Phylogeny of Dynastor and Brassolis butterflies (Lepidoptera: Nymphalidae): a tough nut to crack Author Garzón-Orduña, Ivonne J. Author Penz, Carla M. text Zootaxa 2009 2134 1 22 journal article 10.5281/zenodo.188448 10d3b521-818c-44d2-af1b-df4281f82b80 1175-5326 188448 Brassolis astyra Godart ( Fig. 3 F–H; 8D, I; 9D) Diagnosis: Recognized by a combination of five characters: (1) in both sexes, dorsal FW postmedial orange band clearly bifurcated anteriorly (above M3). In females this band is faded above M3 and inside the discal cell; (2) specimens from the northern end of the distribution ( astyra astyra ) lack a dorsal HW postmedial band, but this band varies from faded to well developed in specimens from southern localities ( astyra philocala ). Dorsal HW postmedial band generally narrower and more jagged in B. astyra philocala than in B. sophorae and B. dinizi ; (3) in both sexes, abdomen orange dorsally from third segment onward; (4) in posterior view, male valva usually with a small point at apex ( Fig. 8 D); (5) in ventral view, anterior section of the female sterigma interrupted at mid point, sclerotized ‘loop’ originates from a fold closer to the midline than in other species. Distribution: Southeastern Brazil ( Casagrande 2004 ). Subspecies: Casagrande (2004) listed two subspecies: nominal astyra ( type locality Brazil ), and philocala Stichel ( Brazil , Rio Grande do Sul). Cladistic analysis Implied weights searches under K=1 and 2 found two most parsimonious trees. Higher values of K (K=3 to 20) consistently yielded the same three most parsimonious trees, one of which is the topology of the strict consensus of these three trees ( Fig. 4 A). The other two are the same trees found under K=1 and 2 ( Fig. 4 B, C). Trees had a fit of 5.2. The three most parsimonious trees show Dynastor and Brassolis as monophyletic sister genera, but differed in the relative positions of B. sophorae , B. haenschi , B. dinizi , and B. isthmia , which had an effect on the resolution of the strict consensus (compare topologies in Fig. 4 ). Unambiguous character transformations for the nodes and terminals were optimized in the strict consensus tree using “hard” polytomy option and are described below ( Fig. 4 A). Four synapomorphies supported the Dynastor + Brassolis clade: last segment of labial palpus shorter than the distal width of second segment (character 1:2, Fig. 8 B), reversed to state 1 in B. haenschi (approximately the same length as distal width of second segment); valva not projecting beyond uncus tip (30:1, Fig. 8 E); groove at the articulating point between valva and appendices angulares limited to the base of valva (34:1); distal opening of phallus located dorsolaterally (38:1). Dynastor + Brassolis had a low Relative Bremer support and a moderate value of jackknife recuperation (30 and 59; Fig. 4 A). The monophyly of the genus Dynastor was supported by seven synapomorphies and one homoplasious transformation. The synapomorphies were: dorsal forewing costal margin striped (4:1, Fig. 7 D), dorsal forewing postmedial band broken (8:1, Fig. 7 D), female foreleg distal tarsomere with ventral spines between the rows of sensilla (24:1, Fig. 9 A), dorsal edge of juxta continuous (37:1), sclerotized tip of phallus forming a small coecum that projects beyond the distal opening (44:1, Fig. 8 H), lateral uncus wings merging to form a single dorsal keel (47:1) and distal portion of uncus tall (48:1, Fig. 8 G). The homoplasious transformation was the ventral outline of phallobase evenly arched (45:0, Fig. 8 H). Relative Bremer support and Jackknife recuperation values for Dynastor were high (100 and 99; Fig. 4 A). We found D. darius and D. napoleon to be sister taxa, and two homoplasious transformations supported this clade: the presence of midleg tibial spurs (21:1), also present in N. panniculus ; and last segment of labial palpus approximately the same length as distal width of second segment (1:1, Fig. 8 C), present also in B. haenschi . This clade had moderated values of Relative Bremer support and jackknife recuperation (50 and 52; Fig. 4 A). The monophyly of Brassolis was supported by 13 synapomorphies and three homoplasious transformations. The synapomorphies of Brassolis were: absence of sensilla styloconica at the proboscis tip (2:0, Fig. 3 E); in males, basal third of forewing costal margin slightly concave (3:1, Fig. 7 A); male dorsal forewing postmedial band extended through discal cell (6:0, Fig. 7 B); presence of a sent organ adjacent to male hindwing vein Cu2 (15:1, Fig. 7 A), in males, absence of a dorsal hindwing marginal line (18:0, Fig. 7 A); in males, presence of dashed pattern of the distal portion of ventral hindwing discal cell (19:1, Fig. 7 C); presence of an anterolateral constriction of the tegumen (26:1, Fig. 8 D); internal surface of valva, basal region conspicuously less sclerotized than distal region (33:1); juxta not reaching the edge of the anellus (36:0, Fig. 8 I); presence of peg-like setae on semi-sclerotized lobe on left side of phallus (41:1, Fig. 8 K); absence of gnathos (49:0); fully sclerotized antrum (55:2); and posterior lobes of the papillae annales more developed ventrally (57:2, Fig. 9 H). The homoplasious transformations were: male midleg tarsus longer than tibia (22:0); presence of a semi-sclerotized lobe on the left side of phallus (40:1), which was also found in D. darius and in one of the outgroup taxa ( Elymnias hypermnestra ); and the transition between pedunculum and vinculum being straight (28:0, Fig 8 F). Relative Bremer support and Jackknife recuperation values for Brassolis were high (81 and 100; Fig. 4 A). Poor resolution of the strict consensus tree reflects the overall morphological similarity within Brassolis ( Fig. 4 A). Brassolis sophorae , B. dinizi , B. isthmia and B. haenschi formed an unresolved clade (sophoraeclade) supported by one homoplasious transformation: presence of a dorsal hindwing postmedial band in males (17:1, Fig. 7 A). Alternative groupings of these species were represented in the other two trees ( Fig. 4 B and C). In all resolutions, however, B. granadensis was the sister taxon of this clade, supported by one homoplasious transformation: the presence of long setae on a semi-sclerotized lobe on the left side of phallus (43:1, Fig. 8 L). Finally, B. astyra appears as sister to all other Brassolis . To identify characters that produced conflicting topologies within the sophorae-clade, we did a search without collapsing zero-length branches, and optimized both ambiguous and unambiguous transformations onto the obtained trees. The results were five alternative topologies for the sophorae-clade ( Fig. 5 ). In these five resolved sub-trees all characters producing ambiguous optimizations (11, 12, 14, 16, 20, 25, 54) but one (54) were coloration characters, and all but one (12) were polymorphic. Trees in Fig. 5 show B. sophorae as sister to each of the species in the sophorae-clade ( Fig. 5 C–E) or as sister to species groups within this clade ( Fig. 5 A–B). Given that polymorphic characters supported ambiguous optimizations within the sophorae-clade, an analysis was done excluding nine polymorphic characters (Appendix 3). Characters 11, 13, 14, 16, 20, 25, 43 and 54 were polymorphic for B. sophorae , and character 17 was polymorphic for B. isthmia , and also for B. astyra (outside the sophorae-clade). Two trees were found that differed greatly in resolution, and the strict consensus of these trees showed the sophorae-clade as a polytomy. We therefore concluded that the exclusion of polymorphic characters did not improve our results (trees not illustrated). As noted above, B. sophorae was polymorphic for eight characters, and we therefore asked whether splitting B. sophorae into monomorphic terminals would improve tree resolution and would provide a more reliable assessment of relationships. To investigate this, we used male-female pairs (whenever possible) collected in eight countries as terminals. This sampling attempted to represent the type locality (country) of most valid subspecies listed by Casagrande (2004) , therefore approximating a range of character variation of B. sophorae . Separating B. sophorae into eight terminals (putative subspecies) showed that all species of Brassolis are associated with one or more subspecies of sophorae ; i.e., the sophorae-clade was not recovered ( Fig. 6 ).