The status of Rhionaeschna galapagoensis ( Currie , 1901 ) with notes on its biology and a description of its ultimate instar larva ( Odonata , Aeshnidae )

El estado de Rhionaeschna galapagoensis (Currie, 1901) con notas sobre su biologia y una descripcion de su ultimo estadio larvario (Odonata, Aeshnidae) Se presenta una caracterizacion morfologica, molecular y comportamental de Rhionaeschna galapagoensis , basada en una serie de especimenes, tanto adultos como larvas, y observaciones realizadas en la isla de San Cristobal, en las Galapagos. Se ha observado que varios de los caracteres propuestos anteriormente para distinguir entre los adultos de esta especie y de su pariente mas proximo, R. elsia , son variables; sin embargo, se ha confirmado que la presencia de una banda negra en la sutura frontoclipeal es un buen caracter diagnostico. Se describe por primera vez el ultimo estadio larvario de R. galapagoensis y se diagnostica de sus parientes mas cercanos mediante una combinacion de caracteres que incluye el angulo agudo entre las apofisis protoracicas, la ausencia de espinas laterales en el sexto segmento abdominal y la longitud de los cercos en relacion con los paraproctos. El analisis molecular confirmo que R. galapagoensis y R. elsia son especies hermanas, y mostro que la distancia genetica entre ellas es la menor entre las especies analizadas, lo cual es previsible dada la edad reciente de las islas Galapagos. Las larvas de R. galapagoensis eran muy comunes y estaban ampliamente distribuidas en los arroyos de montana y en un estanque en el suroeste de San Cristobal. Se observo la formacion de enjambres de decenas de individuos en la vegetacion costera al amanecer que, junto con adultos de Tramea cf. cophysa , se alimentaban de pequenos insectos. Los machos patrullaban pequenas secciones de los arroyos y en un estanque. Solo se observo una copula, que duro unos 10 minutos. Las hembras ovipositaron solas en la vegetacion flotante de los arroyos y el estanque. Nuestras observaciones corroboran que R. galapagoensis y R. elsia son dos especies parapatricas, morfologica y geneticamente cercanas. Las poblaciones de R. galapagoensis en San Cristobal son grandes y aparentemente no estan amenazadas.


Introduction
Islands are evolutionary laboratories where speciation occurs at a high rate due to isolation (MacArthur & Wilson, 1967;Whittaker, 1998).The Galápagos are especially significant in this context in view of their effect on Charles Darwin's theory of evolution by natural selection (Darwin, 1901).Flying animals, with their high dispersal ability, are less prone to island endemism, but archipelagos situated far from the mainland are sufficiently isolated to promote speciation and evolution of novel traits.This is the case of the Azores Islands, whose poor odonata fauna (only four resident species) harbors the only case of parthenogenesis known in the entire order of Odonata (Cordero Rivera et al., 2005).
The Odonata of the Galápagos, with only nine species currently known for the archipelago (Muddeman, 2007;Peck, 1992), are a clear example of the effect of isolation on colonization events.Only one Rhionaeschna species, R. galapagoensis, has been described from the Galápagos Islands, and it is currently considered the only endemic species of the order in the Galápagos archipelago.Rhionaeschna galapagoensis was described by Currie (1901) as Aeshna galapagoensis from a male and a female collected in the Galápagos Island of San Cristóbal.It was later also described from the islands of Santa Cruz and Isabela (Calvert, 1956;Turner Jr., 1967).The male caudal appendages were depicted in Martin (1908), and the species was listed by Campos (1922) from Ecuador.Calvert (1956) included it in the subgenus Neureclipa Navás and provided a redescription and illustrations of the caudal appendages of males and females.Turner (1967) provided a new island record, and Belle (1991) published a few observations about its behaviour.Aeshna galapagoensis was transferred by von Ellenrieder (2003) to the genus Rhionaeschna, accompanied by a characterization, diagnosis, and illustrations of various morphological characters based on examination of the type specimens.The preliminary phylogenetic analysis of the genus based on morphological characters of the adults (von Ellenrieder, 2003) placed R. galapagoensis as the sister species of R. elsia (Calvert) within a clade including the other species previously included in the subgenus Neureclipa by Calvert (1956).These species share the combination of supratriangles usually free, two rows of cells between RP1 and RP2 in Hw beginning at the distal end of the pterostigma or further distally, and male cercus with dorso-distal crest as high or higher than the width of cercus at base, a prominent sub-basal tooth, and external margin concave (von Ellenrieder, 2003).Both species can be distinguished from all remaining species of Rhionaeschna by the combination of rounded clypeal lobes and ventral tubercle of S1 bearing only a few denticles (10 or less) restricted to its apex (von Ellenrieder, 2003).Calvert (1952) described R. elsia without providing any diagnosis from R. galapagoensis.In his monograph of the group (Calvert, 1956), he used the thoracic and membranule color pattern and shape of male caudal appendages in dorsal view to separate them.Von Ellenrieder (2000) redescribed R. elsia, andlater (von Ellenrieder, 2003) showed that thoracic color pattern and dorsal shape of male cerci were variable in R. elsia and unreliable as diagnostic characters, proposing the presence or absence of a black band over the fronto-clypeal suture, and the shape of the anterior hamule anterior tip, female cercus tip, and male cercus dorso-distal crest in lateral view to distinguish between the two species.Needham (1904) provided a brief larval description of R. galapagoensis based on an early instar larva.The ultimate instar larvae of slightly over half of the known species of Rhionaeschna were described by Calvert (1956), De Marmels (1982, 1990, 2001), Limongi (1983), Müller & Schiel (2012), Musser (1962), Novelo-Gutiérrez & González-Soriano (1991), Rodrigues Capítulo (1980), Rodriguez & Molineri (2014), Santos (1966), von Ellenrieder (1999, 2001), von Ellenrieder & Costa (2002), von Ellenrieder & Muzón (2003), and Walker (1958).
Given the scarcity of specimens of R. galapagoensis available for study to date, further data were needed to assess the status of R. galapagoensis as a different taxon from R. elsia.To this end, we: (i) examined morphological characters of adult and ultimate instar larvae, (ii) provided some observations on general and reproductive behaviour, and (iii) used nuclear and mitochondrial DNA sequences to contrast the specific status for these sister taxa.

Material and methods
All observations were performed at San Cristóbal Island (Galápagos Archipelago) between 20 II and 6 III 2014.Most of the island has no road access so we were limited to the populated areas in the southwest of the island.We sampled permanent streams and ponds around the 'Hacienda El Cafetal'.Further observations were carried out at Punta Carola, where young and mature adults were found in swarms feeding on shrubs of Hippomane mancinella Linnaeus.Specimens collected were preserved in 80% ethanol for further study.
In the laboratory, the variability of characters of adult specimens of R. galapagoensis was studied, documented and compared with those of a series of adult R. elsia in order to re-evaluate which diagnostic characters reliably identified the two species.Illustrations were made with the aid of a camera lucida coupled to a Nikon SMZ1500 stereomicroscope.Exuviae were photographed using a Canon Eos 7D mark II camera, and images were combined by means of a procedure of photo-stacking using Adobe Photoshop CS6 software (www.adobe.com).
Ultimate instar larvae were photographed and measured with an AxionCam ICc3 coupled to a ZEISS Discovery V12 and the software Axion Vision version 4.8.Mandibular formula follows Watson (1956).All measurements are given in millimeters; average dimensions are given as average ± standard deviation; hind wing measurement excludes basal sclerites; total length includes caudal appendages; larval wing cases, lateral spines on abdomen, cercus, and paraproct were measured along their inner margin.Abbreviations used throughout the text are as follows: Dept.: Department; Prov.: Province; Fw: forewing; Hw: hindwing; pnx: postnodal crossveins; S1-10: abdominal segments 1 to 10.
Material studied is detailed in appendix 1.

Molecular analysis and phylogenetic reconstruction
DNA extraction and sequencing DNA was extracted from one leg of each adult dragonfly using a GeneJet Genomic DNA Purification kit (Thermo Scientific, Waltham, USA).Three genes were amplified: mitochondrial cytochrome oxidase I (COI); hypervariable D7 region of the large-subunit 28S (rDNA) (28S); and nuclear Histone 3 (H3), using PCR according to Kohli's et al. (2014) protocol.We selected two mitochondrial and one nuclear gene based on their diverse evolutionary rates, to allow us to reconstruct both internal and external branches, respectively (Fritz et al., 1994).Successfully amplified samples were sent to Macrogen (www.macrogen.com)for bidirectional sequencing.

Genetic distances and phylogenetic reconstruction
Forward and reverse sequences were edited in BioEdit version 7.5.0.3 (Hall, 1999) and consensus sequences aligned with SeqMan DNAStar version 5.03 (www.dnastar.com).Variable positions were revised by eye, and only high quality sequences were considered for further analyses.Genetic distances among the seven Rhionaeschna species sequenced were estimated by using Kimura 2-parameter genetic distances (Kimura, 1980) of the three genes separately; COI (20 sequences, 367 bp), nuclear H3 (21 sequences, 251 bp), and 28S (20 sequences, 454 bp).All samples of each species clustered in the same species-group, and genetic distances were estimated between groups.The gamma distribution (shape parameter = 1) was used to modulate the rate of variation among sites with MEGA 6 (Tamura et al., 2013).
Phylogenetic relationships among haplotypes were also estimated using Bayesian inference with the program MrBayes version 3.0 (Huelsenbeck & Ronquist, 2001).To this end, we investigated models of nucleotide substitution of the three genes and ranked them by Akaike information criterion implemented in the program jModelTest (Posada, 2008).The model Generalized Time-Reversible plus Gamma (GTR)+G (Tavaré, 1986) was inferred as the most appropriate model to estimate nucleotide substitutions for two genes (COI and H3) because it allows for a different rate of transitions and transversions as well as unequal frequencies of the four nucleotides (base frequencies).However, the model of nucleotide substitution inferred for the nDNA gene 28S was the Hasegawa-Kishino-Yano model (HKY)+G (Hasegawa et al., 1985), a model that also allows for a different rate of transitions and transversions as well as unequal frequencies of the four nucleotides (base frequencies).Thus, because the nDNA gene 28S has a different model of nucleotide substitution and a low percentage of informative positions to be analyzed alone (see details in Results section) it was not included in the Bayesian analyses.Therefore, only two genes, COI and H3, were concatenated and analyzed together.We conducted two independent runs that consisted of four Markov chains (one cold and three heated chains) each.We ran 100,000 generations, sampling every 10 generations and discarding the first 2,500 (25%) generations (burn-in time).The resulting phylogenetic tree was rooted with Anax amazili and drawn with FigTree version 1.3.1 (http://tree.bio.ed.ac.uk/).

Distribution and biological observations
Rhionaeschna galapagoensis was found in all streams visited, both as adults and as larvae of various instars (table 2).We recorded abundant specimens at the Camarones Stream inside the 'Hacienda el Cafetal' and in a nearby pond (fig.9B).Teneral and mature specimens were found near roads and in the village of Puerto Baquerizo but were particularly common along the coast at Punta Carola and nearby places.

Diagnosis
Adults of R. galapagoensis and R. elsia can be distinguished from all other species of Rhionaeschna by the combination of clypeal lobes rounded and ventral tubercle of S1 bearing only a few denticles (10 or less) restricted to its apex (von Ellenrieder, 2003).Adults of R. galapagoensis differ from those of R. elsia by the presence of a wide black band over the fronto-clypeal suture (fig.1A).In R. elsia, there is no dark color over the fronto-clypeal suture at all or only a faint narrow brown line (fig.1B).
Among the known ultimate instar larvae of Rhionaeschna, still only about 57% of the species in the genus, R. galapagoensis shares only with R. brasiliensis (von Ellenrieder & Costa), R. elsia, and R. marchali the absence of the lateral spines on S6 (Limongi, 1983;Müller & Schiel, 2012;von Ellenrieder & Costa, 2002).Ultimate instar larvae of R. galapagoensis can be recognized from those of R. brasiliensis and R. elsia by the acute angle between the prothoracic supracoxal apophyses (orthogonal to obtuse in R. brasiliensis and R. elsia), and from R. marchali by the well developed prothoracic supracoxal apophyses (absent in R. marchali).Ultimate instar larvae of R. galapagoensis differ further from those In agreement with larval habitat preference, males were seen patrolling sections of streams and around the pond, and females were seen laying eggs in both types of habitat.Nevertheless, individuals were more commonly observed at the streams than at the pond.Mate-searching males patrolled sections of the stream of a few meters, flying at about 30-50 cm above the surface of the water.They remained at the stream for short periods (a few minutes) and were observed in the morning and afternoon.Females approached the stream and oviposited on floating vegetation.
One female was observed laying eggs at 16:00 h.Another female approached the stream at mid-day (12:20 h), was captured in tandem by a patrolling male, and the pair mated perched in a nearby tree for about 10 min (fig.1A).Swarms of tens of individuals (fig.10A, video) were observed feeding just after sunrise around the beaches and shrubland at Punta Carola.The total number of individuals swarming in the area was clearly enormous, although not easily quantifiable.These swarms appeared at about 6:00 h, immediately after sunrise, and dispersed at about 7:00 h, when the sun became stronger.Some specimens of Tramea cf.cophysa were found in the same swarms.These swarms attracted bird predators, and one successful predation event (on a teneral R. galapagoensis) was observed (fig.10B).

Genetic characterization and phylogenetic reconstruction
Alignments of mtDNA COI, H3, and nDNA 28S fragments included 367, 251, and 454 bp positions respectively.Sequences can be accessed at GenBank under accession numbers provided in table 3. The mtDNA COI fragment showed 56 parsimony-informative positions, while there were only 23 and 17 in nDNA H3 and nDNA 28S.Pairwise genetic distances between the seven Rhionaeschna species ranged from 1.7 to 8.8% for mtDNA A, 1.2 to 4.0% mtDNA C and 0.01 to 0.04% for mtDNA B (table 4).Similar topologies were obtained from Bayesian and maximum likelihood phylogenetic reconstructions.The Bayesian posterior probability approach produced a tree with a topology (based on COI and H3) largely resolved, but the two main clades were not well-supported   the length of cerci relative to paraprocts, as detailed above.As these color and morphological differences between the two species were consistent in all adult and larval specimens available to us in this study, we consider that R. galapagoensis and R. elsia can be maintained as separate species, although they are very close due to their recent speciation.
Rhionaeschna galapagoensis was commonly found in streams, far from the coast, perhaps because freshwater ponds are scarce in San Cristóbal.In contrast, the typical habitat of R. elsia is brackish waters in coastal deserts (Müller & Schiel, 2012).Therefore, our results also suggest that the two taxa have different ecological preferences.Further field studies in other islands and other areas of San Cristóbal are needed to confirm our findings.series of specimens of both species available in this study, and cannot be used as diagnostic.The only seemingly reliable character to distinguish between adults of the two species was the black band over the fronto-clypeal suture.The presence or absence of a black band over the fronto-clypeal suture has been found to be a stable character in other species of this genus, always either present or absent in all specimens of a particular species (von Ellenrieder, 2003).The ultimate instar larvae of both species can be recognized from all other Rhionaeschna larvae so far described except for R. brasiliensis and R. marchali by the obsolete to absent lateral spines on S6.These two species differ in the shape of the angle formed between the prothoracic supracoxal apophyses, the length of the lateral spines on S7-9, and of Although the genetic distances between R. galapagoensis and R. elsia are less than the most common 2% divergence between congeneric pairs of animal species (Hebert et al., 2003), other odonate species considered to be good species show similar levels of interspecific divergence.This is the case of the Ischnura elegans-group of species (I.elegans, I. genei, I. graellsii, and I. saharensis) which show less than 1% divergence (Sánchez-Guillén et al., 2014).Consistently with their low genetic divergence, which is to be expected given the young age of the Galápagos Islands, phylogenetic reconstruction confirms the   monophyletic origin of R. elsia and R. galapagoensis as it had been suggested in the preliminary phylogeny of the genus based on morphology (von Ellenrieder, 2003), with both species clustering together in the three reconstructed trees (Bayesian, Maximum likelihood, and Neighbor-Joining).Even though there was no absolute agreement on the position of all species, all three phylogenetic reconstructions also hint at the paraphyletic nature of the Neureclipa-group, with R. galapagoensis and R. elsia clustering with R. marchali or with R. cornigera and R. marchali, rather than with the remaining species of the Neureclipa-group.
In conclusion, morphological, ecological and genetic evidence indicate that R. galapagoensis and R. elsia can be maintained as closely related but separate species.

Fig. 1 .
Fig. 1. A. A pair of Rhionaeschna galapagoensis in copula photographed by Adolfo Cordero-Rivera on 24 II 2014 in Ecuador, San Cristóbal Island, Camarones stream at Hacienda El Cafetal (arrow in insert points at black band over fronto-clypeal suture); B. Male of Rhionaeschna elsia photographed by Dennis Paulson on 26 IV 2014 in Peru, Lima Department, Chorrillos, Pantanos de Villa near Lima (arrow in insert points at fronto-clypeal suture, devoid of a black band).

Fig. 10 .
Fig. 10.Swarm of adults of Rhionaeschna galapagoensis: A. Feeding just after sunrise around the beaches and shrubland at Punta Carola; B. Documented predation event of birds (a Galapagos flycatcher, Myiarchus magnirostris) on a teneral R. galapagoensis.Photos by Adolfo Cordero-Rivera.

Table 1 .
Measurements of ultimate instar larvae of Rhionaeschna galapagoensis.Measurements are given in mm, as average ± standard deviation followed by range in square brackets.