Confirmation of Anopheles ( Anopheles ) calderoni Wilkerson , 1991 ( Diptera : Culicidae ) in Colombia and Ecuador through molecular and morphological correlation with topotypic material

The Central and Southern American mosquito species Anopheles (Anopheles) calderoni Wilkerson, 1991, Anopheles (Anopheles) guarao Anduze and Capdevielle (1949), Anopheles (Anopheles) malefactor Dyar & Knab 1907 and Anopheles (Anopheles) punctimacula Dyar & Knab 1906 (Arribalzagia Series, Laticorn Section) (Harbach 2004) are highly isomorphic and closely related (Wilkerson 1990, 1991). Following reassessment of specimens from both the type series and type locality [Panama, Colon (Canal Zone)] of An. punctimacula, Wilkerson (1990) detected exemplars of An. malefactor in the syntype series and resurrected the nominal species from synonymy with the former. According to Wilkerson (1990, 1991), An. punctimacula is reported from Mexico to northern Colombia and Venezuela, whereas An. malefactor appears to be limited to Panama and northeastern Colombia. So far, the distribution of An. guarao is thought to be limited to its type locality of Caripito, Estado Monagas, Venezuela (Anduze & Capdevielle 1949). Further assessment of An. punctimacula s.l. specimens from Piura, Salitral, Peru, revealed yet another cryptic species, which Wilkerson (1991) formally named An. calderoni. The description suggested that prior records of An. punctimacula in the low-lying areas (< 250 m) of the Pacific Coast of Colombia, Ecuador and Peru (LeviCastillo 1949, Russell et al. 1963, Calderón et al. 1974) could, at least in part, refer to An. calderoni. Based on characters of the adult females, Rubio-Palis and Moreno (2003) confirmed An. calderoni from the states of Táchira and Barinas, in Venezuela. In light of this geographical discordance, its presence in Colombia and Ecuador seems highly likely. That An. calderoni has been found positive for Plasmodium vivax in the Peruvian regions of Lima, Salitral, Querecotillo, Lourdes Paz and Palo Blanco (Kroeger & Alarcón 1993, Calderón et al. 1995), makes this taxonomic query also one of biomedical importance. Literature records and museum holdings indicate only the presence of An. punctimacula and An. malefactor in Colombia (González & Carrejo 2007), and the morphological keys most commonly used to identify Colombian Anopheles (Suárez et al. 1988) predate the description of An. calderoni. Following the detection of atypical An. punctimacula s.l. specimens from various localities in the valley of the Rio Cauca, progeny broods were obtained to facilitate a thoroughly integrated systematic assessment of these Colombian variants. Identifications of archive specimens of An. punctimacula s.l. in the Entomological Museum, Universidad del Valle, Colombia (MUSENUV) were revised, alongside wildcaught material from Ecuador. Financial support: Royal Caribbean International, Consortium for the Barcode of Life, The Sloane Foundation, The ScholarShip Program LLC + Corresponding author: ranulfog@gmail.com Received 29 April 2010 Accepted 31 August 2010 Confirmation of Anopheles (Anopheles) calderoni Wilkerson, 1991 (Diptera: Culicidae) in Colombia and Ecuador through molecular and morphological correlation with topotypic material


MATERIALS AND METHODS
Specimens -Wild-caught adult females of An. punctimacula s.l. were collected in August 2007, between 18 pm-21 pm, either resting or in human landing catches, at Corporacion Autonoma Regional del Valle del Cauca in Laguna de Sonso,Buga,Valle del Cauca,Colombia (03º52'32''N,76º20'53''W) at an elevation of 960 m. Females were induced to oviposit according to the standard protocols of Belkin et al. (1965) and once eggs were laid, females were pinned and stored in the MUSENUV collections and labelled with the same code as the corresponding progeny. Larvae were fed on a mix of brewer's yeast and rabbit food. Some 4th instar larvae were separated, killed by boiling water and stored in acetic acid, glycerol and alcohol fixative (10 parts 60% ethyl alcohol with 1 part glacial acetic acid and 1 part glycerol) for future morphological examination. Larval and pupal exuviae of link-reared specimens were stored in 75% ethanol prior to slide-mounting following the protocols in González and Carrejo (2007). Emergent adults were allowed to feed on 10% sucrose solution and were later killed by freezing. Adults were pinned and labelled with the same code as the mother for further integrated studies.
Morphological assessment -Morphological identifications were carried out using the diagnostic characters for An. punctimacula s.l. (Wilkerson 1990) and the original species description of An. calderoni (Wilkerson 1991). Wing character nomenclature used herein follows that of Wilkerson and Peyton (1990) (Fig. 1, Tables I-III). Measurements were taken using a micrometer mounted in a Nikon SMZ-645 Stereo Zoom Microscope. For each female (n = 79), variation in the overlap of the postsubcostal dark spot with the preapical dark spot (Table I), the respective lengths of the costal wing spots (Table II) and the ratio of the length of the costal wing spot relative to the total wing length (measured from the humeral cross-vein) (Table III) were calculated. The presence of key diagnostic characters of An. calderoni larvae -numbers of branches on seta 3-C and the positioning of seta 1-X in relation to the saddle (Wilkerson 1991) -were assessed in all mounted larvae (n = 144) (Table IV). Verification of specimens held in the archive collections of the MUSENUV and which had been identified as An. punctimacula were also reassessed (Supplementary data).
Molecular analysis -Full locality details of specimens used in the molecular part of this study are given in Table V. DNA extraction, polymerase chain reaction (PCR) amplification and product clean-up took place on board The ScholarShip (M/V Oceanic II) as part of the activities of the Mosquito Barcoding Initiative (MBI). DNA was extracted from the abdomens of pinned specimens using the QIAGEN ® BioSprint robotic DNA extraction machine and the BioSprint 96 DNA Tissue Kit (QIAGEN ® , Crawley, UK), following the manufacturers instructions. The universal LCO and HCO barcoding primers of Folmer et al. (1994) were used to amplify the cytochrome c oxidase subunit I gene (COI) region (658 bp, less primers). The PCR comprised 1 μL template DNA, 1 μL 10 x NH 4 buffer, 0.5 μL dNTPs at 2.5 mM, 0.3 μL each primer at 10 μM, 0.4 μL MgCl 2 at 50 mM and 0.2 μL of Taq polymerase (BioLine, London, England), made up to 10 μL with ddH 2 0. The PCR reaction comprised initial denaturing at 95ºC for 5 min, then 34 cycles of 95ºC for 30 s, 48ºC for 30 min and 72ºC for 45 s, followed by a 5-min extension at 72ºC and a 10ºC hold. Two microliter of each PCR product was visualized on 1% agarose gels, containing 0.5 mg/mL of ethidium bromide. Amplification of the nuclear ITS2 rDNA region was achieved using the 5.8SF and 28SR primers of Collins and Paskewitz (1996), following the protocol listed in Linton et al. (2001).
PCR products were purified using ExoSAP-IT ® (USB, Cleveland, Ohio), following the manufacturers instructions. Bi-directional DNA sequences were generated either in the Sequencing Facility of the Natural History Museum, London, or at the Museum Support Center of the Smithsonian Institution, Maryland, using the Big Dye ® Terminator Kit (PE Applied Biosystems, Warrington, England) and run on ABI 3730 automated sequencers (PE Applied Biosystems). Sequences were edited using   (Fig. 2). Similarity of the sequences generated in this study was compared with those available in GenBank using Blast (blast.ncbi.nlm.nih.gov/) and sequences were analyzed according to standard barcoding methodology (Hebert et al. 2003). Sequence statistics, calculation of pairwise distance parameters using Kimura 2-parameter algorithm (Kimura 1980) and generation of the bootstrapped neighbor-joining tree (10,000 replicates) were carried out in MEGA v.4.0 (Tamura et al. 2007). Details of isofamily specimens of An. calderoni sequenced from Colombia (n = 7) and wild-caught specimens from Ecuador (n = 5) in this study are given in Table V. In addition, one specimen each of the closely related taxa An. punctimacula [Gamboa, Panama (09º06'N, 79º41'W); colls.: J Loaiza, Y Linton, F Ruiz and students of The ScholarShip LLC] and An. malefactor [Yaviza, Panama (08º11'N, 77º41'W); coll.: J Loaiza] were included in the analysis, but no specimens of An. guarao were available for inclusion. The COI electropherograms and specimen details (including exact localities details with georeferences, details of voucher type and housing, GenBank accessions) are available in the public records of the MBI section of the Barcode of Life Data Systems v. 2.5 (boldsystems.org) (Ratnasingham & Hebert 2007) and appear in GenBank as barcode red flag data, indicating their high quality and voucher standards under the following accessions: An. calderoni (Colombia HQ642964-70, Ecuador HQ642971-74), An. punctimacula (Panama HQ622626) and An. malefactor (Panama HQ622625). The ITS2 sequences of An. calderoni generated in this study are available in GenBank under accessions numbers HQ622618-HQ622622.

Morphological assessment of Colombian An. calderoni
Adults -Nearly all the female progeny from families reared for this study exhibited characters concordant with those given in the original description of An. calderoni as follow: upper mesanepimeron with pale scales, wings with pale yellow scales mixed with white scales, distal postsubcostal pale spot on the costa and the postsubcostal pale spot on the first radial (R 1 ) not contiguous, frequently separated by dark scales at their respective ends, i.e. overlap of scales of the distal part of the postsubcostal dark spot with the proximal part of the preapical dark on vein R 1 (Fig. 1A, B); scales on vein R 1 between the proximal part of the presubcostal pale and the distal postsubcostal pale generally dark (Fig. 1B), i.e. R 1 in the subcostal area is usually all dark except for pale spots at the ends.
Measures of the overlap of scales of the distal part of the postsubcostal dark spot with the proximal part of the preapical dark spot of R 1 were found to be heterogeneous   among families (p < 0.0001) and variable even among female siblings (Table I). This value is positive when there is overlap in the spots, but can also have a value of zero when the ends of the spot are immediately adjacent and a negative value when the spots are clearly separated (Fig. 1B).
Overall, eight of the 10 isofamilies examined exhibited overlapping spots (0.003-0.024 mm), however many of the individuals in isofamily 2 and some from isofamilies 7-9 showed either no overlap but immediately adjacent spots or clear separation of the spots (Table I). An even alignment of the postsubcostal spot on C and R 1 , reportedly typical of all other Arribalzagia Series species (Suárez et al. 1988, Wilkerson 1990, 1991, González & Carrejo 2007, was not observed in the An. calderoni specimens examined here. Considering the above observations, it is necessary to add a clarification of the original description of this character as follows: "distal postsubcostal pale spot on vein C generally not contiguous with the postsubcostal pale on vein R 1 , separated (generally) or not by dark scales of the distal part of the postsubcostal dark and the proximal part of preapical dark on R 1 ". For those specimens that agree with the remaining characters yet where the aforementioned overlap does not occur, it is possible complement species identification by examination of the scales between the proximal presubcostal pale and the distal postsubcostal pale spots located on R 1 . These are usually all dark in An. calderoni, whereas they are predominantly pale in An. punctimacula, An. malefactor and An. guarao (Wilkerson 1991). Analysis of costal spots within the isofamilies revealed that some specimens did not comply with a normal distribution, even after data transformation using wing length; hence a non-parametric Kruskal-Wallis test was used to compare the averages of these measurements among isofamilies. Statistically significant differences were found when comparing lengths of the subcostal area (Fig. 3) as a whole among the isofamilies; yet, the relative size comparisons of individual spots within this area, i.e. the size of the most distal presubcostal pale spot, and the proximal and distal postsubcostal pale spots, did not statistically differ amongst isofamilies (Table II). Furthermore, the average values among isofamilies (Table III) were very similar to those presented in the original description (Wilkerson 1991).
The band or spot on hindtarsomere 5 was generally found not to be as pale as in the original description (Wilkerson 1991). Instead it was observed as a darker band, sometimes grayish or pale brown and was not well defined in 91.1% of the progeny specimens examined. This darker band was also noted in 94.6% of specimens collected in human landing catches from the same locality (Valle de Cauca, n = 37).
The original description of An. calderoni makes no reference to variation in the pale and dark spots of the 4th hindtarsomere 4, which are otherwise widely employed as diagnostic characters in several morphological keys for species of the Arribalzagia Series (Gorham et al. 1973, González & Carrejo 2007. There are generally five spots of varying length on hindtarsomere 4: two dark and three pale (basal, apical and central). Spots reportedly differ in length, with the two pale spots (basal, apical) being smaller than the dark spots. In 91.6% of Colombian An. calderoni examined (n = 79), the central pale spot separated the two dark spots; whereas in the remaining specimens the central pale spot was not clear, allowing the two dark spots to merge and appear more like a band. The central pale spot was found to be of variable width, and was mostly light grey in color. This character was even found to be variable between legs of the same individual therefore we discourage the further use of this character for species diagnosis in the Arribalzagia Series.
Larvae -The number of branches in seta 3-C and the position of seta 1-X were assessed in all 144 specimens from 10 isofamilies (Table IV). That seta 1-X is not present on the saddle is a shared characteristic of both An. punctimacula and An. calderoni. In 98.2% of progeny examined, seta 1-X was not found on the saddle itself, with 81.5% attached outside this structure altogether and 16.7% placed on its border (Table IV). Most commonly, An. calderoni larvae exhibited 4-7 branches on seta 3-C (range 1-11; mean 5.4; mode 6). The mean number of branches is slightly higher than the 4.2 reported in the original description (Wilkerson 1991), most probably due to increased sample size. An. punctimacula differs significantly in having an average of 11.5 branches hence the combination of these two characters allows morphological differentiation of larval An. calderoni and An. punctimacula.

DISCUSSION
Revision of MUSENUV archive specimens -Retrospective identification of archive specimens of An. punctimacula and An. malefactor held in the MUSENUV revealed a significant proportion of specimens were infact An. calderoni. Specimens of An. calderoni were detected in collections from 1948-2009, in 35 localities in 20 municipalities, representing 12 departments of Colombia (Supplementary data). Although the distribution of An. calderoni in Colombia appears predominantly coastal (both on the Pacific and Atlantic coasts) (Fig. 3), some of the departments (e.g. Tolima) are far inland. Additionally, several collection sites in Valle de Cauca and Quindio were over 1000 m in altitude (Supplementary data), considerably higher than the 250 m limit previously reported for early reports of An. calderoni in Peru (Wilkerson 1991) and 200-400 m reported in Venezuela (Rubio-Palis & Moreno 2003). Rather our data supports the findings of Calderón et al. (1995) who reported An. calderoni from sea level to 1080 m in Peru, suggesting that the distribution of An. calderoni could be considerably more extensive than at first thought.
Molecular characterization of An. calderoni -ITS2 rDNA sequences generated from two specimens [PE400(2)-29, PE400(2)-31] belonging to the type series of An. calderoni from Salitral (Wilkerson 1991) were identical to those from three specimens from Ecuador (Table V), thus verifying the new country record of An. calderoni in Ecuador. Correlation of the 658 bp COI DNA barcode of one of the verified Ecuadorian An. calderoni specimens (EC229-1) allowed the confirmation the conspecific identity of additional wild-caught adults from two other sites in Ecuador (Guayas, n = 4, Pichincha, n = 1) and progeny broods collected in Valle de Cauca, Colombia (n = 7 from isofamilies 1, 4 and 5) ( Table  V), used in the morphological assessment (Table V).
In total, DNA barcodes were generated for 12 specimens (Colombia, n = 7, Ecuador, n = 5), which varied in AT richness from 66-66.5%. Six unique haplotypes were determined, as follows: haplotypes 1 (n = 6) and 2 (n = 1) (both Valle de Cauca, Colombia), haplotype 3 [Guayas (n = 1) and Pichincha (n = 1), Ecuador] and haplotypes 4-6 (n = 1, each one from Guayas, Ecuador) (Figs 2, 3). Within Colombia, haplotypes 1 and 2 varied by only one base (A/G at base 319) (Fig. 2). Overall, 13 (1.97%) variable bases were noted, 10 of which were parsimony informative. Mean pairwise (K2P) sequence diversity within An. calderoni was 0.70%, well within values of intraspecific variation reported for the barcoding region in other mosquito species (Cywinska et al. 2006, Azari-Hamidian et al. 2010. Haplotypes were unique by country, with mean intra-country pairwise distances of 0.04% and 0.77% within Colombian and Ecuadorian specimens, respectively, and an overall mean inter-country sequence divergence of 1.03%. Amino acid translation of the COI sequences revealed that all nucleotide changes were synonymous, suggesting that these represent population level differences only. Differentiation of An. calderoni, An. punctimacula, An. malefactor and An. guarao -Given that An. calderoni has often been misidentified as An. punctimacula in Colombia (herein), we set out to assess the utility of DNA barcodes for species differentiation between these species and other closely related taxa -An. malefactor and An. guarao. No specimens of An. guarao were available for molecular comparison, but the morphological distinctiveness of this species makes it unlikely to be confused with An. calderoni. The robust boot strap values illustrated on the neighbor-joining tree generated using the COI barcode data of An. calderoni, An. punctimacula and An. malefactor confirmed the separate species status of these taxa (Fig. 4). In comparison to the 0.70% interspecific variation noted within An. calderoni, interspecific comparisons with those of Panamanian An. punctimacula and An. malefactor generated in this study, revealed 5.8% and 10.9% variation, respectively (Fig. 4). COI barcodes of An. malefactor (67.7% AT) were most similar to those of An. punctimacula (67.1% AT), differing by 9.6%.
Neither An. calderoni nor An. malefactor are currently represented in GenBank and thus these sequences also comprise the first DNA sequences reported for these species. Although the GenBank entry AF417719 of Sallum et al. (2002) partially overlapped our An. punctimacula sequence by 93 bases, the sequences generated herein represent the first published DNA barcodes for that species. This study shows that DNA barcodes will aid correct identification of An. calderoni, An. malefactor and An. punctimacula in future studies and will allow the vector status of An. calderoni and An. malefactor to be more accurately assessed.
DNA verification of Ecuadorian and Colombian specimens as An. calderoni along with the morphological reassessment of archived MUSENUV specimens significantly adds to the knowledge of the distribution of An. calderoni in Latin America (Fig. 3). Given that the known distribution of An. calderoni now includes Ecuador and Colombia, as well as Peru and Venezuela, and its prior taxonomic confusion with An. punctimacula and An. malefactor, it is now prudent to carry out further collections to determine the true distributions and specific roles in malaria transmission of these three species across the Neotropical Region.