean Society Society of of London, London, ZoologicalZoological Journal Journal of of the the Linnean Linnean Society Society, 2017,, 2016 179, 767–864 7671 7682 I. L. I. F.L. MAGALHAESF. MAGALHAESET ET AL. AL. SICARIID PHYLOGENY AND REVISION OF SICARIUS 3 of Sicarius and Loxosceles, but showed contentious particles (Duncan, Autumn & Binford, 2007), confer- hexamethyldisilazane (HMDS) or critical-point dried reconnection, retaining up to 1000 trees per replica- support for a monophyletic Sicariidae. As discussed ring a characteristic cryptic appearance to these spi- (fragile parts, such as spinnerets and female genital- tion. Branch supports were estimated by Bremer by them, this result may in part have been caused ders. Additionally, their unique pot-shaped eggcases ia). Dried parts were placed in aluminium stubs (1994) support (using a script that comes with the by the ancient age of the group – Sicariids are are also covered by dirt (Levi & Levi, 1969; Aguilar using adhesive copper tape or carbon stickers, sput- TNT package) and symmetric resampling (Goloboff thought to have started diversifying before the &Mendez,� 1971; Magalhaes~ et al., 2013). Their ter-coated with 10 nm of gold or gold–palladium, and et al., 2003) (estimated based on 1000 pseudorepli- break-up of western Gondwana. Furthermore, Bin- venom, like that of Loxosceles, contains sphin- examined under JEOL JSM-6360L, Quanta 2000 or cates, with a 33% probability of change). Character ford et al. (2008) found evidence that African and gomyelinase D, an enzyme responsible for necrotic Philips FEI XL30 TMP scanning electron micro- optimizations were made in Winclada 1.0.8 (Nixon, Neotropical Sicarius are reciprocally monophyletic. activity. However, the potential activity of the scopes. 1999–2002) and Mesquite 2.75. However, to date no morphological work has been enzyme varies among Sicarius species (Binford et al., Many of the characters used in the analysis have Bayesian inference was carried out in MrBayes 3.2 done to test this hypothesis, or to search for potential 2009; Lopes et al., 2013), and at least some species been taken from the literature (e.g. Platnick et al., (Ronquist et al., 2012) using the Mkv model of evolu- synapomorphies of these clades. This is especially are apparently incapable of causing strong symptoms 1991; Ram�ırez, 2000, 2014; Labarque & Ram�ırez, tion (Lewis, 2001) accounting for rate heterogeneity interesting considering that the first African Sicari- in vertebrates (e.g. Alegre, Meneses & Aguilar, 2012). Morphological scorings were made de novo for among characters (Γ parameter). Two independent inae were described in a separate genus, Hexoph- 1977). They are restricted to xeric habitats, espe- all taxa based on examination of actual specimens, runs of four Markov chains each (one cold and three thalma Karsch, 1879, which was considered a junior cially deserts and tropical dry forests (Magalhaes~ except for L. rufescens (Dufour), scored after the heated) were run for 107 generations, sampling synonym of Sicarius by Simon (1893) and subsequent et al., 2013), and are apparently poor dispersers richly illustrated work of Labarque & Ram�ırez parameters every 103 generations, and discarding authors. Inferring the phylogeny of Sicariidae based (Magalhaes et al., 2014), making them good models (2012); this work was also used to guide the scoring 25% of the initial sampled states as burn-in. Conver- on phenotypic data is desirable for several reasons. for studying the biogeography of these biomes. How- of some cells for the non-sicariid outgroups. Beha- gence of the independent runs and stationarity of First, this would be an independent test of the ever, the lack of taxonomic resolution for the vioural scorings are of our own field observations, the parameters were checked using Tracer 1.6 (Ram- topologies obtained by Binford et al. (2008), and Neotropical species hampers studies about those except when noted. Characters related to venom fea- baut et al., 2014). Branch supports were estimated might improve resolution for some of their clades. interesting aspects of their biology. tures are scored after the findings of Binford et al. as the frequencies of clades in a majority-rule con- Second, morphological phylogenies often reveal Our study has two objectives. We present the first (2009) and Lopes et al. (2013). sensus of the trees sampled in the stationary phase synapomorphies that help in placing taxa for which phylogeny of Sicariidae based on phenotypic data, of the Markov chains. For this analysis, only discrete no molecular data are available, such as fossils; this including morphology, behaviour and venom fea- characters have been used due to limitations of the PHYLOGENETIC ANALYSIS is especially relevant in the context of estimates of tures. This was based on the examination of species Mkv model. divergence times using fossil calibrations. Finally, spanning most of the geographical and morphological Our taxon sampling includes all Neotropical Sicar- phenotypic phylogenies generate a wealth of data on diversity of both Loxosceles and Sicarius, but focus- ius, six African Sicariinae, nine Loxosceles chosen to TAXONOMY morphology as a by-product (see Labarque & ing particularly in the latter genus. The second goal represent the morphological and geographical diver- Ram�ırez, 2012; Ram�ırez, 2014), which might reveal is to present a complete revision of Sicarius from the sity of the genus, and three outgroup taxa in the The format of description follows Magalhaes~ et al. interesting phylogenetic or functional patterns. Neotropical