Increasing the digital repository of DNA barcoding sequences of sand flies (Psychodidae: Phlebotominae)

Sand fly identification is complex because it depends on the expertise of the taxonomist. The females show subtle morphological differences and the occurrence of the species complexes are usual in this taxon. Therefore, a fragment of the cytochrome c oxidase subunit I (COI) gene is used for taxon barcoding to resolve this kind of problem. This study incorporates barcode sequences, for the first time, for Evandromyia cortelezzii and Migonemyia migonei from Argentina. The nucleotide sequence divergences were estimated to generate a neighbour-joining (NJ) tree. The automatic barcode gap discovery (ABGD) approach was employed to find the barcode gaps and the operational taxonomic unit (OTU) delimitation. Other species of the subtribe were included. The frequency histogram of divergences showed a barcoding gap. The ABGD analysis identified 14 operational taxonomic units (OTUs) from 13 morphological species. Sequences of Ev. cortelezzii and Mg. migonei formed well supported clusters and were diagnosed as primary species. These sequences are useful tools for molecular identification of the sand flies of the New World.

nei, e.g. dissection of the last abdominal segments must be done to see the stalk-shaped spermathecal. (6) The subtle morphological differences between females or males and the occurrence of species complexes also make species identification difficult. In the Cortelezzii complex (7) females of Evandromyia cortelezzii and Ev. sallesi are difficult to differentiate because they share diagnostic features of the length of the clypeus and the spermathecal duct. The species can only be distinguished using male characters: Ev. cortelezzii has a hood-shaped paramere and a gonocoxite with a basal tuft of four long setae, and Ev. sallesi has the paramere without an apical hoodshaped structure and the gonocoxite has a basal tuft of five long setae. (6) A fragment of the cytochrome c oxidase subunit I (COI) mitochondrial gene has been widely used in molecular taxonomy to resolve these kinds of issues. (8) Even if several studies have employed COI barcode sequences to identify Phlebotominae species and revealed species complexes in America, (9,10,11) (among others) its use can be problematic in closely related species because of incomplete lineage sorting and introgression events. (12) The strategies designed to delimit species boundaries have been developed, among them is clustering, which does not require a priori definitions of species. (13) Even if more than 220 sand fly species with barcode sequences are available in digital repositories, additional sequences are needed to enhance the efficiency of molecular identification. The present study incorporates, for the first time, barcode sequences of Ev. cortelezzii and the first sequence for Mg. migonei from Argentina and validates COI as an effective marker for taxonomic identification in sandflies because it shows consistent differences between closely related species.
Sand fly adults were captured with mini CDC light traps located at three sites: the Bajo Grande sewage treatment plant (31º23′38″S-64º04′36″W), the San Mar-tín Urban Natural Reserve (31º21′44″S-64º15′47″W), and the Zoo Garden (31º25′34″S-64º10′33″W), once a week actively from 5 pm to 10 am, during March and April, 2018. The sand flies collected were killed by freezing and were stored at -20ºC in 80% ethanol until processed. Species identification was based on the structures of the head and internal and external genitalia of females and males, respectively, which were mounted on the microscope slides using the taxonomic key of Galati. (6) Male sand fly legs (between two and six, according to availability), thorax, and abdomen were separated before morphological preparations for subsequent molecular procedures. Voucher specimens were deposited in the collection of the Centro de Investigaciones Entomológicas de Córdoba, Argentina. The abbreviation of the generic names were according to Marcondes. (14) DNA extractions were obtained from five male specimens of Ev. cortelezzii and one of Mg. migonei. The genomic DNA was extracted using the rapid method described by Pinto et al. (10) The DNA obtained was used as the template to amplify the ~658 bp fragment of the COI barcode region and the amplification was performed as described in Folmer et al. (15) The polymerase chain reaction (PCR) products were electrophoresed in 1% TAE agarose gel stained with GelRed (Biotium Inc, Hayward, USA). All sequencing reactions were carried out in both directions using an ABI3730XL automatic sequencer (Macrogen Inc., Korea) with the same set of PCR primers. The sequences were edited using BioEdit v. 7.2 (16) The primer regions were removed from the sequences. Comparisons with available sequences were performed using Basic Local Alignment Search Tool (BLAST) to check for sequence homology and species identification of Mg. migonei. The COI gene sequences were aligned by nucleotides using the Muscle algorithm. (17) The barcode sequences obtained in this study are deposited in GenBank under accession numbers: MN432906 (Cba1828M); MN432907 (Cba1825M); MN432908 (Cba1829F); MN432909 (Cba1837M); MN432910 (Cba1839M); MN432911 (Cba1856M). Pairwise nucleotide sequence divergences and the mean intra-and interspecific distances were estimated using Kimura two-parameter (K2P) distance (18) implemented in Mega v7. (19) Subsequently, the divergences obtained were used to generate a neighbour-joining (NJ) tree, using Mega v7 to evaluate the clustering pattern among species. The statistical support for the clusters was estimated using the Bootstrap support value (BSV) obtained from 1,000 replicates. Two mosquito sequences of Culex bidens (Diptera: Culicidae) were used as outgroup (GenBank numbers: KY581209 and KY581211). The automatic barcode gap discovery (ABGD) approach (13) was employed to find barcode gaps for the primary species or OTUs delimitation, and to corroborate results obtained in the NJ topology. In this study, the default range of intraspecific divergence was limited between 0.001 and 0.1. Sequences from GenBank were obtained as references in the species delimitation analysis: Ev. evandroi (2), Ev. termitophila (4), Expapillata firmatoi (2), Lutzomyia longipalpis (6), Mg. migonei (5), Pintomyia bianchigalatiae (2), Pi. fischeri (4), Pi. misionensis (2), Pi. monticola (4), and Sciopemyia sordellii (2) from Brazil; Colombia, and Perú. These species were selected because they were recorded from Argentina and belong to the subtribe Lutzomyiina, as Ev. cortelezzii and Mg. migonei. We attempted to assess the degree of divergence of the first barcode sequences for Ev. cortelezzii and Mg. migonei for the science and Argentina, respectively.
Sixty-two specimens were captured and were morphologically identified as follows (Table): Ev. cortelezzii s.s. (n = 12), Cortelezzii complex (n = 49), and Mg. migonei (n = 1). Five males of Ev. cortelezzii and one of Mg. migonei were sequenced. The mean K2P sequence distance within the 13 nominal species was 1.03%. The frequency histogram of mean COI K2P sequence intra-and interspecific divergences may be seen in Fig. 1,  showing that a clear barcoding gap was established. The NJ tree was built using the K2P distances between the specimens, and the primary species delimitation from the strict partition obtained from the ABGD analysis, as shown in Fig. 2. The ABGD analysis provided ten possible partitions for these specimens (Fig. 3). The most conservative partitioning (strict partitioning) identified 14 OTUs (Fig. 2) with values of prior intraspecific divergence between 0.17% and 3.5% because values below 0.17% or above 3.5% can over-or underestimate the number of species, respectively. The 14 OTUs mentioned were identified from 13 morphological species because one of the original groups, Pi. fischeri, was split into two barcode groups (Fig. 2), showing a deep intraspecific variation. The clusters enclosing Pi. fischeri, provisional species (PS)1 and PS2, showed group sequences from Brazil and Perú, separately. A plausible explanation is related to the historical events of vicariance, which can reduce gene flow between subpopulations of the same species, leading to speciation. (20) Over time, morphological differentiation tends to take longer because changes in the morphological traits require changes in the multiple genes. (21) The sequences of Lutzomyia longipalpis from Brazil and Colombia were considered a priori as the two provisional species because the species in Brazil showed a deep intraspecific variation between the regions of the country, as evidenced by Pinto et al. (10) The previously mentioned concept has been supported by the outcomes. The sequences of Pi. monticola were also treated as the two provisional species (PS1 and PS2) because they belong to a cryptic species complex. (10) The sequences of the nominal species Mg. migonei formed a well-supported cluster (100% BSV), with a mean intraspecific K2P distance of 2.24% and was recognised as an OTU by the ABGD partitioning (Fig. 2). It is important to note that Mg. migonei sequences belonging to Argentina differ from those found in Brazil and Colombia (Fig. 2). Novel sequences of Ev. cortelezzii were grouped in a well-supported cluster (100% BSV) and diagnosed as the primary species by the ABGD algorithm (Fig. 2).
Even though all the Ev. cortelezzii specimens that were sequenced had been collected in Córdoba Province in Argentina, it is worth noting that they were not clustered with other congeneric species that were also considered, as shown by Pinto et al. (10) In the latter, sequences of Ev. carmelioni were grouped with Ev. lenti, from specimens collected in the distant states of Espírito Santo and Mato Grosso in Brazil and were not recognised as species by the ABGD partitioning. In our analyses, three species of the genus Evandromyia from Argentina and Brazil were diagnosed as different OTUs.
Mg. migonei and Ev. cortelezzii are widely distributed in Argentina, (3) both species being important in public health. Mg migonei is associated with the agents of visceral Leishmaniasis (VL) and TL, Leishmania braziliensis, and L. infantum, respectively, (22,23) and was considered the putative vector of VL in Santiago del Estero Province. (22) Female specimens of the Cortelezzii complex were found naturally infected with L. braziliensis in Chaco Province. (24) The first sequences of COI barcode region are now available for sand flies for Argentina. The sequence of Mg. migonei in its southernmost known distribution and the sequences of Ev. cortelezzii are the first for the species and allow the molecular identification of the females of the complex. Further studies should be carried out to obtain COI sequences from the males of Ev. sallesi and to solve identification problems of females of the complex. These sequences are useful tools for the future molecular identification of these species in other locations in the New World.