Phylogenetic analysis of the sharpshooter genus Subrasaca Young, 1977 (Hemiptera, Cicadellidae, Cicadellini)

Abstract The South American sharpshooter genus Subrasaca comprises 14 species. Some species of this genus are quite common in the Brazilian Atlantic Rainforest. In this paper, a phylogenetic analysis of Subrasaca, based on a matrix of 20 terminal taxa and 72 morphological characters of the head, thorax, and male and female genitalia, is presented. The analysis yielded six equally most parsimonious trees (197 steps, CI = 0.6091, RI = 0.5722, and RC = 0.3486). The results suggest that Subrasaca is a monophyletic taxon, although the genus branch is not robust. The clade showing the highest bootstrap and Bremer scores is formed by species with longitudinal dark brown to black stripes on the forewings (Subrasaca bimaculata, Subrasaca constricta, Subrasaca curvovittata, and Subrasaca flavolineata), followed by Subrasaca atronasa + Subrasaca austera.


Introduction
The infraorder Cicadomorpha comprises three superfamilies, Cicadoidea (cicadas), Cercopoidea (spittlebugs or froghoppers), and Membracoidea (leafhoppers and treehoppers). According to Hamilton (1999), the monophyly of the Cicadomorpha is well-supported by morphological synapomorphies, including the presence of a complex filter chamber. Cryan (2005), based on molecular data (18S rDNA, 28S rDNA, and histone 3), also supports the monophyly of the Cicadomorpha and suggests the following relationships for the superfamilies: (Membracoidea (Cicadoidea, Cercopoidea)). Based both on morphological and molecular data, the monophyly of the Membracoidea is also well-supported (Evans 1963, Dietrich and Deitz 1993, Hamilton 1999, Cryan 2005. Synapomorphies of the Membracoidea include the enlarged, transverse metathoracic coxae and a pair of rod-shaped lateral apodemes associated with the scutellar suture (Dietrich and Deitz 1993).
The family Cicadellidae (leafhoppers), with over 21,000 described species placed in more than 120 family-group taxa (Oman et al. 1990, Hamilton 1999, includes many species of economic importance because they are vectors of pathogens of cultivated plants (Nielson 1985). According to the morphological phylogeny of Hamilton (1983) and the molecular phylogeny (28S rDNA) of Dietrich et al. (2001), Cicadellidae is a paraphyletic group because treehoppers (Aetalionidae and Membracidae) are derived from leafhoppers. Taxonomically, cicadellids can be distinguished from other membracoids by the mesanepisternum without a hooklike process, separated from the katepisternum by a suture, and hind tibia with setae of longitudinal rows usually large and conspicuous (Dietrich 2005). With over 2,000 known species and a cosmopolitan distribution, Cicadellinae (sharpshooters) is the third largest subfamily of the Cicadellidae (Mejdalani 1998, Takiya 2007, McKamey 2007. According to Young (1968Young ( , 1977Young ( , 1986, this subfamily is divided into two tribes, a cosmopolitan Cicadellini and a New World Proconiini. Sharpshooters feed on the low-nutrient xylem sap of vascular plants. Some species of this group are important vectors of xylem-borne phytopathogenic bacteria (Redak et al. 2004).
The studied specimens belong to the following institutions: Departamento de Entomologia, Museu Nacional, Universidade Federal do Rio de Janeiro (MNRJ, Rio de Janeiro); Coleção Entomológica Prof. José Alfredo P. Dutra, Departamento de Zoologia, Instituto de Biologia, Universidade Federal do Rio de Janeiro (DZRJ, Rio de Janeiro); and Coleção de Entomologia Pe. Jesus S. Moure, Departamento de Zoologia, Setor de Ciências Biológicas, Universidade Federal do Paraná (DZUP, Curitiba). The number of specimens examined of each terminal taxon, their geographical distribution, and collections are listed in Table 1.

Techniques for preparation of specimens and terminology
The techniques for preparation of male and female genital structures follow Oman (1949) and Mejdalani (1998), respectively. The dissected parts are stored in small vials with glycerin, as suggested by Young and Beirne (1958). The first and second pair of valvulae of the ovipositor were mounted on temporary slides with glycerin. The descriptive terminology adopted herein follows mainly Young (1977), except for the facial areas of the head (Hamilton 1981, Mejdalani 1993, 1998 and the female genitalia (Nielson 1965, Hill 1970).  Young (1977) and Wilson et al. (2009).

Cladistic analysis
Morphological characters of the head, thorax, male and female genitalia were included in the unpolarized matrix (Nixon and Carpenter 1993), which was assembled using the Nexus Data Editor (Page 2001). Hypotheses of primary homology were proposed based on the topological identity of the structures (Pinna 1991). All characters were initially scored equal weights. Character states were scored as underscores (_) when inapplicable or as question marks (?) when unavailable. The heuristic search algorithm, as implemented in PAUP* 4.0 (Swofford 2002), was employed for searching the most parsimonious trees. The successive weighting procedure (Carpenter 1988(Carpenter , 1994 was based on the maximum rescaled consistency index (rc) of the characters (Farris 1969(Farris , 1989. The strict consensus method was employed for all original most parsimonious trees. Clade support was estimated by computing 10.000 bootstrap replicates (Felsen-stein 1985) with heuristic search in PAUP* 4.0 and by decay indices (Bremer 1988(Bremer , 1994 in TreeRot 3.0 (Sorenson and Franzosa 2007). Autapomorphic characters were included in the matrix, as suggested by Yeates (1992), but we provide consistency index (CI) values considering all characters as well as only the informative ones.

Results and discussion
The data matrix (Table 2) consists of 72 morphological characters, 35 of the external morphology, 25 of the male genitalia, and 12 of the female genitalia. Among these characters, 51 are binary and 21 are multistate, being 52 informative for the parsimony analysis. The characters, their states, and ci greater than 0.5 are listed below. Although many of the characters are based on color patterns, these are consistent intraspecifically in Subrasaca. Figures 1 (external morphology and male genitalia) and 2 (female genitalia) provide some examples of characters employed in the phylogenetic analysis.

3.
Color of face: (0) black; (1) yellow; (2) light brown; (3) black with yellow central region enclosing black macula; (4) black with cream central macula and two orange maculae on anterior portion; (5) black with orange macula on posterior portion; (6) black with orange macula on anterior portion; (7) yellow with black Y-shaped macula; (8) yellow with black central portion enclosing yellow macula; (9) cream with black central portion and anterior region with orange macula; (A) yellow with brown streaks and median stripe; (B) black with orange lateral portions; (C) yellow with small black maculae on anterior portion. ci = 0.9.

Main aspects and discussion of the phylogenetic analysis
The  Fig. 3a.
The successive weighting analysis yielded one tree, which is also one of the six original trees, with length = 80, CI = 0.8249 (excluding uninformative characters = 0.7199), RI = 0.7641, and RC = 0.6303 (Figs 3b, 4). Thirty-two characters had maximum weight (= 1.0) and 40 had lower weights. Twenty characters were parsimony-uninformative. Figure 4 gives bootstrap estimates (when > 50%) and Bremer support indices for the clades recovered under equal weights. Apomorphies of this tree are given in Table 3.
Other phylogenetic (e.g., Felix and Mejdalani 2011) or purely taxonomic (e.g., Mejdalani et al. 2014) studies on the Cicadellini highlighted the need for more precise definitions of various genera of this tribe. In the introduction of his impressive monograph on the New World Cicadellini, Young (1977: 10) expressed his perception of this problem as follows: "The Cicadellini are an intricate group. Their morphology suggests rapid radiation and often shows small discontinuities compared with those found in many of the Proconiini." Small discontinuities are precisely what we have observed between Subrasaca and the genera here employed as outgroups. In any case, the cladistic analysis allowed us to propose a more objective definition of the genus.
Two clades appeared in all six most parsimonious trees and were fairly robust in the analysis (Fig. 4). The clade formed by the species with longitudinal dark brown to black stripes on the forewings (S. bimaculata, S. constricta, S. curvovittata, and S. flavolineata) had the highest percentage of bootstrap (= 83%) and was supported by seven apomorphic conditions (Table 3, node 26), including the conspicuous set of dark brown to black longitudinal stripes on the forewings (character 25, state 1; Fig. 1b, c). The Bremer support of this clade was 3. It shows the following internal relationships in all trees (Fig. 3a): (S. constricta (S. bimaculata (S. flavolineata, S. curvovittata))). These four species were described in detail by Silva et al. (2013a). The group is distributed in the Atlantic Forest from northeastern (state of Bahia) Table 3. Apomorphy list for clades of Fig. 4 of the phylogenetic analysis of Subrasaca and outgroup taxa. Non-homoplastic characters are in bold. to southern Brazil (state of Paraná). The second clade, formed by S. atronasa + S. austera, is supported by nine apomorphic conditions (Table 3, node 33), including a whitish-yellow transverse band on middle portion of pronotum (character 14, state 1; Fig. 4, node 33). This clade also had relatively high bootstrap (= 82%) and Bremer (= 2) scores. These two species, which are known only from the state of Santa Catarina (Zanol and de Menezes 1982) in southern Brazil (Atlantic Forest), were described by Young (1977), who considered them "very close" to each other (Young 1977: 479). Although with low support scores, the clades formed by S. ignicolor + S. diminuta and S. nigriventris + S. rachelae were recovered in all most parsimonious trees (Fig.  4, nodes 23 and 30, respectively). Unlike the species with longitudinal dark brown to black stripes on the forewings (node 26), those with contrasting orange marks (S. diminuta + S. ignicolor, S. nigriventris + S. rachelae, and S. rubra) did not form a monophyletic group in any of the six most parsimonious trees.  Table 3. Most sharpshooter images from Wilson et al. (2009).