Org Divers Evol (2017) 17:409–420 DOI 10.1007/s13127-017-0325-7

ORIGINAL ARTICLE

Phylogenetic analysis of the Rambur, 1842 (: ), based on morphological characters of larvae and mitochondrial DNA sequences

Mónica Torres-Pachón1 & Rodolfo Novelo-Gutiérrez1 & Alejandro Espinosa de los Monteros2

Received: 21 September 2016 /Accepted: 27 February 2017 /Published online: 17 March 2017 # Gesellschaft für Biologische Systematik 2017

Abstract The study of the evolutionary interrelationships phylogenies; nonetheless, there is historical congruence among the encompassed in the Neotropical genus among them. Within Argia, five clades were consistently Argia (Zygoptera: Coenagrionidae) has been neglected. The recovered. Most of those clades have been identified, at least goal of this study is to infer the phylogenetic relationships in part, in previous phylogenetic and taxonomic studies. among 36 species of Argia Rambur, 1842, using complemen- Indubitably, the morphological characters from larvae have tary data sets (i.e., larval morphology and mitochondrial historical signal useful for cladistic and taxonomic inference. DNA). The morphological data set comprises 76% of the Therefore, it should be a priority to pay more attention to this larvae currently described for this genus and includes 97 source of characters. morphological characters. From those, 47 characters have not been previously used in taxonomic studies involving drag- Keywords 16S rRNA . Cytochrome oxidase subunit I . onflies’ larvae. This is the first cladistic study based on larvae Coenagrionidae . Bayesian inference . Larval morphology . morphology for species within the suborder Zygoptera. Data Parsimony . Zygoptera partitions were analyzed individually, as well as total evi- dence, using parsimony and Bayesian inference as criteria for optimal-tree selection. The results support the monophyly Introduction of the North American species of Argia. This genus can be identified by the combination of eight synapomorphies, four Within the order Odonata, Argia is the most diverse genus, of which are exclusively found in Argia. According to the with 128 described species (Garrison and von Ellenrieder optimal trees, the individual data sets (i.e., morphology and 2015; Schorr and Paulson 2016). This genus is distributed DNA sequences) have a high level of homoplasy, resulting in exclusively in the New World, with Mexico as the country soft polytomies and low support for several nodes. The with the largest number of described species (49 spp., corre- specific relationships of the terminal units differ between the sponding to 38.28% of the total species) (González-Soriano and Novelo-Gutiérrez 2007, 2014), relative to the species Electronic supplementary material The online version of this article recorded for Central America (15 spp.), USA (32 spp.), Brazil (doi:10.1007/s13127-017-0325-7) contains supplementary material, (29 spp.), and Colombia (22 spp.) (González-Soriano 2012). which is available to authorized users. Argia was described by Rambur in 1842 with only two species: Argia impura and Argia obscura. The latter was sub- * Alejandro Espinosa de los Monteros sequently considered as a junior synonym of Agrion [email protected] fumipenne (Garrison et al. 2010). To date, some 150 names of species have been proposed. From these, 128 could be 1 Department of Biodiversidad y Sistemática, Instituto de Ecología, A.C., Carretera antigua a Coatepec 351, El Haya, considered valid names (Garrison and von Ellenrieder 2015; 91070 Xalapa, Veracruz, Mexico Schorr and Paulson 2016). 2 Department of Biología Evolutiva, Instituto de Ecología, A.C., Garrison (1994) conducted a synopsis of the species within Carretera antigua a Coatepec 351, El Haya, 91070 Xalapa, Veracruz, the genus Argia for North America and the north of Mexico Mexico and admitted 29 species. However, the species of Mexico, 410 Torres-Pachón M. et al.

Central America, and South America need to be reviewed, Phylogenetic studies that have included the genus Argia since several names that appear in the bibliography cannot be associated with any species (Garrison et al. 2010). A number of phylogenetic studies at the level of family, sub- Likewise, the lack of updated keys, the existence of descrip- family, and other genera of the order Odonata have included tions scattered in the bibliography and the lack of illustrations, some species of the genus Argia. The location of these species and the great diversity of species of Argia make the identifi- changed within the topologies of trees according to the traits cation of these species a complex task (Garrison et al. 2010). (i.e., morphological, molecular, or both) and the optimality On the other hand, the of larvae is still criteria used (i.e., parsimony, maximum likelihood, and deficient, as only 36% (46 spp.) of the 128 recognized Bayesian inference) (Rehn 2003; Saux et al. 2003; Bybee species are known, with half of these having been et al. 2008; Pessacq 2008;YuandBu2011; von Ellenrieder described over the past 10 years (Pérez-Gutiérrez and 2013;Dijkstraetal.2014). Montes-Fontalvo 2011). Moreover, some of the descrip- In a research on the phylogenetical reassessment of the tions conducted in the first half of the twentieth century Coenagrionidae subfamilies, O’Grady and May (2003) in- lack detailed illustrations of the diagnostic traits, which cluded the species Argia sedula, Argia moesta,andArgia makes them difficult to interpret. bipunctulata in their analysis. As a result, in one of the trees proposed, Argia was recovered as a polyphyletic group. In another phylogenetic tree, Argia was a monophyletic group, Phylogenetic studies including larvae with A. bipunctulata as the basal clade, whereas A. sedula and A. moesta were sister species. Within the order Odonata, the phylogenetic hypotheses have Carle et al. (2008), in a phylogenetic study focused on the been proposed based primarily on morphological traits of superfamily , included the species A. moesta adults (Westman et al. 2000;May2002; Carle and Kjer and Argia vivida. These were not recovered as sister species. 2002; von Ellenrieder 2002, 2013;Rehn2003;O’Grady and A. moesta appeared as a sister taxon of Neoneura May 2003; Pessacq 2008; Yu and Bu 2011;Blankeetal. (), newly integrated to the Coenagrionidae as 2013) or exclusively on molecular information (Chippindale the subfamily Protoneurinae (Dijkstra et al. 2014). et al. 1999; Misof et al. 2000;Artissetal.2001;Guanetal. In a study on the redefinition of the families in the suborder 2013;Dijkstraetal.2014; Kim et al. 2014), in a few cases on Zygoptera, Dijkstra et al. (2014) included two species of Argia the analysis of a combination of data sets (morphology of (i.e., Argia oculata and Argia sp.), which formed a monophy- adults + larvae + fossils + DNA sequences) (Baskinger et al. letic group and were the sister species of the genera Neoneura, 2008; Bybee et al. 2008; Caesar and Wenzel 2009). However, Epipleoneura,andDrepanoneura. However, the authors studies that explore new morphological traits of larvae (e.g., found that the support values for some coenagrionid groups length and proportion between structures) are scarce. were low and the phylogenetic relationships among them Furthermore, the assessment of those characters in a phyloge- changed if Argia was excluded. As a result, the authors sug- netic contest to diagnose monophyletic clades is a novel gest conducting further phylogenetic studies on this genus and attempt. At the family level, the only works that can be other related genera. mentioned have been conducted with Aeschnidiidae The only preliminary Bayesian inference analysis focused (Odonata: Anisoptera; Fleck et al. 2002), on molecular traits and some morphological features for the (Odonata: Anisoptera; Fleck 2011), and Epiophlebiidae species of Argia living in Northern Mexico was conducted by (Odonata: Anisozygoptera; Blanke et al. 2015). Similarly, Caesar and Wenzel (2009). These authors included in their studies based on the combination of larval morphology and analyses nine morphological traits of adults of both sexes, molecular information include those for Leucorrhinia one trait of larvae, and sequences of the mitochondrial 16S (Odonata: ) (Hovmoller and Johansson 2002)and ribosomal RNA (rRNA) gene. These authors reported two Micromacromia (Odonata: Libellulidae) (Fleck et al. 2008). trees based on molecular information alone, which include Phylogenetic hypotheses based solely on traditional three polytomies. The combination of morphological and taxonomic traits of larvae have not been proposed within the molecular data improved the resolution in the topology of suborder Zygoptera. However, Carle and Louton (1994), the third proposed tree, which was the authors’ main phylo- Novelo-Gutiérrez (1995), Ware et al. (2007 ), Caesar and genetic hypothesis. In this phylogeny, the Argia species group Wenzel (2009), Ballare and Ware (2011), and Dijkstra et al. was recovered as monophyletic (Caesar and Wenzel 2009). (2014) stressed the importance of morphological exploration However, the clade formed by Argia nahuana, Argia alberta, of new traits and the use of the traditional taxonomic traits of Argia leonorae, Argia agrioides, Argia hinei, Argia pallens, larvae, as a potential solid source of information to understand Argia fumipennis, A. sedula, Argia pulla,andArgia rhoadsi the evolutionary relationships between the Odonata species displayed a basal polytomy, suggesting that these data are from a phylogenetic framework. insufficient to understand how these species are interrelated. Phylogeny of Argia (Odonata) 411

These same authors contend that their findings illustrate the the following entomological collections: IEXA-INECOL importance of morphological data to supplement the molecu- (Xalapa, Mexico), IORI (Gainesville, USA), and Dr. Natalia lar information for the genus Argia, thus warranting further von Ellenrieder (California, USA). investigation given their likely information potential. They mention that the Argia larvae remain poorly studied and that Codification of morphological traits larval traits are a potential source of phylogenetic information. The above points to the need to investigate the morphology of A total of 97 traits (Online Resource 2) were obtained, 57 of larvae while exploring morphological traits, aiming to which were qualitative and 40 quantitative. For the quantita- improve the resolution and support to clades where the tive traits, states were delimited based on Thiele’s(1993)and evolutionary interrelations of the species within the clade Wiens’ (2001) formulas. The missing characters that could not remain unknown (Caesar and Wenzel 2009). be encoded for a lack of the structure in the specimen exam- Accurate identification of is, without a doubt, a ined were entered with the symbol B?,^ while non-applicable major problem, especially for holometabola . This is a characters were encoded as B-^. consequence of the lack of studies that recognize morpholog- ical characters useful to set species boundaries in ontogenetic DNA extraction, amplification, and sequencing stages other than the adults. This is particularly crucial for taxa like Odonata in which juvenal instars are often used as indi- We sequence the 16S rRNA gene in the BSistematica cators of environmental quality. Thus, the first objective of our Filogenetica^ laboratory at the Instituto de Ecología (Xalapa, research is to shed light on the identification of Argia larvae by Mexico), while the cytochrome C oxidase subunit I (COI) means of robust morphological characters. As a second objec- gene sequences were obtained by the BOLD Systems interna- tive, we will explore the phylogenetic relationships of a group tional project laboratories. Muscle tissue was extracted from of 36 Argia species based on those larval morphological traits the legs and abdomen of adult specimens preserved in acetone and sequences of the 16S rRNA and cytochrome oxidase and larval specimens preserved in 96% ethanol. DNA extrac- subunit I mitochondrial genes. tion was performed using the kit ZR Tissue & DNA MiniPrep (Zymo Research, California, USA) following the manufacturer’s instructions. The genomic DNA was stored Materials and methods in refrigerators at −20 °C. A 548-bp fragment of the 16S rRNA gene was amplified using primers LR-J-12887 (5′- Examination of specimens CCGGTCTGAACTCAGATCACGT-3′) and LR-N-13398 (5′-CGCCTGTTTAACAAAAACAT-3′).Foreachofthe The larvae of 46 species were examined: 35 species of the PCR reactions, a total volume of 30.7 μl was used, composed genus Argia and 11 species as an outgroup, with a total of of 16 μl double-distilled water (ddH2O), 3 μl buffer, 3 μl 320 specimens examined. The following species were used dNTPs, 1.5 μlMgCl2,3μl of each primer, 0.2 μlTAQ,and as an outgroup: Hetaerina americana and Hetaerina 1 μl DNA. PCR conditions were as follows: 94 °C for 3 min, vulnerata (), alacer and Lestes alfonsoi followed by 30–40 cycles at 94 °C as the denaturing temper- (), Amphipteryx agrioides (Amphipterygidae), and ature for 1 min, 30–50 °C as the annealing temperature for civile, Enallagma novaehispaniae, Ischnura 1 min, 74 °C as the extension temperature for 2 min, and a denticollis, Ischnura ramburii, digiticollis,and final extension at 72 °C for 5 min. The amplified PCR Telebasis salva (Coenagrionidae). These species were includ- products were visualized in 1% agarose gels stained with ed following the criteria of Rehn (2003), Bybee et al. (2008), ethidium bromide and purified with GeneJET PCR Caesar and Wenzel (2009), and Dijkstra et al. (2014). The Purification (Thermo Scientific, Lithuania, EU). The direct species A. nahuana was not included within the morphologi- sequencing was performed using the BigDye Terminator cal analysis because we did not have access to larvae of this v3.1 Cycle Sequencing Kit. Sequences were read in the ABI species. Refer to the Online Resource 1 for a comprehensive PRISM® 310 Genetic Analyzer (Applied Biosystems) at list of all the taxa examined. Instituto de Ecología, A.C. The specimens selected belonged to the final instar (F0), To obtain the sequencing of the COI gene, two samples of which was characterized by evident wing venation in the leg tissue for each of the Argia species and the outgroup pterothecae. When this was not available, F1 or F2 specimens species were sent to the BOLD Systems international project were used, as their morphological differences are not signifi- laboratories. The protocols of this project consisted in using a cant with respect to instar F0. Males and females of all taxa plate on which 5 ml of the insect lysis buffer and 0.5 ml of were examined. Structures’ length was recorded in proteinase K were poured; afterwards, the tissue sample from millimeters, and the terminology of Corbet (1953) was used each specimen was immersed in this mixture and incubated at for the labium. The specimens studied were kindly loaned by 56 °C for at least 6 h. The following DNA extraction steps 412 Torres-Pachón M. et al. were performed according to the protocol of the Canadian phylogenetic signal. In all instances, the robustness of clades Center for DNA Barcoding (CCDB) DNA Extraction was estimated by 1000 pseudoreplicates in a bootstrap test. (Ivanova et al. 2006). A 680-bp fragment was amplified using Character states were mapped on the most parsimonious tree the primers LC-OR-1490 (5′-GGTCAACAAATCAT using the slow optimization command in Winclada (Nixon AAAGATATTGG-3′) and HC-OR-2198 (5′-TAAA 2002) to infer character transformation and to recover group CTTCAGGGTGACCAAAAAATCA-3′) (Folmer et al. synapomorphies. 1994; Ivanova and Grainger 2006). For each reaction, a total Our molecular data set was supplemented with sequences volume of 12.5 μl was used, made of 6.25 μltrehalose,2μl retrieved from GenBank. The final composition of the molec- ddH2O, 1.25 μlof10×buffer,0.625μlMgCl2,0.125μlof ular and total evidence matrices is shown in Online Resource each primer, 0.0625 μldNTPs,0.06μlpolymerase,and2μl 4. These two matrices were analyzed through Bayesian infer- of template DNA (Ivanova et al. 2006). The typical ence (BI) using a maximum likelihood criterion for selecting amplification conditions for the COI gene include an initial the optimal trees. To reconstruct the BI trees, we used a gen- denaturation period at 94 °C for 1 min, 5 cycles at 94 °C for eral model for character evolution for the morphology data. 30 s, annealing at 45–50 °C for 40 s, and an extension at 72 °C Character state transformation was set to variable following a for 1 min, followed by 30–35 cycles at 94 °C for 30 s, 51– gamma probability. The evolutionary model of each molecu- 54 °C for 40 s, and 72 °C for 1 min, with a final extension at lar partition as chosen by jModelTest v 2.0.2 (Guindon and 72 °C for 10 min. Sequencing was carried out following the Gascuel 2003; Darriba et al. 2012) was used as prior (i.e., protocol for the BigDye Terminator v3.1 Cycle Sequencing TPM1u + gamma + I, gamma shape = 0.485, invariant kit, and the protocols for the Canadian Center for DNA sites = 0.396 for 16S rRNA; TrN + gamma + I, gamma Barcoding (Ivanova and Grainger 2006). shape = 0.301, invariant sites = 0.461 for COI). To allow for In all cases, the PCR products were sequenced in differences in the evolutionary characteristics of each parti- both directions, and the differences or ambiguities be- tion, substitution models were set as unlinked. Tree space tween the strings were resolved visually after examining was explored using MrBayes v. 3.2.2 (Ronquist et al. 2012). the electropherograms of each sequence using the soft- For these data sets, two parallel Markov chain Monte Carlo ware Sequencher v5.0.1 (Gene Codes Corporation, Ann (MCMC) analyses were run simultaneously, each for a mini- Arbor, Michigan, USA). All nucleotide sequences were mum of 20 million generations, sampling every 1000 genera- blasted in GenBank in order to corroborate gene identity tions. A 25% burn-in value was set (Ronquist et al. 2012)to and to compare them with homologous sequences of only include trees after the -lnL values reached an asymptote. other odonates. Finally, a majority rule consensus tree was calculated, show- ing nodes with a posterior probability (pp) higher than 0.5. All Phylogenetic analysis trees were drawn using FigTree v 1.4.2 (Rambaud 2009).

To assess the phylogenetic signal contained in each data set, three alternative character matrices were used for the cladistic Results analysis. The first was based on larval morphological traits (Online Resource 3), the second was based on molecular The phylogenetic hypotheses recovered in the various analy- characters (i.e., 16S rRNA + COI sequences), whereas the ses were consistent with one another. In general, the lack of third was a total evidence matrix (i.e., larval morphology + resolution within the internal group was responsible for most 16S rRNA + COI). differences observed between phylogenies. The maximum For the morphology matrix, the main optimality criteria parsimony analysis recovered 19 trees with a length of 579 used during the tree selection was maximum parsimony. The steps (consistency index = 0.237; retention index = 0.562). reconstruction was performed using the program TNT v1.5 The consensus tree showed a polytomy at the base of the (Goloboff et al. 2008). To start, ten million random trees were genus Argia, resulting in an almost total lack of resolution that created and loaded in RAM. The tree space was searched prevented us from inferring the precise evolutionary relation- during the phylogenetic analyses by calling the implicit enu- ships between its species. A puzzling finding is that Argia was meration (ienum) command. The trees retained were subjected not recovered as a monophyletic group. Within the basal to branch swapping with the tree bisection and reconnection polytomy, a clade formed by species of the genera (TBR) algorithm. To recover the best tree, this analysis was Enallagma, Ischnura, Lestes, and Telebasis was recovered. run separately ten times. All characters were considered unor- Additionally, the bootstrap values for the few nodes recovered dered and non-additive. The consistency index, retention were low (i.e., <50%). In order to reduce the phylogenetic index, and tree length were calculated as tree descriptors. uncertainty resulting from the lack of resolution in the consen- When multiple equally parsimonious trees were obtained, a sus tree, the morphological matrix was reanalyzed by applying majority rule consensus was computed to summarize the implied weights (Goloboff 1993). The analysis was repeated Phylogeny of Argia (Odonata) 413 by varying the concavity value (i.e., k = 2 to 6) in order to clade keeps virtually the same phylogenetic relationships obtain the trees that minimized homoplasia and maximized recovered with the morphological data; however, the the data fit. The best fit was achieved with k =3 total evidence suggests that A. nahuana should be part (fit = 47.001; rearrangements = 3,297,373), recovering only of this group, being the basal species. two trees. The strict consensus tree (Fig. 1)showsthegenus To recover the evolutionary history of the characters and Argia as a monophyletic group; however, the support for this their states, these were mapped in the two phylogenetic trees node was low (bootstrap <50%). Even when the internal struc- recovered in the parsimony analysis corrected by implied ture of the Argia clade is almost completely resolved, all its weights. Figure 4 shows the evolutionary scenario inferred nodes received poor bootstrap values (<80%). The most basal from larval morphological characters based on one of these species within the genus were A. alberta and Argia tibialis. trees. Given the phylogenetic uncertainty, we only report the The clades that were consistent with the results recovered in synapomorphies recovered for the stable clade groups. The the molecular analyses are mentioned below. At the upper end optimization of these synapomorphies is identical under both of the phylogeny, we recovered the Bcarlcooki^ clade (these phylogenetic scenarios. Of the 97 characters encompassed the names are assigned only as reference, but they lack a strict matrix of larval morphological traits, 48 had not been previ- taxonomic identity), which showed the following evolution- ously used to explore the phylogenetic relationships and/or set ary relationships (Argia carlcooki (Argia plana ((A. vivida, taxonomic proposals in odonates (Online Resource 2). The Argia extranea)(Argia immunda, A. bipunctulata)))). The genus Argia can be recognized by the combination of eight Brhoadsi^ clade clustered the species (A. rhoadsi (A. pulla, character states: (i) robust and long body shape [character 1: A. sedula)). The Bpercellulata^ clade shows a polytomy at its state 1], (ii) absence of premental setae [31: 1], (iii) presence base, grouping six species with the following historical struc- of basidorsal setae in the prementum [35: 1], (iv) two ramifi- ture (Argia percellulata (Argia cuprea, Argia tezpi)(Argia cations in the palp [42: 2], (v) long and sharp apical palp joergenseni (Argia barretti, Argia harknessi))). The ramifications [43: 4], (vi) length of abdominal segments 9 Btarascana^ clade is formed by (Argia tarascana (Argia and 10 that is 1.08 to 1.77 times larger than the length of male munda, A. pallens)). Finally, the relationship of sister species gonapophysis [73: 0], (vii) epiproct with no lateral carina [91: between Argia lugens and Argia funcki was recovered regard- 1], and (viii) tracheation of the epiproct not visible [94: 1]. The less of the weighting scheme applied, the taxa composition relationship between A. lugens and A. funcki is supported by used, or the optimality criterion selected. the combination of six character states: (i) the second The phylogenetic hypothesis exhibiting the lowest resolu- flagellomere 0.76 times shorter to equal to the length of the tion was the one recovered based on the mitochondrial DNA first flagellomere [18: 1], (ii) second flagellomere longer than sequences only. The monophyly of the genus Argia received a the pedicel [22: 0], (iii) 18 to 22 basidorsal setae in the support value of pp =0.87(Fig.2). Virtually, all the internal prementum [36: 2], (iv) length of the external edge of the palp group is collapsed in a large basal polytomy. Internally, the greater than the length of the movable hook [45: 1], (vi) length only clades that maintained their phylogenetic identity were of medium leg that is 2.54 to 2.83 times larger than the length rhoadsi and (A. lugens, A. funcki) with pp = 0.85 and 0.97, of the medium tibia [54: 0], and (vi) epiproct length that is respectively. 1.54 to 2.12 times greater than the maximum epiproct width The BI analysis based on the total evidence (Fig. 3)sup- (excluding the tip) [95: 0]. The tarascana clade is supported by ports the monophyly of the genus Argia (pp = 1), as well as the the combination of five synapomorphies: (i) antennae are 9.78 monophyly of the species groups structured based on morpho- to 12.94 times greater than the scape length [8: 2], (ii) pedicel logical traits. The total-evidence phylogeny shows a higher is 1.80 to 2.33 times greater than the scape length [9: 2], (iii) resolution level than the mitochondrial DNA phylogeny; how- second flagellomere is 2.14 to 2.82 times greater than the ever, the degree of phylogenetic uncertainty remains high, scape length [11: 2], (iv) fourth flagellomere is 0.45 to with a large number of unresolved nodes. The basal position 0.73 times shorter than the pedicel length [16: 1], and of A. alberta and A. tibialis was corroborated. Both species (v) presence of an outer lateral carina in the paraprocts form a monophyletic group in these phylogenetic hypotheses; [82: 0]. Three characters differentiate the species that however, the support for this node is practically negligible comprise the percellulata clade: (i) length of cephalic lobe (pp = 0.52). The next clade is formed by the sister species smaller than the eye length [6: 1], (ii) spatulate shape of A. lugens and A. funcki; this relationship is well supported the male gonapophysis [71: 1], and (iii) the lateral (pp = 1). The percellulata clade is supported only by a gonapophysis of the female is bluntly pointed [74: 1]. pp = 0.62. Unlike the morphological phylogeny, here, the The rhoadsi clade can be distinguished by the following species Argia oenea was recovered as part of this clade, characters: (i) pedicel is 1.80 to 2.33 times greater than while the species A. tezpi was left out. The rhoadsi the scape length [9: 2] and (ii) long and acuminate clade is corroborated with a pp = 0.91, while the epiproct tip [89: 1]. Finally, the carlcooki clade is charac- tarascana clade yields a pp = 0.89. Finally, the carlcooki terized by the following combination of characters: (i) 414 Torres-Pachón M. et al.

Amphipteryx agrioide Fig. 1 Strict consensus tree of Hetaerina vulnerata two equally parsimonious trees Hetaerina americana 100 Lestes alacer inferred from the morphological Lestes alfonsoi 100 Enallagma civile traits of larvae. During the search, Enallagma novaehispa Ischnura denticollis implied weights were applied 55 Ischnura ramburii Telebasis digiticoll based on k = 3 (fit = 47.001; Telebasis salva rearrangements = 3,297,373). Argia alberta

Argia tibialis “funcki” Numbers are the bootstrap Argia chelata Argia lugens support; only scores greater than Argia funcki Argia tonto 50% are shown. Names between Argia oculata “tarascana” quotes are for reference only and Argia lacrimans Argia tarascana do not represent formal 51 Argia munda Argia pallens taxonomic groups Argia moesta Argia translata Argia oenea “percellulata” Argia percellulata Genus Argia cuprea Argia tezpi Argia joergenseni

Argia barretti Argia Argia harknessi Argia ulmeca

Argia apicalis “rhoadsi” Argia anceps Argia rhoadsi 60 Argia pulla Argia sedula Argia hinei 56 Argia emma

Argia fumipennis “carlcooki” Argia carlcooki Argia plana Argia vivida Argia extranea 79 Argia immunda Argia bipunctulata head width is 2.22 to 2.76 times greater than the antennae Discussion length [5: 2], (ii) second flagellomere shorter than the pedicel length [22: 1], (iii) third flagellomere shorter than The genus Argia appeared as a monophyletic group in the the scape length [23: 1], (iv) triquetral paraprocts present three phylogenetic hypotheses proposed here, which is [76: 1], and (v) triquetral epiprocts present [87: 4]. consistent with the phylogenetic reassessment work for the

Fig. 2 Bayesian inference tree Amphipteryx agrioides 1 Hetaerina sanguinea recovered from the phylogenetic 1 Hetaerina vulnerata Hetaerina americana signal contained in the 1 Lestes praemorsus 0.99 mitochondrial genes (i.e., 1 Lestes helix Lestes alacer 16S rRNA and COI). 0.88 1 Telebasis digiticollis Numbers indicate posterior Telebasis salva 0.97 Telebasis obsoleta probability values 0.81 0.93 Telebasis dunklei 1 Ischnura senegalensis 0.8 Ischnura elegans 0.74 Ischnura denticollis Ischnura barberi 0.92 Enallagma civile 0.96 Enallagma circulatum 0.55 Enallagma sp. Argia tezpi Argia moesta Argia tarascana Argia inmunda 0.87 Argia chelata Argia nahuana Argia carlcooki Argia fumipennis Argia cuprea Argia munda

Argia translata Genus Argia 0.51 Argia hinei Argia anceps Argia tonto 0.89 Argia percellulata “percellulata I” Argia joergenseni 0.93 Argia barretti “percellulata II” Argia harknessi 0.96 Argia extranea Argia plana 0.97 Argia lugens “funcki” Argia funcki 0.85 Argia sedula 0.96 Argia pulla “rhoadsi” Argia rhoadsi Argia ulmeca Argia oculata 0.2 0.56 Argia oenea Phylogeny of Argia (Odonata) 415

Amphipteryx agrioides Hetaerina sanguinea 1 1 Hetaerina vulnerata Hetaerina americana 1 Lestes praemorsus 1 Lestes helix 0.99 Lestes alacer 0.97 Lestes alfonsoi 1 Telebasis digiticollis 0.98 Telebasis salva 0.96 Telebasis obsoleta 0.99 Telebasis dunklei Ischnura senegalensis 0.94 Ischnura elegans Enallagma novaehispaniae Ischnura barberi 0.54 Ischnura denticollis 1 Ischnura ramburii 0.57 Enallagma civile 0.84 Enallagma circulatum 0.77 Enallagma sp. 0.52 Argia alberta Argia tibialis 1 Argia lugens “funcki” 1 Argia funcki Argia tonto 0.52 Argia oculata 0.64 Argia lacrimans Argia moesta 0.54 Argia ulmeca Argia tezpi 0.54 Argia translata Argia oenea 0.57 Argia cuprea 0.62 Argia percellulata “percellulata” Genus 1 Argia joergenseni 0.2 Argia barretti 0.94 Argia harknessi Argia hinei

Argia anceps Argia Argia apicalis Argia fumipennis Argia emma 0.58 0.56 Argia sedula 0.91 Argia pulla “rhoadsi” 0.66 Argia rhoadsi Argia chelata 0.89 Argia tarascana 0.54 Argia pallens “tarascana” 0.82 Argia munda Argia nahuana 0.78 Argia carlcooki Argia plana 0.87 Argia extranea Argia vivida “carlcooki” 0.89 Argia immunda 1 Argia ºbipunctulata

Fig. 3 Phylogenetic hypothesis recovered by Bayesian inference from the total evidence data set

Argia species of North American distribution performed by dramatically between larvae and adults. Character states Caesar and Wenzel (2009). Even when the specific evolution- homologous to those observed in Argia have been described ary interrelationships of the species varied according to the in the traditional taxonomic works on larvae of species in the phylogenetic analysis matrix used, at least five species groups families Amphipterygidae (Novelo-Gutiérrez 1995), were consistent. The phylogenetic uncertainty resulting from Calopterygidae (Zloty et al. 1993), Lestidae (Novelo- the lack of resolution in the consensus trees is due to the high Gutiérrez and González-Soriano 2003), and Coenagrionidae level of homoplasy in the partition of both morphological and (Novelo-Gutiérrez 2005). Both the new characters explored as molecular data sets. In spite of this, we believe that the present those historically used in taxonomic and phylogenetic studies analysis provides evidence of the use of non-molecular traits reflected a high degree of homoplasy within the tree (Fig. 4). to address taxonomic and phylogenetic issues in hemimetab- Therefore, our internal group fails to display unique characters olous insects such as the odonates. to allow the identification of individual clades. The only ex- ceptions to the above were found in the carlcooki clade and for The importance of the morphological traits of larvae the sister species A. funcki and A. lugens. Triquetral epiproct occurs only in the species belonging to the carlcooki group; on The analyses conducted agree that Argia is a natural evolu- the other hand, the presence of 18 to 22 basidorsal setae in the tionary group. The scenario on the evolution of morphological prementum of the larvae indicates that these should belong to characters of larvae recovers a combination of eight characters the species A. funcki or A. lugens.Atthelevelof for the node of this genus. Of these, the robust and long body autapomorphies in terminal groups, only three larvae species shape, the presence of basidorsal setae in the prementum, may be differentiated based on diagnostic traits. The speci- labial palp with two ramifications, and the long and sharp mens of A. alberta were the only one that showed a cephalic apical palp ramifications are character states unique to the lobe length equal to the eye length. A. barretti specimens were Argia species included in this study. Therefore, the recogni- the only one that showed the length of the outer edge of the tion of any of these characteristics in odonate larvae would palp equal to the length of the movable hook, and the maxi- indicate that they likely belong to the genus Argia.Thistype mum width of the paraproct is equal to the maximum width of of evidence is of the most importance to determine the taxo- the epiproct. Moreover, larvae with middle femur length nomic identity of specimens where ontogenetic stages change shorter than the hind femur length possibly belong to 416 Torres-Pachón M. et al.

Outgroup 629 30 48 81 95 Argia alberta 2 1 0 2 2 0 45 79 Argia tibialis Calvert

“funcki” Argia chelata 135 4243 3173 9194 1 1 280 89 91 Argia chelata 6 34 48 49 1 1 2 4 1 0 1 1 0 0 1 0 210 1113 14 15 17 74 Argia funcki 47 50 52 57 78 1 1 0 0 18 22 36 45 54 95 2 3 2 1 2 2 1 1 1 0 2 1 0 0 1 0 0 0 1 19 53 69 75 78 Argia lugens 1 0 1 1 0 53 62 89 95 Argia tonto 75 82 0 0 1 0 53 55 90 Argia oculata 0 1 0 0 1 2 6 23 59 73 78 Argia lacrimans “tarascana” 2 1 1 1 1 67 68 69 0 4 95 13 48 78 Argia tarascana 8 9 11 16 82 1 1 1 1 0 0 0 1 10 12 17 45 58 62 Argia munda 2 2 2 1 0 15 19 24 54 81 84 3 2 1 1 2 0 2 1 0 1 2 0 2 4 20 30 36 37 5356 57 61 67 80 Argia pallens 0 0 1 0 0 0 0 1 1 1 0 0 81036 78 85 94 Argia translata 72 95 2 3 0 0 0 0 13 22 54 1 0 2 4 18 23 48 49 50 5374 89 Argia oenea 1 0 0 0 1 1 1 0 0 1 0 2 1 34 58 15 4752 59 60 67 6869 89 1 2 Argia moesta a 2 0 1 1 1 0 0 0 1 618 19 34 54 Argia cuprea 21415 45 48 49 89 0 1 1 0 1 “percellulata” 0 1 2 0 0 0 1 36 58 85 94 Argia tezpi 53 72 95 0 1 0 0 0 1 0 6 14 45 52 80 90 Argia barretti 671 74 66 67 0 1 2 1 3 2 63 64 65 63 64 65 66 1 1 1 0 0 1518 30 50 53 5457 61 95 Argia harknessi 1 1 1 1 0 0 0 34 74 2 1 2 1 1 1 1 1 1 b 0 2 Argia joergenseni 50 Argia percellulata 1 23 45 48 56 60 83 5810 1118 48 59 81 Argia hinei Argia ulmeca2 20 49 62 95 1 1 2 1 1 1 02 3 2 1 0 1 0 36 37 50 55 85 86 94 1 1 1 0 0 Argia emma 18 19 37 51 73 77 88 1 1 0 0 0 0 0 6 58 81 92 1 1 0 1 0 1 1 Argia fumipennis 8 11 30 53 1 2 2 1 “rhoadsi” Argia rhoadsi 9 89 2 2 0 0 19 2122 51 58 86 94 c 84 91 48 50 76 78 80 87 Argia pulla 436 49 2 1 0 1 1 1 2 0 0 0 0 0 0 3 1 0 3 1 0 0 59 73 Argia sedula 1 0 10 17 20 2130 45 55 58 Argia carlcooki 2 50 73 1 1 1 1 0 1 0 0 36 53 56 57 73 “carlcooki” 0 1 1 522 2376 87 Argia plana 5 9 23 30 50 13 52 82 1 0 1 1 0 Argia bipunctulata 2 1 1 1 4 59 6066 67 68 69 90 1 2 0 0 0 d 0 1 0 83 86 1 1 0 0 0 0 1 10 18 36 51 52 Ar gia immunda 1 0 63 64 65 1 1 1 1 2 6 49 51 53 58 0 0 0 10 19 20 21 54 62 74 89 Argia anceps 2 80 Argia extranea 1 1 1 0 2 1 1 1 1 0 0 1 1 1 3 13 29 30 37 41 46 51 52 5758 75 77 22 23 45 55 6179 95 Argia apicalis Argia vivida 1 1 1 0 1 1 0 1 1 0 2 1 1 1 0 1 2 1 1 Fig. 4 Evolutionary scenario for the larval morphology inferred by direct the specific character, and the number below indicates the character state. optimization of characters on one of the two most parsimonious trees Images illustrate the four synapomorphies that are unique to the genus recovered in the phylogenetic analysis after applying implied weights. Argia: short and robust body shape [1:1] (a), basidorsal setae present Black circles indicate transformation of unique characters; white circles [35:1] (b), two palp ramifications [42:2] (c), and apical palp indicate parallelism or reversion. The number above the circle indicates ramifications as long as the sharp hooks [43:4] (d)

A. carlcooki. In any case, all Argia species used revealed clear the other hand, there is still great phylogenetic uncertainty and unique combinations of larval morphological that, elucidating with certainty, restrains the detailed evolu- autapomorphies that enable an easy identification to species tionary history of the species in this genus. However, it is level (Fig. 4). As in the present study, the work of von possible to recover general patterns of species associations, Ellenrieder (2002) and Rehn (2003) on other Odonata groups which are consistent in both morphological and molecular demonstrates the importance of using morphological charac- characters. The total-evidence BI shows a first cladogenesis ters of larvae. Some of the clades included in their phyloge- event where an ancestor separates and gives rise to a lineage netic hypothesis were resolved and supported by unique syn- leading to A. alberta and A. tibialis. This evolutionary rela- apomorphies, which corresponded to traditional taxonomic tionship is biogeographically consistent, since these two spe- traits of larvae of (von Ellenrieder 2002)and cies have a restricted distribution from Northern Mexico to Zygoptera (Rehn 2003). Canada. In spite of this, there are alternative hypotheses, which do not agree with our findings. According to Caesar Phylogenetic hypotheses and Wenzel (2009), A. tibialis should be more closely related to Argia apicalis, A. tarascana,andA. tezpi, rather than with The results obtained from the various analyses undertaken A. alberta; this latter species, in turn, shares an immediate provide evidence of the monophyly of the genus Argia.On common ancestor with A. nahuana and A. leonorae. Phylogeny of Argia (Odonata) 417

The monophyletic group formed by A. funcki and slightly prominent ligula. A. pulla and A. rhoadsi share A. lugens is the most stable relationship, since it is re- a small body size, a short tapered male gonapophysis covered in all analyses irrespective of the criterion used with sharp and divergent tips and yellow gills with (Figs. 1, 2, 3,and4). Similarly, these two species were irregular brown spots. In their analysis, Caesar and compared morphologically in the larval stage and pro- Wenzel (2009) found that these three species share a posed as close by Novelo-Gutiérrez and Gómez-Anaya common immediate ancestor, as observed in our various (2006), who confirmed their assignment to the subgenus phylogenies (i.e., morphology, molecular, and total Hyponeura Selys proposed by Gloyd (1968). Likewise, evidence). Garrison (1994) concludes that adults of A. lugens and Even when the tarascana clade [A. tarascana, A. pallens, A. funcki are very similar. In this case, our results con- and A. munda] is recovered in both the morphological (Fig. 1) firm the hypothesis of Caesar and Wenzel (2009), where and the total evidence (Fig. 1) analyses, the identity of this A. funcki and A. lugens formed a node of evolutionary group is maintained primarily based on larval characters. The sister species. molecular characters alone were not able to recover this group The association of species found for the percellulata in a statistically significant manner. For Novelo-Gutiérrez group [i.e., (A. cuprea, A. oenea,(A. barretti, (1992), A. tarascana is related to Argia anceps by having A. harknessi), (A. percellulata, A. joergenseni)); Fig. 3] three long palp setae, laminar gills with a similar arrangement has been considered previously. Novelo-Gutiérrez (1992) of spiny setae on the lateral surface of the carina, and tapered related A. harknessi to A. oenea because they share gonapophysis in both males and females. According to this similarities in the shape of the cerci in male larvae. author, A. munda is more closely related to Argia tonto by Garrison (1994), based on the morphology of the adults, displaying evenly colored antennae, two palpal setae, and a considered that A. cuprea shows considerable similari- male gonapophysis of tapered shape. Subsequently, Garrison ties with A. oenea and A. barretti. However, the same (1994) considered that the morphological similarity between author found similarities between these three species adults suggests that A. pallens should be closer to and other taxa such as Argia translata and A. tezpi; A. agrioides, A. fumipennis, A. hinei , A. leonorae,and these were not consistently grouped within our A. nahuana,whileA. munda should be related to A. vivida, percellulata clade but, nonetheless, were found as A. extranea,andA. plana. Despite the lack of historical sup- evolutionarily close to this clade (Figs. 1 and 4). Von port and the phylogenetic uncertainty contained in the DNA Ellenrieder (2007) described the larva of A. joergenseni sequences, the consistency observed in the larval morpholog- and included it within the Argia species group of larvae, ical data sets points to the need to conduct further in-depth characterized by a prominent ligula and one seta in analyses to test the evolutionary identity of this lineage. the palp, which associates it closely with A. translata. Last, at least some of the relationships found in the Novelo-Gutiérrez (2008) related the larvae of A. barretti carlcooki clade [(A. nahuana, A. carlcooki,(A. plana, tothelarvaeofA. harknessi and A. joergenseni,asthe (A. vivida, A. extranea,(A. immunda + A. bipunctulata)))); three species share a strongly prominent ligula, labial Fig. 3] are supported by historical precedents. According to palp with one seta, femur with dark strips that are wider Novelo-Gutiérrez (1992), the A. plana larva is close to than the intermediate pale spaces, spatulate shape of the A. vivida by having a small and pubescent body, a nearly male gonapophysis, and pale antennal escape with the straight rear margin of the head, antennae that are shorter rest of flagellomere of brown color. Caesar and Wenzel than the head, square prementum, a conical male (2009) conclude that there is a close evolutionary rela- gonapophysis, and triquetral gills. Garrison (1994) related tionship between A. cuprea,andA. harknessi.More the adults of A. plana, A. vivida, A. extranea, A. immunda, recently, Novelo-Gutiérrez and Gómez-Anaya (2012) and A. nahuana. In the taxonomic key by Westfall and May mention that A. percellulata shows a great similarity to (1996), the species A. extranea, A. vivida, A. bipunctulata,and A. barretti, A. harknessi, Argia insipida,and A. immunda were grouped by sharing the following morpho- A. joergenseni by sharing larval characters that include logical characters in the larval stage: a marginal fringe of stout abundant claviform setae on the middle of sternite 8, setae running at least three fourths the length of the ventral and male cerci globose, and bluntly pointed spatulate dorsal margins of the lateral gills, usually three setae in the gonapophyses in males, beset with claviform setae on palp, femora of variable coloration but often distinctly dark, ventral border. and a very long lateral carina in each lateral gill, resulting in a Similarly, the proximity of the species included in the markedly chitinous appearance of gills in dorsal view. In the rhoadsi monophyletic group [i.e., (A. sedula (A. pulla, hypothesis proposed by Caesar and Wenzel (2009), A. plana A. rhoadsi)); Figs. 2 and 3] has been previously pro- shares a node as the evolutionary sister species of A. extranea; posed. Novelo-Gutiérrez (1992) recognized the relation- likewise, A. vivida and A. extranea are recovered as evolution- ship between these three species characterized by a arily close species. 418 Torres-Pachón M. et al.

Conclusion References

The strong phylogenetic uncertainty caused by the high Artiss, T., Schultz, R. T., Polhemus, D. A., & Simon, C. (2001). homoplasy observed in both the morphological and the Molecular phylogenetic analysis of the genera Libellula, molecular characters is an issue that is not unique to our Ladona,andPlathemis (Odonata: Libellulidae) based on mitochon- drial cytochrome oxidase I and 16S rRNA sequence data. Molecular study. Rehn (2003) recovered a polytomy between the genus Phylogenetics and Evolution, 18,348–361. Argia (Coenagrionidae), , and other Ballare, E. F., & Ware, J. L. (2011). Dragons fly, biologists classify: an cenagrionid genera. Likewise, Dijkstra et al. (2014)reported overview of molecular odonate studies, and our evolutionary under- low support values of posterior probabilities in the phyloge- standing of dragonfly and (Insecta: Odonata) behavior. International Journal of Odonatology, 14,137–147. netic relationships of the genus Argia with the other Baskinger, G. M., Ware, J. L., Kornell, D. D., May, M. L., & Kjer, K. M. coenagrionid genera included in their study. On the other (2008). A phylogeny of Celithemis inferred from mitochondrial and hand, the use of sequences of the COI and 16S rRNA genes, nuclear DNA sequence data and morphology (Anisoptera: – together with morphological information, does resolve the Libellulidae). Odonatologica, 37,101 109. Blanke, A., Büsse, S., & Machida, R. (2015). Coding characters from phylogenetic relationships for some odonate groups at various different life stages for phylogenetic reconstruction: a case study on taxonomic levels (Kambhampati and Charlton 1999;Misof dragonfly adults and larvae, including a description of the larval et al. 2000;Brownetal.2000;Artissetal.2001;Misof head anatomy of superstes (Odonata: et al. 2001; Stoks and McPeek 2006;Kjeretal.2006; Epiophlebiidae). Zoological Journal of the Linnean Society, 174, 718–732. Kiyoshi and Sota 2006;Wareetal.2007, 2009; Bybee et al. Blanke, A., Greve, C., Mokso, R., Beckman, F., & Misof, B. (2013). An 2008;Dammetal.2010; Lee et al. 2010; Lin et al. 2010). The updated phylogeny of Anisoptera including formal convergence above supports the idea that total-evidence analyses should be analysis of morphological characters. Systematic Entomology, 38, considered to be the primary basis for taxonomic and system- 474–490. Brown, J. M., McPeek, M. A., & May, M. L. (2000). A phylogenetic atic decision-making. perspective on habitat shifts and diversity in the North American This study is part of a continued effort to understand the Enallagma . Systematic Biology, 49,697–712. phylogenetic relationships within the genus Argia and other Bybee, S., Ogden, T., Branham, M., & Whiting, M. (2008). odonate groups that have been relegated historically by the Molecules, morphology and fossils: a comprehensive approach to odonate phylogeny and the evolution of the odonate wing. specialists. We recommend, for future studies, to explore Cladistics, 24,477–514. critically the different methods to encode continuous charac- Caesar, R. M., & Wenzel, J. W. (2009). A phylogenetic test of classical ters and increase the number of specimens of each species species groups in Argia (Odonata: Coenagrionidae). Entomologica examined (Garcia-Cruz and Sosa 2006). Additionally, we pro- Americana, 115,97–108. pose that future phylogenetic studies include a greater number Carle, F. L., & Louton, J. A. (1994). The larva of Neopetalia punctata and establishment of fam. nov. (Odonata). Proceedings of species of Argia in the adult stage, mainly those distributed of the Entomological Society of Washington, 96,147–155. in South America, and include supplementary molecular Carle, F. L., & Kjer, K. M. (2002). Phylogeny of Libellula Linnaeus markers such as nuclear gene sequences within the total evi- (Odonata: Insecta). Zootaxa, 87,1–18. dence data sets. Carle, F., Kjer, K., & May, M. (2008). Evolution of Odonata, with special reference to Coenagrionoidea (Zygoptera). Systematics &Phylogeny,66,37–44. Acknowledgments We are grateful to Enrique González-Soriano (IB- Chippindale, P., Davé, V., Whitmore, D., & Robinson, J. (1999). UNAM), Rosser W. Garrison, Natalia von Ellenrieder (California, USA), Phylogenetic relationships of North American damselflies of the and William Maufray (IORI-FSCA, Gainesville, USA) for the loan of genus Ischnura (Odonata: Zygoptera: Coenagrionidae) based on specimens. MTP thanks the Mexican Consejo Nacional de Ciencia y sequences of three mitochondrial genes. Molecular Phylogenetics Tecnología (CONACYT) for the M.Sc. Grant No. 414099/263565. and Evolution, 11,110–121. Thanks also to Janet Nolasco-Soto for her guidance in the laboratory Corbet, P. S. (1953). A terminology for the labium of larval Odonata. and to Hector Jaime Gasca Álvarez for the edition of photographs. Entomologiste, 86,191–196. Thanks to Torbjørn Ekrem (Norway) and the BOLD Project for their Damm, S., Dijkstra, K. D. B., & Hadrys, H. (2010). Red drifters and dark support in obtaining the COI gene sequences, to the staff at the residents: the phylogeny and ecology of a Plio-Pleistocene dragon- Servicios Escolares Office at Instituto de Ecología, A.C. (a.k.a. fly radiation reflects Africa’s changing environment (Odonata, INECOL) for their assistance in the inter-institution administrative Libellulidae, Trithemis). Molecular Phylogenetics and Evolution, procedures and support in the student mobility, and to the World 54,870–882. Dragonfly Association (WDA) for a grant to attend the 2013 ICO. This Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jModelTest manuscript has been improved by the comments and suggestions made 2: more models, new heuristics and parallel computing. Nature by Dr. Andreas Wanninger (Chief Editor, ODAE) and the two anony- Methods, 9,772. mous reviewers. Dijkstra, K. D. B., Kalkman, V., Dow, R., Stokvis, F. R., & van Tol, J. (2014). Redefining the damselfly families: a comprehensive molec- ular phylogeny of Zygoptera (Odonata). Systematic Entomology, 39, Compliance with ethical standards 68–96. Fleck, G. (2011). Phylogenetic affinities of Petaluridae and basal Conflict of interest The authors declare that they have no conflict Anisoptera families (Insecta: Odonata). Stuttgarter Beitraege zur of interest. Naturkunde Serie A, 4,83–104. Phylogeny of Argia (Odonata) 419

Fleck, G., Nel, A., Bechly, G., & Escuillié, F. (2002). The larvae of the Kiyoshi, T., & Sota, T. (2006). Differentiation of the dragonfly genus Mesozoic family Aeschnidiidae and their phylogenetic implications Davidius (Odonata: ) in Japan inferred from mitochon- (Insecta, Odonata, Anisoptera). Palaeontology, 45,165–184. drial and nuclear gene genealogies. Zoological Science, 23,1–8. Fleck, G., Brenk, M., & Misof, B. (2008). Larval and molecular Kjer, K. M., Carle, F. L., Litman, J., & Ware, J. (2006). A molecular characters help to solve phylogenetic puzzles in the highly phylogeny of Hexapoda. Arthropod Systematics & Phylogeny, 64, diverse dragonfly family Libellulidae (Insecta: Odonata: 35–44. Anisoptera): the Tetrathemistinae are a polyphyletic group. Lee, E. M., Hong, M. Y., Kim, M., Kim, M. J., Park, H. C., Kim, K. Y., Organisms, Diversity & Evolution, 8,1–16. Lee, I. H., Bae, C. H., Jin, B. R., & Kim, I. (2010). The complete Folmer, O., Black, M., Hoeh, W., Lutz, R., & Vrijenhoek, R. (1994). mitogenome sequences of the palaopteran insects Ephemera DNA primers for amplification of mitochondrial cytochrome oxi- orientalis (Ephemeroptera: Ephemeridae) and Davidius lunatus dase subunit I from diverse metazoan invertebrates. Molecular (Odonata: Gomphidae). Genome, 52,810–817. Marine Biology and Biotechnology, 3,294–299. Lin, C. P., Chen, M. Y., & Huang, J. P. (2010). The complete mitochon- Garcia-Cruz, J., & Sosa, V. (2006). Coding quantitative character data for drial genome and phylogenomics of a damselfly, Euphaea formosa phylogenetic analysis: a comparison of five methods. Systematic support a basal Odonata within the Pterygota. Gene, 468,20–29. Botany, 31,302–309. May, M. (2002). Phylogeny and taxonomy of the damselfly genus Garrison, R. W. (1994). A synopsis of the genus Argia of the United Enallagma and related taxa (Odonata: Zygoptera: States with keys and descriptions of new species, Argia sabino, Coenagrionidae). Systematic Entomology, 27,387–408. A. leonorae,andA. pima (Odonata: Coenagrionidae). Transactions Misof, B., Anderson, C. L., & Hadrys, H. (2000). A phylogeny of the of the American Entomological Society, 120,287–368. damselfly genus Calopteryx (Odonata) using mitochondrial 16S – Garrison, R. W., & von Ellenrieder, N. (2015). Damselflies of the genus rDNA markers. Molecular Phylogenetics and Evolution, 15,5 14. Argia of the Guiana Shield (Odonata: Coenagrionidae). Zootaxa, Misof, B., Rickert, A. M., Buckley, T. R., Fleck, G., & Sauer, K. P. 4042,1–134. (2001). Phylogenetic signal and its decay in mitochondrial SSU Garrison, R. W., von Ellenrieder, N., & Louton, J. A. (2010). Damselflies and LSU rRNA gene fragments of Anisoptera. Molecular Biology – genera of the New World: an illustrated and annotated key to the and Evolution, 18,27 37. Zygoptera. Baltimore: The Johns Hopkins University Press. Nixon, K. C. (2002). WinClada version 1.00.08 [software]. Ithaca, New Guindon, S., & Gascuel, O. (2003). A simple, fast and accurate method to York: Published by the author. estimate large phylogenies by maximum-likelihood. Systematic Novelo-Gutiérrez, R. (1992). Biosystematics of the larvae of the genus Biology, 52,696–704. Argia in Mexico (Zygoptera: Coenagrionidae). Odonatologica, 21, – Gloyd, L. K. (1968). The synonymy of Diargia and Hyponeura with the 39 71. genus Argia (Odonata: Coenagrionidae: Argiinae). Michigan Novelo-Gutiérrez, R. (1995). The larva of Amphipteryx and a reclassifi- Entomologist, 1,271–274. cation of Amphipterygidae sensu lato based upon the larvae (Zygoptera). Odonatologica, 24,73–87. Goloboff, P. (1993). Estimating character weights during tree search. Novelo-Gutiérrez, R. (2005). La larva de Enallagma novaehispaniae Cladistics, 9,83–91. Calvert, 1902 (Odonata: Zygoptera: Coenagrionidae). Folia Goloboff, P., Farris, J., & Nixon, K. (2008). TNT, a free program for Entomológica Mexicana, 44,219–224. phylogenetic analysis. Cladistics, 24,774–786. Novelo-Gutiérrez, R. (2008). Description of the last instar larva of Argia González-Soriano, E. (2012). Argia mayi, a new species from Mexico barretti Calvert (Zygoptera: Coenagrionidae). Odonatologica, 37, (Zygoptera: Coenagrionidae). Organisms Diversity & Evolution, 3, – – 367 373. 261 265. Novelo-Gutiérrez, R., & González-Soriano, E. (2003). The larvae of González-Soriano, E., & Novelo-Gutiérrez, R. (2007). Odonata of Lestes alfonsoi González & Novelo (Zygoptera: Lestidae). Mexico revisited. In B. K. Tyagi (Ed.), Biology of (pp. Odonatologica, 32,289–294. – 105 136). Jodhpur: Scientific Publishers. Novelo-Gutiérrez, R., & Gómez-Anaya, J. A. (2006). A description of the González-Soriano, E., & Novelo-Gutiérrez, R. (2014). Biodiversidad larvae of Argia funcki (Selys, 1854) (Odonata: Zygoptera: de Odonata en México. Revista Mexicana de Biodiversidad, 85, Coenagrionidae). Proceedings of the Entomological Society of – 243 251. Washington, 108,261–266. Guan, Z., Dumont, H., Yu, X., Han, B. P., & Vierstraete, A. (2013). Novelo-Gutiérrez, R., & Gómez-Anaya, J. A. (2012). Description of the Pyrrhosoma and its relatives: a phylogenetic study (Odonata: larva of Argia percellulata (Odonata: Coenagrionidae). – Zygoptera). International Journal of Odonatology, 16,247 257. International Journal of Odonatology, 15,45–50. Hovmoller, R., & Johansson, F. (2002). A phylogenetic perspective on O’Grady, E., & May, M. (2003). A phylogenetic reassessment of the larval spine morphology in Leucorrhinia (Odonata: Libellulidae) subfamilies of Coenagrionidae (Odonata: Zygoptera). Journal of based on ITS1, 5.8S, and ITS2 rDNA sequences. Molecular Natural History, 37,2807–2834. Phylogenetics and Evolution, 30,653–662. Pérez-Gutiérrez, L. A., & Montes-Fontalvo, J. M. (2011). Description of Ivanova, N. V., & Grainger, C. (2006). Pre-made frozen PCR and se- the last stadium larvae of Argia medullaris Hagen in Selys and quencing plates. CCDB Advances Methods Release, 1,1–3. A. variegata Förster (Odonata: Coenagrionidae). International Ivanova, N. V., de Waard, J., & Hebert, P. D. N. (2006). An inexpensive, Journal of Odonatology, 14 ,217–222. automation-friendly protocol for recovering high-quality DNA. Pessacq, P. (2008). Phylogeny of Neotropical Protoneuridae (Odonata: Molecular Ecology Notes, 6,998–1002. Zygoptera) and a preliminary study of their relationship with related Kambhampati, S., & Charlton, R. E. (1999). Phylogenetic relationship families. Systematic Entomology, 33,511–528. among Libellula, Ladona and Plathemis (Odonata: Libellulidae) Rambaud, A. (2009). FigTree, 1.4.2. Institute of Evolutionary Biology, based on DNA sequence of mitochondrial 16S rRNA gene. University of Edinburgh. http://tree.bio.ed.ac.uk/software/figtree. Systematic Entomology, 24,37–49. Accessed 25 Mar 2016. Kim, M. J., Jung, K. S., Park, N. S., Wan, X., Kim, K. G., Jun, J., Rehn, A. (2003). Phylogenetic analysis of higher-level relationships of Yoon,T.J.,Bae,Y.J.,Lee,S.M.,&Kim,I.(2014).Molecular Odonata. Systematic Entomology, 28,181–239. phylogeny of the higher taxa of Odonata (Insecta) inferred Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D. L., Darling, A., from COI, 16S rRNA, 28S rRNA, and EF1-α sequences. Hohna, S., Larget, B., Liu, L., Suchard, M. A., & Huelsenbeck, J. P. Entomological Research, 44,65–79. (2012). MrBayes 3.2: efficient Bayesian phylogenetic inference and 420 Torres-Pachón M. et al.

model choice across a large model space. Systematic Biology, Ware, J., May, M., & Kjer, K. (2007). Phylogeny of the higher 61,539–542. (Anisoptera: Odonata): an exploration of the most Saux, C., Simon, C., & Spicer, G. (2003). Phylogeny of the dragonfly and speciose superfamily of dragonflies. Molecular Phylogenetics and damselfly order Odonata as inferred by mitochondrial 12S ribosom- Evolution, 45,289–310. al RNA sequences. Annals of the Entomological Society of America, Ware, J., Simaika, J. P., & Samways, M. J. (2009). Biogeography and 96,693–699. divergence time estimation of the relict Cape dragonfly genus Schorr, M., & Paulson, D. (2016). World Odonata list. Slater Museum of Syncordulia: global significance and implications for conservation. Natural History, University of Puget Sound. http://www.ups.edu/ Zootaxa, 2216,22–36. x6140.xml. Accessed 18 Aug 2016. Westfall, M. J., & May, M. L. (1996). Damselfies of the North America. Stoks, R., & McPeek, M. A. (2006). A tale of two diversifications: recip- Gainesville: Scientific. rocal habitats shifts to fill ecological space along the pond perma- Westman, A., Johansson, F., & Nilsson, A. N. (2000). The phylogeny of – nence gradient. The American Naturalist, 168,50 72. the genus Leucorrhinia and the evolution of larval spines Thiele, K. (1993). The holy grail of the perfect character: the cladistic (Anisoptera: Libellulidae). Odonatologica, 29,129–136. – treatment of morphometric data. Cladistics, 9,275 304. Wiens, J. (2001). Character analysis in morphological phylogenetics: von Ellenrieder, N. (2002). A phylogenetic analysis of the extant problems and solutions. Systematic Biology, 50,689–699. Aeshnidae (Odonata: Anisoptera). Systematic Entomology, 27, Yu, X., & Bu, W. (2011). A preliminary phylogenetic study of 437–467. with focus on the Chinese genera Sinocnemis von Ellenrieder, N. (2007). The larva of Argia joergenseni Ris Wilson & Zhou and Priscagrion Zhou & Wilson (Odonata: (Zygoptera: Coenagrionidae). Odonatologica, 36,89–94. Zygoptera). Hydrobiologia, 665,195–203. von Ellenrieder, N. (2013). A revision of Metaleptobasis Calvert (Odonata: Coenagrionidae) with seven synonymies and the Zloty, J., Pritchard, G., & Esquivel, C. (1993). Larvae of the Costa Rican Hetaerina (Odonata: Calopterygidae) with comments on distribu- description of eighteen new species from South America. – Zootaxa, 3738,1–155. tion. Systematic Entomology, 18,253 265.