Publisher: CSIRO; Journal: IS:Invertebrate Systematics

Article Type: research-article; Volume: ; Issue: ; Article ID: IS12030

DOI: 10.1071/IS12030; TOC Head:

High species turnover of the ant genus Solenopsis (Hymenoptera : Formicidae) along an altitudinal gradient in the Ecuadorian Andes, indicated by a combined DNA sequencing and morphological approach

Thibaut DelsinneA,C, Gontran SonetB, Zoltán T. NagyB, Nina WautersA, Justine JacqueminA and Maurice LeponceA

ABiological Evaluation Section, Royal Belgian Institute of Natural Sciences, Rue Vautier 29, B-1000 Brussels, Belgium.

BJoint Experimental Molecular Unit, Royal Belgian Institute of Natural Sciences, Rue Vautier 29, B-1000 Brussels, Belgium.

CCorresponding author. Email:

Solenopsis is a widespread ant genus. T and the identification of its species is notoriously difficult. Hence, investigation of their distribution along elevational gradients is challenging. Our aims were (1) to test the complementarity of the morphological and DNA barcoding approaches for Solenopsis species identification; , and (2) to assess species diversity and distribution along an altitudinal gradient in the Ecuadorian Andes. Ants were collected in five localities between 1000 and 3000 m aslabove sea level. In total, 24 morphospecies were identified along the gradient and 14 of them were barcoded. Seven morphospecies were confirmed by the molecular approach. Three others, occurring sympatrically and possessing clear diagnostic characters, showed low genetic divergence. Representatives of a further four morphospecies were split into nine clusters by COI and nuclear wingless genetic markers, suggesting the existence of cryptic species. Examination of gynes revealed potential diagnostic characters for morphological discrimination. Solenopsis species were found up to the an altitude altitudinal record of 3000 m. Most morphospecies (20/ of 24) were found at a single elevation. Our results suggest a high species turnover along the gradient, and point to the use of morphological and DNA barcoding approaches as necessary for differentiating among Solenopsis species.

IS12030

High altitudinal species turnover of Solenopsis ants.

T. Delsinne et al.

Manuscript received 18 April 2012, accepted 16 September 2012, published online dd mmm yyyy

Introduction

Solenopsis Westwood, 1840 is a large myrmicine ant genus encompassing 183 species, with a worldwide distribution (Guénard et al. 2010; Bolton 2012). The most well known species of this genus inflict a painful sting and are known as fire ants. Some fire ants, like such as Solenopsis invicta, have become important invasive pests (Tschinkel 2006). Other members species are referred to as thief ants because some of them are known to steal food from other ants (Pacheco 2007). About half of all described Solenopsis species are found in the Neotropical region (Fernández and Sendoya 2004). Solenopsis nests can be found virtually everywhere, in soil, leaf litter, dead wood, epiphytes or plant cavities (Creighton 1950). Workers forage from deep in the soil (Ryder Wilkie et al. 2007) to high in the forest canopy (Blüthgen et al. 2000). They are one of the most frequently encountered ant genera in ground-dwelling ant communities (Ward 2000; Donoso and Ramón 2009; Braga et al. 2010) and, as a consequence of their diversity and abundance, they are considered to be of significant ecological importance in the Neotropics (Ward 2000).

Members of the genus can be easily differentiated from other Myrmicinae by the 10-segmented antenna with a 2-jointed club, the propodeum rounded and unarmed, the petiole and postpetiole nodes well developed, and the clypeus longitudinally bicarinate with an isolated median seta (Ettershank 1966; Bolton 2003). Identification to specific level is, however, extremely difficult. Mackay and Mackay (2002) indicated that ‘identification is nearly impossible’. Creighton (1930), in his incomplete revision of the New World Solenopsis, wrote: ‘Carlo Emery once characterised the genus Solenopsis as the crux myrmecologorum. That the term is apt no one who has experienced the difficulties of the group will deny, least of all the author who, at the end of three years of study, still finds the «cross» a heavy burden’. In fact, the literature abounds with quotations describing similar opinions. For instance: ‘the genus Solenopsis is no favorite of ant taxonomists’ (Thompson 1989), ‘at least some of [Solenopsis] are exceedingly difficult to classify’ (Smith 1943), ‘the members of the thief ant group of the genus Solenopsis have had a notorious reputation of being difficult to identify for over 70 years (.) … this reputation is merited’ (Pacheco 2007). Snelling (2001) indicated, while describing S. maboya, a new thief ant from Puerto Rico, that ‘maboya is the Taino (Arawak) word for a perverse spirit, and seemed appropriate, given the challenging nature of the taxonomy of this group of ants’.

The difficulty to of identifying specimens of Solenopsis is explained by two factors. First, worker morphology monotonously lacks diagnostic characters. Thief ants are tiny in size, often less than 2 mm long, which complicates the recognition of morphological characters. Fire ants are larger but polymorphic, presenting a continuum of sizes in the same nest. For instance, workers of S. invicta range from 2.65 mm to 6.16 mm in body size (Tschinkel et al. 2003). Moreover, all species of fire ants and several thief ant species exhibit intraspecific variation in morphological traits which may exceed interspecific differences (Pitts et al. 2005; Pacheco 2007; Ross et al. 2010). Males and gynes may be less uniform morphologically and offer additional characters for species identification (Creighton 1950). However, these reproductive castes are less frequently encountered and rarely associated with workers (Creighton 1950; Pacheco 2007). Second, most species were inadequately described based onon the basis of limited material, mainly between the end of the 19th and beginning of the 20th centuries. The use of numerous trinomials and quadrinomials has generated serious taxonomical confusion (Pacheco 2007). Creighton (1930) attempted to revise New World Solenopsis but most thief ants were not included in his work since he planned to treat them in a separate publication which never eventuated. Trager (1991) restricted his revision to fire ants but, even afterwards, species delimitation often remains problematic (Ross et al. 2010). More recently, Pacheco (2007) revised New World thief ants, proposed eight species complexes and numerous new synonymies and other changes, recognised 83 species and presented keys for the identification of workers. Unfortunately, this thesis does not conform to the rules of the International Code of Zoological Nomenclature, making it impossible to acknowledge his taxonomical changes (William Mackay, pers. comm.).

Thus, the α-taxonomy of Solenopsis is still confused, which represents a serious impediment for ant biodiversity inventories, and ecological work in general. Most collected species are misidentified (Thompson 1989) or are only simply recorded only as morphotypes. For instance, 13 Solenopsis species were sampled in the Nouragues Research Station, in French Guiana, but only two of them could be assigned to a valid name (Groc et al. 2009). Similarly, only one of 15 species collected during a thorough inventory in Ecuadorian Amazonian forests was named (Ryder Wilkie et al. 2010).

Recently, the use of DNA barcodes, short mitochondrial DNA sequences of the cytochrome oxidase oxidase1 I (COI) gene, has been proposed to facilitate species identification and discovery (Hebert et al. 2003; Janzen et al. 2009). This method is acknowledged as a useful explorative tool to provide estimates of species numbers, especially in very diverse and poorly understood taxonomic groups (Wiemers and Fiedler 2007; Jansen et al. 2009; Strutzenberger et al. 2011; Tänzler et al. 2012). In particular, DNA barcoding has proved useful in complementing morphological species determination in biodiversity surveys of ants (Smith et al. 2005; Fisher et al. 2008; Fisher and Smith 2009), in facilitating caste association (Fisher et al. 2008), and assisting in the discovery of new cryptic ant species (Schlick-Steiner et al. 2006; Fisher et al. 2008; Menke et al. 2010). Nevertheless, the barcoding approach possesses several pitfalls and shortcomings (reviewed in Rubinoff et al. 2006; Jinbo et al. 2011) with barcoding success rate varying among taxa (Elias et al. 2007; Jansen et al. 2009; Wild 2009). Therefore, species hypotheses based on DNA barcodes should be supported by additional, independent nuclear markers (Ross et al. 2010; Smith et al. 2011) or other data such as morphology, geography, ecology or behaviour (Yassin et al. 2010).

For unknown reasons, a previous attempt to amplify the COI marker from thief ants was not successful (Pacheco 2007) and, so far, most genetic studies of Solenopsis have focussed on fire ants (Ross and Shoemaker 2005; Shoemaker et al. 2006; Ross et al. 2010). These analyses have identified genetically independent lineages within variable and widespread taxa. Further, the combined use of mitochondrial and nuclear markers have revealed cryptic species (Ross et al. 2010).

Identification of Solenopsis species is expected to be particularly challenging along elevational gradients, where it is frequently found that ants once considered to belong to a single widely distributed species turned out to be several cryptic species with parapatric distributions and restricted altitudinal ranges (Lattke 2003). Our aims in this study were (1) to test the complementarity of morphological and DNA barcoding approaches for species identification in the genus Solenopsis; , and (2) to assess species diversity and distribution along an altitudinal gradient in the Ecuadorian Andes, which is considered as to be a biodiversity hotspot for many taxa.

Materials and methods

Ant sampling

Ants were collected between 2007 and 2011 in seven forested sites spread among five localities between 1000 m and 3000 m above sea level with elevational steps of 500 m between localities. Study sites were selected in the Podocarpus National Park and two adjacent protected areas (Reserva Biológica San Francisco and Copalinga private reserve), on the eastern range of the South Ecuadorian Andes, in the provinces of Loja and Zamora-Chinchipe. Details about the study area are provided in Beck et al. (2008). Five reference sites were selected: Copalinga Private Reserve -–blue trail (called thereafter ‘1050 m-C’; Lat., Long.: S 4°5¢S, W 78°57¢W’), Copalinga Private Reserve- – red trail (‘1420 m’; S 4°5¢S, W 78°58¢W’); Reserva Biológica San Francisco- – Transect T1 (‘2070 m-R1’; S 3°58¢S’, W 79°5¢W), El Tiro (‘2500 m’; S 3°59¢S’;, W 79°7¢W’) and Cajanuma-Podocarpus National Park (‘3000 m’; S 4°6¢S’;, W 79°10¢W’). Two supplementary sites were sampled at 1050 and 2070 m: Bombuscaro-Podocarpus National Park (‘1050 m-B’; S 4°6¢S’, W 78°58¢W’) and Reserva Biológica San Francisco-NUMEX (‘2070 m-R2’; S 3°58¢S’, W 79°4¢W’). Distance The distance between sites ranged from 2 to 20 km. At each site, ants present in quadrats of leaf litter were extracted by the Winkler method (54 m2 extracted per site). In addition, we searched for Solenopsis nests in dead wood, soil and vegetation in an attempt to document reproductive castes in association with series of workers. Specimens were preserved in 96% ethanol (denatured with diethyl ether) and sorted to morphospecies based onon the basis of criteria proposed by Pacheco (2007), such as expression of clypeal teeth, number of ommatidia, number of mandibular teeth, scape length, body colour, pattern and extent of sculpture, shape and size of body tagma, expression of anteroventral petiolar process and size of cephalic punctures. We used the phenetic species concept and expected that a certain degree of difference in morphological characters indicated potential reproductive isolation.

A few specimens from each morphospecies were pinned and photographed. Images were taken with a Leica DFC290 camera attached to a Leica Z6APO stereomicroscope. A series of images was taken by focusing on different levels of the insect, using the Leica Application Suite v38 (2003–2011) and combined with CombineZP (Hadley 2010). Final processing of images was done in Adobe Photoshop CS5. Original images were deposited in Morphbank (collection numberno.: 801203; http://www.morphbank.net/801203). Voucher specimens were deposited at the Royal Belgian Institute of Natural Sciences, Brussels, Belgium, and at the Universidad Técnica Particular de Loja, Loja, Ecuador.

Laboratory method

About 10 260 Solenopsis specimens were collected during this study (0 up to 4605 specimens/ per site). Multiple representatives (n = 2–70) of each of the most abundant Solenopsis morphospecies were selected for DNA analysis. Seven morphospecies represented by less fewer than 5 five individuals were discarded. Analyses were carried out on 187 Solenopsis specimens (Supplementary material S1). Total genomic DNA was isolated from the complete ant body using the commercial NucleoSpin Tissue Kit (Macherey-Nagel, Germany). After DNA extraction, specimens were preserved as vouchers in absolute ethanol. Amplification of the mitochondrial cytochrome cytochromec oxidase subunit subunitI (COI) marker was carried out in polymerase chain reaction (PCR) using the primer pair LF1 and LR1 (Smith et al. 2005) modified from Hebert et al. (2004a) and the universal primers LCO1490 and HCO2198 (Folmer et al. 1994). When amplification systematically failed, DNA quality was checked on 1.2% agarose gel, and smaller DNA fragments were amplified using the primer combination LCO1490 and LCO-ANTMR1D-RonIIdeg_R (Fisher and Smith 2008) modified from Simon et al. (1994). Amplification of the nuclear wingless (wg) marker was performed for a selection of 1–3 sample(s) per COI haplogroup using primers wg578F (Ward and Downie 2005) and wg1032R (Abouheif and Wray 2002). Each PCR was prepared in a total volume of 25 µL containing 2 µL of DNA template and 0.03 U/ µL–1 of Platinum® Taq DNA polymerase (Life Technologies, USA), 1´X PCR buffer, 0.2 mm dNTPs, 0.4 μm of each primer, 1.5 mm MgCl2. PCR protocol followed the profile of 94°C for 3 min; 5 cycles of 94°C for 30 s, 45°C for 30 s and 72°C for 60 s; 36 cycles of 94°C for 30 s, 50°C for 30 s and 72°C for 60 s; followed by a terminal elongation step at 72°C for 7 min, and subsequent storage at 4°C. PCR products were visualised on ~1.2% agarose gel, and purified using the NucleoFast 96 PCR Plate (Macherey-Nagel, Germany). PCR products were sequenced with an ABI 3130xl automated capillary sequencer using BigDye v1.1 chemistry and following the manufacturer’s instructions (Life Technologies, USA).

Genetic data analysis

DNA sequences were checked for quality and aligned by hand. No internal stop codons were detected. Homologous fragments of COI sequences of Solenopsis available in GenBank and BOLD (Ratnasingham and Hebert 2007) were added to the dataset provided that no characters were missing. As the length of the sequences obtained varied from 237 to 658 bp, three datasets were created: one including a maximum number of samples but with short sequences (237 bp) and two with longer sequences but including fewer samples (310 and 631 bp). Distributions of pairwise uncorrected distances were plotted for all genetic datasets using the R language and environment for statistical computing and graphics version ver. 2.14.2 (R developmental core team) and package ape v2.7–-3 (Paradis et al. 2004). For an overview of the pairwise genetic distances, a neighbour-joining tree with bootstrapping (1000 replicates) was constructed based onon the basis of the uncorrected distance matrix of the 631 631-bp dataset and using MEGA v5.01 (Tamura et al. 2011). Putative species delimitation was performed using uncorrected distances without a phylogenetic tree reconstruction in which the assumption of monophyly would be doubtful based on a single gene and incomplete sampling (Taylor and Harris 2012). In the absence of species identifications, intraspecific distances could not be distinguished from interspecific distances and no optimal threshold distance could be defined for species delimitation. For this reason, different threshold values were used. Since intraspecific distances are expected to be generally lower than interspecific distances— – forming a ‘barcoding gap’—( – (Hebert et al. 2004b), local minima in the distribution of genetic distances can be used as tentative threshold distance to test for delineation of species. All local minima of the density of the pairwise distances were determined using the function localMinima of package spider v1.1–-2 (Brown et al. 2012) and were used as thresholds. Based onOn the basis of the literature, we also selected threshold values of 2% and 10%. The former was proposed as a standard distance for ants (Smith et al. 2005; Smith and Fisher 2009) and the latter represented an extreme value rarely surpassed by intraspecific distances (e.g. Smith and Fisher 2009; Yassin et al. 2010).