This research sheds light on the phylogenetic position of the recently described Caribbean species Duartettix montanus. Morphologically most similar to the Melanoplinae, especially the North American genus Melanoplus, it was assigned to that subfamily. The Orthoptera Species File, curiously, assigns Duartettix to the South American tribe Dichroplini. The present molecular phylogenetic analysis of portions of four mitochondrial genes, however, strongly rejects that allocation and instead supports an association with the northern melanoplines. Within the context of an “Out-of-South America” hypothesis proposed earlier for the subfamily's origin, it is speculated that the antecedents of Duartettix arose from ancestors on their northward movement, traversing a series of island arcs that intermittently connected the two Americas during the late Cretaceous. Of possible taxonomic interest, phylogenetic information is also provided for a number of South American melanoplines, notably members of the tribe Jivarini, which have not been previously analyzed using molecular methods.
Duartettix montanus, a recently described genus and species (Perez-Gelabert & Otte 2000), presents some interesting taxonomic difficulties. The species occurs in the high mountainous valleys of the Dominican Republic and, apparently, nowhere else. Morphologically, the species bears a superficial resemblance to North American melanoplines, specifically the genus Melanoplus; yet, it resembles no one species overall. Perez-Gelabert & Otte (2000) point to the species' geographic proximity to the North American continent as further evidence of its probable link with the northern melanoplines. For reasons not entirely clear, the Orthoptera Species File (OSF2) assigns Duartettix to the South American tribe Dichroplini (Otte et al. 2006). This note seeks to clarify the phylogenetic position of Duartettix in relation to a selection of melanopline grasshoppers distributed in North America, South America and Eurasia. The Neotropical taxa encompass material from two major tribes, Dichroplini and Jivarini. Including the latter may be significant because most members are also adapted to high altitudes (Rowell & Carbonell 1977). This study also includes the South American species, Apacris rubritorax, whose tribal affiliation is presently unknown (Amédégnato et al. 2003).
Materials and Methods
Species along with sources are listed in Table 1. Included are 19 South American species, 12 North American species, 4 Eurasian species and 1 specimen of Duartettix montanus. DNA was extracted using either the DTAB/CTAB method (Philips & Simon 1995) or the QIAGEN DNeasy tissue kit (Mississauga, Canada). Portions of the mitochondrial genes cytochrome b (cytb), cytochrome oxidase subunits I and II (COII and COII), and NADH dehydrogenase subunit II (ND2) were PCR amplified and sequenced. Parsimony, weighted parsimony analysis, maximum likelihood and Bayesian analyses were performed using the programs PAUP* (Swofford 2001) and MrBayes (Huelsenbeck & Ronquist 2001). Application of the program Modeltest (Posada & Crandall 1998) identified the model GTR+G+I as best fitting the data and accordingly this model was used in ML searches. Schistocerca gregaria, (Forskål) a member of the subfamily Cyrtacanthacridinae, served as the outgroup. Levels of branch support were estimated through 1000 bootstrap replicates using parsimony and by calculating Bayesian posterior probabilities (PP). Further procedures concerning molecular and data manipulation can be found elsewhere (Litzenberger & Chapco 2003).
Species analyzed, location and GenBank accession numbers of mtDNA sequences.
Results and Discussion
Sequence data consisting of a maximum of 1716 bases have been deposited in GenBank (Table 1). Across the four genes, 926 sites were variable and of these, 538 were phylogenetically informative. ML yielded a single best tree with -LnL = 15658.84. Maximum resolution was achieved using Bayesian methods and weighted parsimony following Farris's (1969) iterative reweighting scheme and by counting transversions at third codon positions only. Figs 1A and 1B depict relationships uncovered by these two methods. The Bayesian approach considered a variety of models and the one that emerged with the highest likelihood was also GTR+G+I. The two techniques generally yielded the same broad associations with different levels of support.
Duartettix.— The base composition of mitochondrial DNA in Duartettix consists of 35.1% (A), 16.1% (C), 14.3% (G), and 34.4% (T), well within the range of all melanoplines thus far examined (Litzenberger 2002). Contrary to its tribal designation in OSF2, the species is not phylogenetically related to either South American tribe, Dichroplini or Jivarini. Instead, Duartettix is very strongly associated with the northern melanoplines, a result that supports Perez-Gelabert & Otte's (2000) viewpoint. Unfortunately, its exact placement is indeterminate. Parsimony suggests an affiliation with Dactylotum, a member of the North American tribe Dactylotini (Vickery 1997, see also Litzenberger & Chapco 2003). Both species are, in turn, connected to the clade encompassing Melanoplus, Hypochlora and Hesperotettix. Bootstrap support for these associations, however, is not large. In analyses based on the Bayesian approach, Duartettix emerges as part of an unresolved polytomy within the northern group. In the context of the “Out-of-South America” scenario proposed for the subfamily's origin (Chapco et al. 2001, Amédégnato et al. 2003), if Duartettix had branched off from ancestors on their northward movement, traversing the series of “proto-Antilles” island arcs that intermittently connected the two Americas (Pitman et al. 1993), one might expect the species to be basal to the North American-Eurasian clade and internal to the South American clade. An alternative possibility, suggested—albeit weakly—by the parsimony result, is that Duartettix may have evolved more recently, from northern melanoplines. Application of an “orthopteroid clock” places, in rapid succession, the times separating Duartettix from South American taxa and from North American taxa at approximately 78 and 73 Mya, respectively (this clock was calibrated using transversional substitutions, for which there is evidence of linear accumulation over time, and fossil data that link Caelifera and Ensifera—see references and data in Chapco et al. 2001). These times would be in accord with the first scenario, but until a better resolution of relationships within the northern taxa—Duartettix cluster—is achieved, one can only speculate on the precise sequence of events.
Remaining taxa.—The inclusion of additional South American species in the analysis leads to results which further substantiate the “Out of South America” hypothesis for the origin of the Melanoplinae, given that the southern taxa occupy a basal and paraphyletic position to the northern species. All analyses identify Jivarini as the more basal of the two South American tribes, and probably the more ancient (see also Amédégnato et al. 2003). Both tribes emerge as monophyletic assemblages.
Within Dichroplini, the grassland species Scotussa, Leiotettix, Ronderosia and Atrachelacris, comprise the “Paranense-Pampeano” group (Cigliano & Ronderos 1994). However, bootstrap support for this group's integrity is weak (54%), and support using Bayesian methods is nonexistent, although the four genera are strongly associated with four species of Dichroplus (99% PP). Dichroplus is clearly not monophyletic. Colombo et al. 2005 analyzed a greater number of species of Dichroplus, using two genes and morphology, and their investigation also found the genus to be polyphyletic. In the present work, D. vittiger and D. maculipennis consistently emerged together in all analyses. Both belong to the same species group, but to different subgroups (Otte et al. 2006). The two remaining Dichroplus species, elongatus and pratensis, were also linked, strongly in the Bayesian analysis, weakly using parsimony. Both belong to different species groups in the OSF2 (Otte et al. 2006). In the Colombo et al. 2005 investigation, these species were either part of a large unresolved polytomy when genes were analyzed, or were topologically separated when genes and morphology were analyzed in combination. Additional genes and, especially, species (see Zwickl & Hillis 2002) need to be studied to resolve relation-ships within that genus. Apacris, previously unassigned to tribe, could conceivably be regarded as part of Dichroplini.
While Jivarini is monophyletic, the genus Jivarus is not; instead, it is paraphyletic with J. antisanae occupying a position external to Urubamba and two other species of Jivarus. Using parsimony, Argemiacris is positioned externally to the remaining species, followed by Nahuella. Bayesian methods reverse those placements. According to Ronderos' (1978) morphological studies, Argemiacris is phylogenetically close to Urubamba, a finding not supported by the present analysis.
The strong connection between Barytettix and Aptenopedes, two species on opposite sides of the southern United States, is some-what surprising. Perhaps one or both genera were, in the past, distributed more widely and over time became restricted in their distributions.
It should be noted that when sequences were scrutinized for possible internal stop codons or deletions, which could signify that nuclear sequences of mitochondrial origin or “Numts” (Bensasson et al. 2000) had been amplified, all but Neopedies brunneri were free of these features. The latter possessed two deletions within the ND2 gene, one three nucleotides long and another, one nucleotide long. Nevertheless, the two Neopedies species sequences, when aligned, were linked with high bootstrap and posterior probability values.
I thank C. Amédégnato (Muséum national d' Histoire naturelle, France), C. S. Carbonell (Universidad de la República, Uruguay), M. Cigliano (Museo de La Plata, Argentina), and D. Otte (The Academy of Natural Sciences of Philadelphia, USA) for providing recently collected or museum specimens. I am grateful to D. Otte, in particular, for supplying the Duartettix montanus material. I thank E. Chapco and an anonymous reviewer for their comments on the writing. Lastly, G. Litzenberger's technical work on this project is gratefully acknowledged. This research was funded by a grant from the Natural Sciences and Engineering Research Council of Canada.