Research Article |
Corresponding author: Paula Rozo-Lopez ( paula.rozo.lopez@gmail.com ) Academic editor: Art Borkent
© 2015 Paula Rozo-Lopez, Ximo Mengual.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Rozo-Lopez P, Mengual X (2015) Mosquito species (Diptera, Culicidae) in three ecosystems from the Colombian Andes: identification through DNA barcoding and adult morphology. ZooKeys 513: 39-64. https://doi.org/10.3897/zookeys.513.9561
|
Colombia, one of the world’s megadiverse countries, has a highly diverse mosquito fauna and a high prevalence of mosquito-borne diseases. In order to provide relevant information about the diversity and taxonomy of mosquito species in Colombia and to test the usefulness of DNA barcodes, mosquito species collected at different elevations in the departments of Antioquia and Caldas were identified combining adult morphology and barcode sequences. A total of 22 mosquito species from eight genera were identified using these combined techniques. We generated 77 barcode sequences with 16 species submitted as new country records for public databases. We examined the usefulness of DNA barcodes to discriminate mosquito species from the Neotropics by compiling 1,292 sequences from a total of 133 species and using the tree-based methods of neighbor-joining and maximum likelihood. Both methodologies provided similar results by resolving 105 species of mosquitoes separated into distinct clusters. This study shows the importance of combining classic morphological methodologies with molecular tools to accurately identify mosquitoes from Colombia.
Culicomorpha , nematocerous Diptera, Neotropical Region, species identification, combining methodologies
Culicidae currently comprises 3,543 formally recognized species distributed throughout most types of habitats and ecosystems of the world (
Mosquito identification is traditionally based on dichotomous keys constructed from morphological features taken for a particular life stage or gender (
The DNA barcode method was postulated by
In mosquitoes, the effectiveness of the COI barcode marker for specimen identification has been tested in surveys of Canada (
Colombia is located in the northwest of South America. It comprises a variety of biogeographic regions that have contrasting biophysical characteristics and high environmental variability (
In order to improve the mosquito knowledge in Colombia, we tested the effectiveness of the barcoding methodology to support the reliable identification of Neotropical species of mosquitoes, previously identified with morphological characters, collected over three different altitudinal ecosystems of the Colombian Andes (Antioquia and Caldas departments).
The study area is located in the west central Andean region in Colombia. Fieldwork was conducted during September 2013 in rural areas of Antioquia and Caldas departments (Fig.
Map of Colombia indicating the sampling sites of mosquitoes collected in this study: Belmira, Antioquia, paramo, 3,200 masl (red circle); Rio Sucio, Caldas, cloud forest, 1,960 masl (green circle); La Pintada, Antioquia, tropical dry forest, 660 masl (blue circle); Supia, Caldas, rural area, 1,150 masl (pink circle). Modified from Instituo Geográfico Agustín Codazzi (www.igac.gov.co) and Wikimedia Commons (by Addicted04).
Mosquito adults were collected using one malaise trap and three Centers for Disease Control-CDC light traps in each habitat type (A or B) of the three biomes, totaling six sites. CDC traps collected for 14 hours, between 5:00 pm and 7:00 am of the next day during two nights in each of our six locations plus the rural area of Supia. Malaise traps were placed and left for 48 hours in each location. Additional specimens were obtained by aspirating mosquitoes attracted to humans during the placement and operation of the traps. All data in the sampled localities were plotted following the protocols of
Wild-caught adults were killed using ethanol (90%) fumes in a lethal chamber to ensure DNA preservation. All mosquitoes were kept dry and individually transferred to labeled 1.5 ml tubes. Each tube was labeled, pierced with a mounted needle (to allow the escape of moisture) and placed in plastic bags containing silica gel. Specimen mounting techniques were conducted using the protocols of Walter Reed Biosystematics Unit-WRBU (
All the specimens collected were identified by female morphology and male genitalia. Since there is no single morphological key to facilitate the identification of mosquitoes of Colombia, genus level identifications were made with multiple approaches using the dichotomous keys of
Voucher specimens and associated genitalia preparations are stored in the entomological collections of the Zoologisches Forschungsmuseum Alexander Koenig (ZFMK), Bonn (Germany). Genomic DNA extracts are stored at –80 °C in the biobank collections of the ZFMK for future reference. Details of reference material are listed by genus in Suppl. material
DNA was extracted from legs and occasionally abdomens (for specimens without legs or with deficient amplification) following standard protocols of the commercially available DNeasy Blood & Tissue Kit (QIAgen®). The COI barcode region was amplified using the forward primer LCO1490 (5’-GGTCAACAAATCATAAAGATATTGG-3’) and the reverse primer C1N2191 (alias Nancy) (5’CCCGGTAAAATTAAAATATAAACTTC-3’) (
To test variation in the barcoding region across a greater geographical area, barcoding sequences of mosquito species listed for the Neotropics were downloaded from BOLD and GenBank between December 2013 and February 2014 for all identified species with a minimum length of 480 bp of COI barcoding region, no stop codons and alignment without gaps (Suppl. material
Lutzomyia longipalpis (Lutz & Neiva, 1912) (Diptera: Psychodidae) was constrained as outgroup. We also included several other genera from three different families as outgroups, i.e. Chironomidae [Cricotopus bicinctus (Meigen, 1818), Chironomus decorus Goetghebuer, 1927, Chironomus kiiensis Tokunaga, 1936, Dicrotendipes tritomus (Thienemann & Kieffer, 1916), and Tanytarsus guerlus (Roback, 1957)], Simuliidae [Simulium ochraceum Walker, 1861, Simulium inaequale (Paterson & Shannon, 1927), Gigantodax abalosi Wygodzinsky, 1958, and Gigantodax basinflatus Wygodzinsky & Coscaron, 1989] and Dixidae [Dixella sp.]. A total of 11 outgroup taxa were included. All outgroup taxa sequences were downloaded from BOLD (Suppl. material
We used similarity methods based on the match between the query sequence and the reference database [e.g. BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and BOLD Identification System (IDS) (http://www.boldsystems.org/index.php/IDS_OpenIdEngine)] and clustering in Tree-based identifications (using Neighbour-joining and maximum likelihood approaches) in order to analyze the DNA barcode region of the mosquitoes collected in Colombia and assign individuals to a given species.
The Neighbour-Joining (NJ) tree (
To perform the maximum likelihood analysis, the data set was divided into three partitions according to the codon positions of COI (first, second, and third positions). We determined the best choice of model for each partition using PartionFinder v1.1.0 (
A total of 77 mosquito specimens were collected from four sampling sites during our study (Table
Mosquito species collected. Four collection sites of the Colombian Andes during September 2013. One specimen collected in Rio Sucio (habitat A) still remains undetermined due to damage and lack of conspecific sequences for comparison.
Mosquito species | Sites | ||||||
---|---|---|---|---|---|---|---|
Supia 1,150 masl | Rio Sucio 1,960 masl | La Pintada 660 masl | Belmira 3,200 masl | ||||
Species | Habitat A | Habitat B | Habitat A | Habitat B | Habitat A | Habitat B | |
Anopheles neomaculipalpus | 1 | ||||||
Coquillettidia nigricans | 1 | ||||||
Culex coniger | 10 | ||||||
Culex conspirator | 5 | ||||||
Culex spp. [coronator complex] | 1 | 1 | 3 | ||||
Culex declarator | 1 | 10 | |||||
Culex educator | 1 | ||||||
Culex erraticus | 6 | 2 | |||||
Culex erythrothorax | 1 | ||||||
Culex lactator | 3 | ||||||
Culex lucifugus | 6 | ||||||
Culex nigripalpus | 1 | 2 | |||||
Culex spinosus | 1 | ||||||
Culex spissipes | 1 | ||||||
Culex theobaldi | 1 | ||||||
Culex (Culex) sp. | 1 | ||||||
Culex (Melanoconion) sp. | 4 | 1 | |||||
Haemagogus janthinomys | 2 | ||||||
Haemagogus lucifer | 1 | ||||||
Ochlerotatus angustivittatus | 1 | 1 | |||||
Ochlerotatus euiris | 1 | ||||||
Psorophora cingulata | 1 | ||||||
Psorophora ferox | 2 | 1 | |||||
Trichoprosopon evansae | 1 | ||||||
Wyeomyia luteoventralis | 1 | ||||||
Undetermined | 1 | ||||||
Total | 4 | 16 | 1 | 45 | 10 | 1 | 0 |
Twenty one species belonging to seven genera were successfully identified by morphological characteristics of adult females and male genitalia: Coquillettidia nigricans Coquillett, 1904 [n=1 female], Culex conspirator Dyar & Knab, 1906 [n=5 males], Cx. corniger Theobald, 1903 [n=8 females], Cx. declarator Dyar & Knab, 1906 [n=8; 5 females, 3 males], Cx. educator Dyar & Knab, 1906 [n=1 male], Cx. erraticus [n=2 males], Cx. erythrothorax Dyar, 1907 [n=1 female], Cx. lactator Dyar & Knab, 1906 [n=2 females], Cx. lucifugus Komp, 1936 [n=3 males], Cx. nigripalpus Theobald, 1901 [n=2; 1 female, 1 male], Cx. spinosus Lutz, 1905 [n=1 male], Cx. spissipes Theobald, 1903 [n=1 female], Cx. theobaldi Lutz, 1094 [n=1 male], Haemagogus janthinomys Dyar, 1921 [n=2 females], Hg. lucifer Howard, Dyar & Knab, 1913 [n=1 female], Ochlerotatus angustivittatus Dyar & Knab, 1907 [n=2; 1 female, 1 male], Oc. euiris Dyar, 1922 [n=1 female], Psorophora cingulata Fabricius, 1805 [n=1 female], Ps. ferox von Humboldt, 1819 [n=3; 2 females, 1 male], Trichoprosopon evansae Antunes, 1942 [n=1 female], and Wyeomyia luteoventralis [n=1 female]. One specimen of the genus Anopheles was identified by morphology to the group: neomaculipalpus / punctimacula. It was only possible to identify the species using its gene sequence.
All the collected specimens were successfully amplified and sequenced, generating a total of 77 sequences with lengths ranging from 618 to 699 bp. Only high quality sequences were retained. Sequences from 14 species are submitted as new records for public databases: Cq. nigricans, Cx. conspirator, Cx. educator, Cx. lactator, Cx. lucifugus, Cx. spissipes, Cx. theobaldi, Hg. janthinomys, Hg. lucifer, Oc. angustivittatus, Oc. euiris, Ps. cingulata, Tr. evansae, Wy. luteoventralis.
All the DNA sequences obtained were compared to those available, by 25 January 2015, in GenBank and BOLD Identification System. At the level of genus, BLAST accurately discriminated 92% of the genera previously identified, whereas BOLD accurately discriminated 70% of genera previously identified. Although BLAST represented a higher percentage of accurate discrimination of genera in our queries, five specimens with a score of more than 89% were wrongly matched: Trichoprosopon was mismatched as Anopheles, Wyeomyia was mismatched as Sabethes Robineau-Desvoidy, 1827, Haemagogus was mismatched as Ochlerotatus (or Aedes), and for the sequence of Haemagogus lucifer mixed results of Spilogona Schnabl, 1911, Haematobia Lepeletier & Serville, 1828 (family Muscidae), and Culex were obtained.
By adding GenBank, BOLD, and unpublished sequences to our data set, it was possible to obtain taxon coverage of 68% of the mosquito genera and 34% of the mosquito species listed for Colombia. Moreover, our data set coverage for Neotropical mosquito species corresponds to 58% of mosquito genera and 12% of mosquito species. The data set of the COI sequences of the Neotropical mosquitoes comprises a total of 1,292 barcode sequences belonging to 133 species and 21 genera (with a minimum length of 640 bp). The alignment was unambiguous: no gaps and amino acid translations without stop codons, indicating that all sequences represented functional protein coding genes, not pseudogenes. The analyzed region starts at the position 45 and stops at position 693 of the COI gene of the mitochondrial genome of Drosophila melanogaster Meigen, 1830 (AJ400907) (
The Neighbour-Joining and Maximum likelihood analyses showed 105 species, from the 133 Neotropical species with available barcodes, separated into distinct clusters. Genera represented by more than one taxon formed cohesive assemblages of five clusters [Anopheles, Orthopodomyia, Psorophora, Toxorhynchites Theobald, 1901, Uranotaenia Lynch Arribalzaga, 1891] in the NJ analysis and eight [Anopheles, Coquillettidia, Culex, Orthopodomyia, Psorophora, Stegomyia Theobald, 1901, Toxorhynchites, Uranotaenia] in the ML analysis.
The NJ tree based on Tamura-Nei genetic distances (Fig.
Neighbour-Joining tree of the barcoding sequences of mosquito species listed for Neotropics, based on Tamura-Nei genetic distances. Terminal branches have been collapsed in order to save space (see original tree in Suppl. material
The likelihood score for the best ML tree was –31,760.30891. The overall topologies of the ML (Fig.
Maximum likelihood tree of the barcoding sequences of mosquito species listed for the Neotropics. Terminal branches have been collapsed in order to save space (see original tree in Suppl. material
In this study 22 species belonging to eight genera and 11 subgenera were identified by combining morphological and molecular methodologies. Although our data and sampling size were limited, the combination of methodologies provided better success in species identification. Overall, congruence between morphology and barcode grouping, based on cluster monophyly with high (more than 95%) bootstrap support, was found in 18 of the 22 morphologically defined taxa (82%). Similarity methods based on the match between the query sequence and the reference database (more than 98% identity between BOLD/BLAST) only discriminated three mosquito species of this study. Similarity methods based on the match between the query sequence and the reference databases represent the least suitable method to discriminate mosquito species in this study.
Tree-based diagnostics are graphic criterion for species recognition, which describes genetic similarity in a visually satisfying style (Goldstein and DeSalle 2010). An important advantage of using a tree-based approach is that they present a direct sense of the statistical reliability and do not retrieve a positive result if no matching diagnostics are found (unlike the best match algorithm of BLAST and BOLD which retrieves the closest match but requires the user to interpret its reliability) (Goldstein and DeSalle 2010). Here, we compared two tree-based methods (neighbour-joining and maximum likelihood), which provided similar results. The NJ tree computed was in general agreement with previously published taxonomy based approaches (
The usefulness of DNA barcodes to discriminate mosquito genera of our dataset by using a tree-based approach was poorly supported. Approximately half of currently recognized genera represented by two or more species formed stable clusters in the NJ tree. Similarly, generic monophyly was weakly improved with the ML approach. In other insect genera, the monophyletic clusters based on DNA barcodes varies greatly depending on clustering method: e.g. Lepidoptera: Ithomiinae 50 to 61% recovered generic monophyly (
In tree-based methods, the non-monophyly at the species level represents the greatest challenge for taxon sampling and threshold approach (
Culex was the most diverse and species rich genus in the sampled mosquitoes. Culex is a cosmopolitan genus and one of the largest groups of the family Culicidae (768 species divided among 26 subgenera). There are many areas of uncertainty regarding phylogenetic relationships within the genus, as well as some problems with the identification of some species. In our study, many difficulties arose attempting to identify specimens of Culex species. In many species, female identification is very problematic due to polymorphisms and ambiguities of the morphological characters (
Classification of the species within the genus Anopheles has always been challenging. The current system of subgeneric classification is based primarily on characteristics of the male genitalia (
The most important factor affecting the accuracy of species identification through public databases is the coverage and reliability of available sequences (
It is clear that reference libraries with properly identified sequences will facilitate the association of conspecific specimens and the detection of identification errors. The DNA barcodes produced in this work allowed for the identification of females and damaged specimens, which could not be done using morphological characteristics alone. This study highlights the potential of barcoding methodology to resolve taxonomic problems associated with limitations in morphological identification, but also draws attention to its limitations to discriminate species in some Neotropical mosquito genera, especially in Anopheles (Nyssorhynchus) species complexes and some Culex species. A larger taxon barcode library with correctly identified vouchers will help future studies. Despite the limitations in our survey, the DNA barcodes produced in this work are an important contribution to increase the scope of reference libraries with properly identified sequences and to assist in the identification of mosquito species from Colombia.
Many thanks to James Pecor (WRBU, USA) for confirming the identification of some specimens and for sharing useful taxonomic literature. Also, thanks to Claudia Etzbauer (ZFMK, Germany) for her support while obtaining molecular data and to two anonymous reviewers for improving our manuscript with helpful comments. Special thanks to David Fowler (UTK, USA) for logistical support during fieldwork.
Mosquito specimens collected in the present study
Data type: occurrence
Explanation note: Full collection site details, including geo-references and environmental conditions, of the mosquito taxa identified in this study.
COI mosquito sequences downloaded from NCBI and BOLD
Data type: Text
Explanation note: Sequences from NCBI and BOLD. Sequences downloaded between December 2013 and February 2014 for all identified species with a minimum length of 480 bp of COI barcoding region, no stop codons and possible alignment among the majority of the sequences. Total of 1,159 sequences. In blue: Additional species sequences from other geographic areas for those Neotropical groups without available sequences from the Neotropics.
Mosquito specimens collected in Uraba during 2009 (unpublish data)
Data type: occurrence
Explanation note: Additional sequences of mosquitoes collected in rural areas of Uraba (Antioquia, Colombia) during 2009, including sampling information.
Outgroup taxa used in the present study
Data type: Text
Explanation note: Outgroup taxa used in the present study. Sequences downloaded from BOLD between December 2013 and February 2014.
NJ tree based on 1,292 sequences of Neotropical mosquitoes
Data type: picture
Explanation note: Original Neighbour-Joining tree of the barcoding sequences of mosquito species listed for the Neotropics, based on Tamura-Nei genetic distances.
NJ tree based on 1,292 sequences of Neotropical mosquitoes
Data type: multimedia
Explanation note: Original Neighbour-Joining tree of the barcoding sequences of mosquito species listed for the Neotropics, based on Tamura-Nei genetic distances (.tre).
ML tree based on 1,292 sequences of Neotropical mosquitoes
Data type: picture
Explanation note: Original Maximum likelihood tree of the barcoding sequences of mosquito species listed for the Neotropics.
ML tree based on 1,292 sequences of Neotropical mosquitoes
Data type: multimedia
Explanation note: Original Maximum likelihood tree of the barcoding sequences of mosquito species listed for the Neotropics (.tre).