High Levels of Diversity in Anopheles Subgenus Kerteszia Revealed by Species Delimitation Analyses

The Anopheles subgenus Kerteszia is a poorly understood group of mosquitoes that includes several species of medical importance. Although there are currently twelve recognized species in the subgenus, previous studies have shown that this is likely to be an underestimate of species diversity. Here, we undertake a baseline study of species delimitation using the barcode region of the mtDNA COI gene to explore species diversity among a geographically and taxonomically diverse range of Kerteszia specimens. Beginning with 10 of 12 morphologically identified Kerteszia species spanning eight countries, species delimitation analyses indicated a high degree of cryptic diversity. Overall, our analyses found support for at least 28 species clusters within the subgenus Kerteszia. The most diverse taxon was Anopheles neivai, a known malaria vector, with eight species clusters. Five other species taxa showed strong signatures of species complex structure, among them Anopheles bellator, which is also considered a malaria vector. There was some evidence for species structure within An. homunculus, although the results were equivocal across delimitation analyses. The current study, therefore, suggests that species diversity within the subgenus Kerteszia has been grossly underestimated. Further work will be required to build on this molecular characterization of species diversity and will rely on genomic level approaches and additional morphological data to test these species hypotheses.


Introduction
Malaria remains one of the most important infectious human diseases globally. In 2020 alone, there were an estimated 241 million malaria cases across 85 countries [1]. Although A total of 272 specimens, representing 10 of the 12 formally recognized species of Anopheles Kerteszia, collected over 30 years from across Central and South America, were included in the current analyses. Newly generated mtDNA COI barcode sequences and associated data, along with publicly available sequences, are available on the Barcode of Life Database (BOLD) under Dataset: DS-KERT. This dataset comprised An. bambusicolus (n = 3), An. bellator (n = 14), An. boliviensis (n = 10), An. cruzii (n = 62), An. homunculus (n = 17), An. laneanus (n = 8), An. lepidotus (n = 7), An. neivai (n = 107), An. pholidotus (n = 10), and An. rollai (n = 23) from Brazil, Colombia, Ecuador, French Guiana, Trinidad and Tobago, Panama, and Venezuela. In addition, an unidentified Kerteszia specimen (NAMRU6_2014_342) from Peru and ten further "Culicidae spp." sequences closely related to Kerteszia using NCBI Blast alignment, were also included. Verified specimens of An. auyantepuiensis and An. gonzalezrinconesi were unavailable for analyses. All specimens were morphologically identified using available morphological keys and original species descriptions [25][26][27]. Morphological voucher specimens were retained where available, and included point-mounted adults of both sexes, slide-mounted associates of male genitalia, and immature stages (larval exuviae, Le; pupal exuviae, Pe). Extracted DNA was retained in frozen repositories of the National Museum of Natural History, Smithsonian Institution, USA (USNM), and the Natural History Museum, England (NHMUK).

DNA Barcodes
Specimens for this study were accrued by members of the Mosquito Barcoding Initiative (MBI). Legs and/or abdomens from 90 vouchered morphologically identified specimens were used for DNA extraction following the phenol-chloroform method, using the AutoGen or QIAgen BioSprint automated extraction platform. Amplification of the COI barcode region (658-bp) was performed using the Folmer LCO1490 and HCO2198 primers [28] and the PCR conditions described in Ruiz et al. [29]. Sequencing was carried out in both directions using the Big Dye ® Terminator Kit on an ABI 3730 automated sequencer (PE Applied Biosystems). Sequences were edited in Sequencher TM v. 4.8 (Genes Codes Corporation, Ann Arbor, MI, USA). An additional 182 mtDNA COI sequences, publicly available in the GenBank (www.ncbi.nlm.nih.gov/genbank; accessed on 10 October 2022) and BOLD (www.boldsystems.org; accessed on 10 October 2022) databases, were included in the study (Supplementary Table S1). The full sequence dataset was then aligned, first by nucleotides using the Muscle algorithm [30] implemented in SeaView [31], and then by amino acid using TranslatorX [32].

Exploratory Data Analysis
The NeighborNet algorithm implemented in SplitsTree [33] was used to construct a network to explore phenetic relationships and assess whether the data nature was tree-like, with a bifurcating phylogeny, or network-like, with reticulation and conflicting phylogenetic signals.

Phylogeny
A 70% majority-rule consensus tree was constructed by maximum likelihood using 1000 replicates from the ultrafast bootstrap approximation approach (UFBoot) implemented in IQ-TREE 2 [34]. The UFBoot support values are more unbiased and support for a clade should be considered at ≥95%. Data were partitioned by codon position and optimal model selection was performed using the -m MFP option among a range of models (-mset mrbayes). The selection of an appropriate outgroup followed the recommendations of Grant [35], by including the closest known sister taxa and successively expanding the outgroup sample until ingroup topology was shown to be stable in at least two iterations. Outgroup sampling included specimens from the following subgenera: Nyssorhynchus (An. albitarsis, MF381591; An. oryzalimnetes, HQ335345; An. darlingi, GQ918272; An. evansae, MF381711; An. nuneztovari, MF381680; An. strodei, NC037808; An. braziliensis, NC037791); Anopheles (An. minor, MF381684); and Stethomyia (An. kompi, NC037827; An. nimbus, NC037811).

Species Delimitation
The first method used for the species delimitation analyses was the Assemble Species by Automatic Partitioning (ASAP [23]), which uses pairwise genetic distances and ascending hierarchical clustering to build a list of best partitions. Partitions are ranked by score, which is a combination of two metrics: the probability of panmixia and the barcode gap width. The method does not require any a priori knowledge of species number/composition, or biological information, such as a phylogeny or intraspecific distances. We also used the Barcode of Life Database (BOLD: www.boldsystems.org; accessed on 10 October 2022) Barcode Index Numbers (BINs). BOLD BINs are produced from the Refined Single Linkage (RESL) analysis of the entire BOLD database, which clusters sequences into operational taxonomic units (OTUs) using a graph analytical approach [24]. However, not all specimens/clusters had been assigned BOLD BIN numbers at the time of manuscript submission (11 from 82 specimens of An. neivai s.s.), due to delays associated with the automated BIN analysis running on BOLD. The Multi-rate Poisson tree processes (mPTP [22]) approach was also used. This approach is a phylogeny-aware delimitation method that uses differences in mutation rate in a phylogenetic tree to distinguish interspecific and intraspecific diversity. Markov Chain Monte Carlo (MCMC) was used to assess confidence, using 4 independent runs.

Phylogenetic Network and Tree
The phylogenetic network demonstrates the tree-like versus network-like nature of the data, as well as the clustering of sequences into tentative species ( Figure 1). As can be seen, there is a strong network-like structure among the deeper phylogenetic relationships, while the relationships among the tips and at the level of species tend to be bifurcating and more tree-like. These relationships are also consistent with the phylogenetic tree ( Figure 2); all species and species complex clades are supported by the ultrafast bootstrap approximation at values ≥95%. However, relationships between the members of complexes and those found at more basal positions are all poorly supported or resolved.
Linkage (RESL) analysis of the entire BOLD database, which clusters sequences into operational taxonomic units (OTUs) using a graph analytical approach [24]. However, not all specimens/clusters had been assigned BOLD BIN numbers at the time of manuscript submission (11 from 82 specimens of An. neivai s.s.), due to delays associated with the automated BIN analysis running on BOLD. The Multi-rate Poisson tree processes (mPTP [22]) approach was also used. This approach is a phylogeny-aware delimitation method that uses differences in mutation rate in a phylogenetic tree to distinguish interspecific and intraspecific diversity. Markov Chain Monte Carlo (MCMC) was used to assess confidence, using 4 independent runs.

Phylogenetic Network and Tree
The phylogenetic network demonstrates the tree-like versus network-like nature of the data, as well as the clustering of sequences into tentative species ( Figure 1). As can be seen, there is a strong network-like structure among the deeper phylogenetic relationships, while the relationships among the tips and at the level of species tend to be bifurcating and more tree-like. These relationships are also consistent with the phylogenetic tree (Figure 2); all species and species complex clades are supported by the ultrafast bootstrap approximation at values ≥95%. However, relationships between the members of complexes and those found at more basal positions are all poorly supported or resolved.

Species Delimitation
The species delimitation using the distance-based ASAP approach yielded 28 putative species (vertical bars in Figure 2). BOLD BINs were associated with all clusters delimited by ASAP. In some cases, multiple BINs were detected per ASAP cluster (BOLD BINs associated with taxon names in Figure 2). The mPTP delimitation was not included because the phylogenetic tree, upon which mPTP relies, was poorly supported, and the SplitsTree network suggested that the data were not strongly tree-like.
Species such as An. lepidotus (BOLD:AAJ2789), An. bambusicolus (BOLD:AAF0600), and An. cruzii (BOLD:AAG3843) are well-resolved taxa in the phylogenetic tree (≥98% bs support) and are separately clustered in the species delimitation analyses. However, all other taxa in the genus show at least some evidence of species complex structure.

Species Delimitation
The species delimitation using the distance-based ASAP approach yielded 28 putative species (vertical bars in Figure 2). BOLD BINs were associated with all clusters delimited by ASAP. In some cases, multiple BINs were detected per ASAP cluster (BOLD BINs associated with taxon names in Figure 2). The mPTP delimitation was not included because the phylogenetic tree, upon which mPTP relies, was poorly supported, and the SplitsTree network suggested that the data were not strongly tree-like.
Species such as An. lepidotus (BOLD:AAJ2789), An. bambusicolus (BOLD:AAF0600), and An. cruzii (BOLD:AAG3843) are well-resolved taxa in the phylogenetic tree (≥98% bs support) and are separately clustered in the species delimitation analyses. However, all other taxa in the genus show at least some evidence of species complex structure.
Anopheles neivai is a highly diverse species complex and is delimited into eight specific clusters using ASAP analyses. Clusters are denoted An. neivai s.s. and An. neivai 2 through to An. neivai 8, with minimum inter-cluster distances ranging from 3.12% (between An. neivai 2 and An. neivai 3) to 10.09% (between An. neivai 3 and An. neivai 6), suggesting considerable inter-specific differences. In addition, An. neivai 6 has two BINs assigned to it (BOLD:AEL2207 and BOLD:AEH3455), although these are each represented by a single sequence and separated by a distance of just 1.10%. These two BINS were, thus, considered representative of a single taxon (An. neivai 6) in the current study. There is also some support for a complex structure in An. homunculus. Although the ASAP analysis identifies it as a single cluster, the BOLD BIN analysis splits it into two clusters (BOLD:ACB9054 and BOLD:AAF0613) separated by a distance of 2.71% ( Figure 2; Table 1). Anopheles bellator is also split into two clusters (topotypic An. bellator s.s. from Trinidad and Tobago, BOLD:AAJ2798; An. bellator 2 from São Paulo, Brazil, BOLD:AAF0614) and separated by an inter-cluster distance of at least 4.11%. Anopheles laneanus s.s. (BOLD:AAN3565) is fully resolved across all delimitations. A previously published sequence in GenBank listed as An. cruzii (GenBank: KU551284) is herein denoted as An. laneanus 2 (BOLD:ADV0367), with which it forms a sister relationship. These are separated by a minimum inter-cluster distance of 3.44%.
The remaining An. cruzii specimens are found as a single cluster (BOLD:AAG3843), although variation within this cluster is high, with a maximum intra-cluster distance of 3.22%. A single specimen (GenBank: KU551285) from the state of the type locality (Rio de Janeiro, Brazil) is found as a sister to the main An. cruzii clade in the phylogenetic tree and is separated by a distance of 2.12%.

Discussion
Our combination of phylogenetics and species delimitation analyses of Kerteszia mosquitoes reveals high species diversity within this subgenus (Table 1; Figures 1-3).
Laneanus complex and Anopheles cruzii Previous studies of An. cruzii have found that this taxon exists as a species complex. Specimens collected in Brazil from Itatiaia in the state of Rio de Janeiro, and Florianópolis in the state of Santa Catarina, were found to be significantly differentiated at nuclear gene fragments [19]. In addition, chromosomal polymorphism has also been reported, with three forms found among populations from the Brazilian state of São Paulo (Ramirez & Dessen, 2000). More recently, a phylogenetic analysis of the COI barcode suggested the existence of three species in the An. cruzii complex [17], although genetic distances were not given. Herein, we find cryptic species boundaries between An. cruzii and An. laneanus. One specimen initially identified as An. cruzii (GenBank: KU551284), from the municipality of São Paulo, is clearly resolved from the main An. cruzii group, clusters with An. laneanus s.s. in both the tree and network and has inter-cluster distances (>3.44%). This level of variation is suggestive of specific differences given that COI variation within species of mosquito [36][37][38] and other insects [24,39,40] does not generally exceed 2%. The same sister relationship was found for this specimen and An. laneanus s.s. in a previous phylogenetic study using combined COI-ND4 genes, although not with the COI barcode alone [17]. Considering the relationship of this specimen with An. laneanus s.s. (which includes specimens from the An. laneanus type locality of Campos do Jordão), we have denoted it An. laneanus 2, pending further investigation. There is little support for additional species delimitation among the remaining An. cruzii specimens, although the maximum intra-cluster distance was higher than expected at 3.22%. The genetic diversity in An. cruzii sensu lato, its close association with the laneanus complex, and its identity as a potential vector of human pathogens, such as Plasmodium [15] and Boraceia virus [41,42], raises important questions about effective approaches to vector incrimination and control in Brazil's Atlantic Forest.
Neivai complex Specimens identified as An. neivai showed the highest level of genetic diversity in the current study, with pairwise differences of 14.6% among the most divergent specimens. Species delimitation recovered at least eight tentative species. A basal split within the An. neivai clade appears to separate a northern Andean/central American clade from a Guiana Shield clade. Within the former clade, five clusters are identified, two of which (An. neivai s.s. and An. neivai 3) were found across considerable geographic distances. Anopheles neivai s.s. was identified based on its geographical proximately to the type locality of An. neivai (~5 km to Portobelo, Panama). The An. neivai s.s. cluster retains considerable genetic diversity (maximum intra-cluster distance = 3.15%), mainly due to genetic distances between Panama and Colombia specimens, but these distances were not sufficient to allow for delimitation in our analyses (minimum inter-group distance < 1.20%). This pattern of diversity within An. neivai s.s. may represent population differentiation between Panama and Colombia specimens. Within the latter Guiana Shield clade, at least three clusters were identified. These clusters were found in collections made by Silva-do-Nascimento et al. [43] in northwest Roraima State, one of which (An. neivai 5) was also collected in the Sinnamary Commune of French Guiana by Talaga et al. [44], demonstrating a geographical distribution that extends for more than 1000 km through much of the Guiana Shield. Anopheles neivai is a confirmed vector of important human pathogens. In the Pacific lowlands of Colombia, it has been found to be naturally infected with Plasmodium falciparum, P. vivax [45], and Guaroa virus [46], which causes febrile illness in tropical regions of Central and South America [47,48]. It was also found to be infected with yellow fever virus in Panama [49].
The identification of considerable diversity within An. neivai s.l. has important implications for vector incrimination in northern South America. An. neivai and An. neivai 8 (the latter referred to as "An. neivai nr. neivai 4" in Ahumada et al. [20]) were both found in the department of Valle del Cauca, Colombia, where populations of An. neivai s.l. were reportedly infected with Guaroa virus [46]. In addition, specimens of An. neivai s.s., collected by López-Rubio et al. [50], were found close (within approximately 40 km) to where historic examples of yellow fever virus-infected An. neivai s.l. occurred [49]. Efforts in vector incrimination in the region will, therefore, need to consider the cryptic diversity within An. neivai s.l., and further investigate potential differences in epidemiologically and ecologically important traits.
Bellator complex Anopheles bellator specimens were molecularly partitioned into two tentative species: topotypic An. bellator s.s. from Trinidad and Tobago, and An. bellator 2, which were identified in geographically disparate collections in the Atlantic Forest region of São Paulo, Brazil, and the Amazon region of Putumayo, Colombia, respectively, and separated by distances of approximately 4000 km. Anopheles bellator was previously considered an important vector of Plasmodium in Trinidad and Tobago, being highly anthropophilic and experimentally susceptible to infection, and it was primarily responsible for the transmission of what became known as "bromeliad malaria" on the islands [9]. In the Atlantic Forest, An. bellator 2 (as An. bellator) reportedly displays considerable endophilic and endophagic behavior [51], although data on Plasmodium infection are scant [52].
Homunculus complex Although specimens identified as An. homunculus were not partitioned with ASAP, they were partitioned with the BOLD BIN analyses. However, the maximum intra-specific distance of the ASAP single cluster is 4.1%, and the minimum inter-specific distance of the two clustered proposed by BINs is 2.7% and is suggestive of a species complex structure. In the case of the two clusters of the BIN partition, one is identified from Trinidad and Tobago and the other from Brazil's Atlantic Forest. Additionally, it appears that the BIN cluster found in Trinidad and Tobago may also be present on the continental mainland. Although not included in the species delimitation analyses, two short COI sequences (307 bp) from two Venezuelan individuals were uniquely invariable with sequences from Trinidad and Tobago (GenBank: OQ272333 and OQ272334). Further sampling and genetic analysis of An. homunculus specimens, particularly from its type locality of Restrepo, Department of Meta, Colombia, and elsewhere in northern South America, are necessary to clarify the identity of An. homunculus s.s. and improve our understanding of geographic and evolutionary relationships among island and continental populations.
Pholidotus complex A simple pairwise analysis of all An. pholidotus specimens found the maximum intra-taxon distance to be considerable (>7%). Our analyses determined that An. pholidotus was likely a species complex and split this taxon into four clusters. The first was distributed across Colombia, Venezuela, and Peru. The remaining three clusters were each represented by a single specimen; two were collected in the Department of Antioquia in northern Colombia [53], while the third was collected in the Department of Cajamarca in northwestern Peru. Two "Culicidae sp" sequences (GenBank: KY11725 and KY117252; Rosero-García et al. [53]) were also included in this study due to their close identity to Kerteszia species following a Blast analysis. Our study allows us to tentatively associate them with the Pholidotus complex. However, without specimens collected from the type locality in northern Panama, we were unable to associate a cluster with the type species.
Anopheles lepidotus Although the morphological similarity of An. lepidotus with An. pholidotus has mistakenly implicated the former as a malaria vector in the Department of Tolima, Colombia [26,54], the current study shows that An. lepidotus is well resolved as a single cluster in both in tree-and distance-based delimitations and found from Roraima in Brazil to the Province of Orellana in Ecuador, covering almost 1500 km of the Amazon Basin. This cluster may correspond to the type specimen from the Department of Meta, Colombia [25], but molecular data from topotypic An. lepidotus are needed to confirm this. The absence of available specimens from the type locality of An. pholidotus (Bocas del Toro province, Panama) means that we cannot confidently ascribe any cluster as An. pholidotus s.s. The potential existence of sibling species in Colombia further complicates our understanding of the competency of these different entities to transmit malaria or other pathogens in the region.
Anopheles bambusicolus Unique among Kerteszia species, Anopheles bambusicolus is the only species whose larval habitat is found not in bromeliads, but generally in unbroken bamboo internodes. The species' southern limits are along the Atlantic Forest of southeastern Brazil, while its northern limits are found in the Department of Meta, Colombia, where its type locality exists (La Unión, Department of Meta, Colombia). The specimens included in the current study come from the Province of Orellana, Ecuador, more than 300 km from the type locality, and comprise just a single COI haplotype. Further sampling of this species is, thus, required to properly characterize its intra-specific diversity and determine whether this cluster is representative of the type specimen.
Boliviensis complex As can be seen from the polytomy present in the phylogenetic tree, evolutionary relationships among An. boliviensis and An. rollai are poorly resolved. However, there is strong evidence for both existing as species complexes. Anopheles boliviensis 1 was found in Columbia, while An. boliviensis 2 and An. boliviensis 3 were both collected in the western foothills of the Peruvian Andes, approximately 350 km apart. Anopheles boliviensis 1 and An. boliviensis 2 form a sister relationship, separated by at least 4.3%, while their evolutionary relationship with An. boliviensis 3 is unresolved, and separated by at least 4.6%. We are unable to identify which cluster represents the type species, as our most geographically proximate specimen (An. boliviensis 3) was collected more than 250 km from the type locality in Songo, Bolivia. Anopheles boliviensis is not considered an important malaria vector, while the importance of orthobunyaviruses Anopheles A and Anopheles B previously found infecting the species remains largely unknown [55]. However, the existence of multiple species within a Boliviensis complex may yet reveal diverse medical importance across its range.
Rollai Complex Two clusters, An. rollai 4 and An. rollai 5, originating from Venezuela (states of Mérida and Táchira) and Peru (Cajamarca Province), respectively, are consistently resolved. The remaining An. rollai specimens are resolved as a single cluster in the ASAP analysis, but split into three clusters in the BOLD BIN analysis, and denoted herein An. rollai 1, An. rollai 2, and An. rollai 3 (BOLD:ABZ5384, BOLD:AAI4555, and BOLD:ADK7866, respectively). Both An. rollai 1 and An. rollai 2, like An. rollai 4, are found in the state of Mérida, Venezuela, while An. rollai 3 is found in the department of Antioquia, Colombia. The specimens of An. rollai 3 (GenBank: KY117248-KY117250), collected by Rosero-García et al. [53] in Colombia and denoted Culicidae sp. therein, and of An. rollai 5, collected by the U.S. Naval Medical Research Unit (NAMRU-6) in Cajamarca state, northern Peru, appear to be new country records for the rollai complex. However, the geographic proximities of An. rollai 1, An. rollai 2, and An. rollai 4 specimens to the An. rollai type localities (in the states of Barinas and Táchira, Venezuela) make it difficult to assign clusters to the type species. Biological processes, such as incomplete lineage sorting and introgression, can confound attempts to reconstruct the evolutionary history of closely related species. Given the close evolutionary and geographic relationships among clusters within the Rollai complex, robust phylogenetic inferences will require accounting for such processes using approaches such as the multispecies coalescent model on multi-locus data. Further molecular and morphological analyses of collections at these type localities, as well as from across the Rollai complexes range, will, therefore, be needed to characterize the evolutionary history of the complex and determine which clusters represent the type species. locality of Campos do Jordão), we have denoted it An. laneanus 2, pending further investigation. There is little support for additional species delimitation among the remaining An. cruzii specimens, although the maximum intra-cluster distance was higher than expected at 3.22%. The genetic diversity in An. cruzii sensu lato, its close association with the laneanus complex, and its identity as a potential vector of human pathogens, such as Plasmodium [15] and Boraceia virus [41,42], raises important questions about effective approaches to vector incrimination and control in Brazil's Atlantic Forest.  Finally, five specimens listed only as Culicidae spp. (GenBank: KY117253-KY117257) collected in Colombia by Rosero-García et al. [53] were included in our study, due to the Blast proximity to Kerteszia species. These were shown to comprise two haplotypes recovered in two distinct clusters. Their relationship with other members of Kerteszia is difficult to determine due to poor support present in the phylogenetic tree; they are separated from their nearest neighbors by a distance of 4.3% and are herein denoted as Kerteszia sp.
Missing taxa Unfortunately, we did not have access to specimens of An. gonzalezrinconesi or An. auyantepuiensis, both of which are endemic to Venezuela. The inclusion of these two species will be important in future studies to establish their relationships among the diversity described here and to characterize the deeper evolutionary relationships among species in Kerteszia. Of course, it remains a possibility that some of the unnamed clusters determined here may indeed correspond to either An. gonzalezrinconesi or An. auyantepuiensis.
The findings of our study show much higher levels of diversity within the subgenus Kerteszia than previously reported. Among the ten currently recognized species that were included in our study, our analyses found at least twenty-eight tentative species present, and some evidence of further cryptic species structure. Previously collected specimens that could only be identified to the level of family are herein denoted members of the subgenus. Despite being able to reveal the existence of several species complexes, COI barcode data were unable to resolve deeper evolutionary relationships among species in the subgenus Kerteszia. Similarly poor support among these relationships has also been found in the phylogenetic analysis of Kerteszia mitogenomes [56].
Future work on describing the species diversity and phylogeny of the subgenus will require the analyses of a range of more informative loci, potentially via whole genome sequencing, and using topotypic material where possible. This can allow for assessments of species diversity while accounting for the confounding influences of incomplete lineage sorting and introgression [57][58][59], as well as provide a means to better describe basal nodes and well-supported phylogenetic relationships among members of the subgenus. Furthermore, several of the clusters found, which are tentatively ascribed species, are found as singletons; clearly, further sampling of these clusters is also required to better characterize species diversity within Kerteszia.

Conclusions
DNA barcoding has proven useful in exposing cryptic biodiversity in mosquitoes. Here, we assessed samples belonging to 10 of the 12 described species in the Anopheles subgenus Kerteszia. Species delimitation using the distance-based ASAP approach yielded a staggering 28 putative species-almost trebling the currently described taxa in the subgenus. Our findings highlight the importance of using topotypic material to clearly establish the identity of the nominotypical members of complexes, in order to allow formal descriptions of novel species and advocate the use of DNA barcoding approaches in baseline assessments of species diversity, as well as to shape more comprehensive studies of the evolutionary biology of medically important species.