Background
Anopheles nuneztovari sensu lato was originally described in San Carlos, State of Cojedes, western Venezuela. It is geographically distributed from eastern Panama to northern South America [
1] and is considered one of the four most important malaria vectors in northern South America, together with
A. darlingi,
A. albimanus and
A. aquasalis [
2,
3].
Anopheles nuneztovari
s.l. has been long recognized to be an important human malaria vector in Colombia and Venezuela, presenting endo and exophagic behaviours, besides high levels of anthropophily and infection rate [
4,
5]. Differently from Colombia and Venezuela, the Brazilian populations of this species, which are predominantly zoophagic, were not considered a malaria vector in past decades (1940s–1970s). However, with the development of more sensitive techniques for detecting malaria parasites,
A. nuneztovari
s.l. has been reported to be infected with
Plasmodium species in five states of the Brazilian Amazon region [
6‐
12], and was recently considered an important local vector in the State of Amapá, Brazil [
12]. Supporting these findings, experimental infection studies conducted with
A. nuneztovari s.l. from Manaus (MN), Brazil, reported a high infection rate for
Plasmodium vivax [
13]. In fact, the “populations” of the Brazilian Amazon region feed preferentially on bovines rather than humans, and this behaviour is probably the limiting factor to transmitting human malaria [
12].
In view of its importance as malaria vector in northern South America, a great number of studies was conducted with
A. nuneztovari
s.l. from Colombia/Venezuela (malaria vector) and the Amazon Basin (non-malaria vector) [
14‐
24], aiming to understand the distinct patterns of malaria transmission across its geographic range. The results indicated that
A. nuneztovari
s.l. could encompass two ecologically and genetically distinctive geographic populations, but no strong evidence of one cryptic species complex—as previously thought—was found, reflecting a very recent evolutionary history [
14‐
24].
Based on phylogenetic analyses of three molecular markers and the comparison of the male genitalia aedeagus apex, Calado et al. [
25] demonstrated that the
A. nuneztovari from Colombia and Venezuela are likely to be distinct from the Brazilian Amazon specimens studied by the authors [
25]. Considering these results,
A. goeldii was revalidated from synonymy with
A. nuneztovari [
25]. Scarpassa [
26] observed differences in the length of the pre-humeral dark spot (PHD) and the length of the subcostal pale spot (SCP) on the wings of the adult females from Buenaventura and Tibú (Colombia) compared with the specimens of the Brazilian Amazon, and—based on a taxonomic key—they were identified as
A. rangeli. However, the 4th instar larvae, male genitalia and eggs were identified as
A. nuneztovari. Sant’Ana et al. [
27] reported similar findings in the revision of
A. goeldii. Recently,
A. dunhami was also included in
A. nuneztovari
s.l., based on phylogenetic analyses performed with three markers [
28]. Thus, the taxonomic status of
A. nuneztovari
s.l. now includes:
A. nuneztovari
s.s. which occurs in Colombia and western Venezuela,
A. goeldii in the Brazilian Amazon region, and
A. dunhami, found in the Brazilian Amazon [
25,
29‐
31] and Colombia [
32]. In the Brazilian Amazon, however,
A. dunhami shows overlap with
A. goeldii in a large geographic area [
31]. The role of
A. nuneztovari
s.s. as malaria vector has been elucidated, but the role of each lineage and/or
A. goeldii species within the Brazilian Amazon still remains to be clarified, despite the reports of infection by
Plasmodium spp. [
6‐
12].
Anopheles dunhami is abundant in forests and has zoophilic behaviour [
33], and it was never found infected with the malaria parasite, or was wrongly identified as
A. goeldii or other
A. nuneztovari s.l. lineages.
Along with the presence of three cryptic species in
A.
nuneztovari
s.l., two genetic lineages were reported in the Brazilian Amazon, based on ITS2 sequences [
17] and mtDNA-RFLP [
18,
23]. More recently, Mirabello and Conn [
24] detected five lineages with the
white gene, three of which (1, 4 and 5) in the Brazilian Amazon + Suriname, and two (2 and 3) in Colombia/Venezuela. The two lineages observed in Colombia and Venezuela could correspond to
A. nuneztovari
s.s., whereas the three of the Brazilian Amazon + Suriname could correspond to
A. goeldii [
24]. Scarpassa and Conn [
31] proposed the existence of four lineages across its geographic range, based on the fragment at the 3′ end of the
COI gene. Specimens from Bolivia/Colombia/Venezuela grouped to a single cluster (subclade II-C), and may represent
A. nuneztovari s.s. The specimens from the Brazilian Amazon + Suriname grouped to three clusters (Clade I and subclades II-A and II-B) and may represent
A. goeldii and other species within
A. nuneztovari s.l. The inconsistence in the number of lineages among studies [
17,
18,
23,
24,
31] could be related especially to sampling strategies within the Brazilian Amazon region.
Given the evidence above, there is no doubt that
A. nuneztovari consists of a cryptic species complex in northern South America, and that the Brazilian populations comprise two or more genetic lineages or species. Therefore, analyses with multi-markers will be needed to test the hypothesis of multiple species in the Brazilian Amazon region. The definition of these lineages could help understanding how they contribute to the malaria transmission in this region, especially because they may show differences in ecology, behavior and
Plasmodium susceptibility, with consequent implications on management, surveillance and control measures. The aim of this study was to investigate whether the
A. nuneztovari
s.l. lineages detected [
17,
18,
23,
24,
31] should actually have species status, based on the analyses of specimens from five localities of the Brazilian Amazon with the mitochondrial (barcode region) and nuclear (12 microsatellites loci) markers. Two of the five localities [Autazes (AU), Abacate da Pedreira (AP)] were sampled for the first time in this study, whereas the other localities [MN, Careiro Castanho (CS), Tucuruí (TU)] were previously studied by Scarpassa and Conn [
31], using a fragment of the 3′ end of the
COI gene. This fragment has demonstrated to be highly variable in anopheline species, being therefore amply used in the population genetics and phylogeographic studies of this group [
34]. In contrast, the barcode region (Folmer region) that consists of a 648 bp fragment at the 5′ end of the
COI gene has emerged as the standard barcode region, because it presents a low rate of intra-specific and high inter-specific variation (barcoding gap), thus permitting the characterization of each taxonomic group or unit [
35]. This region has shown to be a valuable tool in the identification of species complexes in anophelines [
36‐
38]. The microsatellite markers, developed and characterized previously for
A. nuneztovari
s.l. [
39,
40], were analysed for the first time in the present study. These markers are appropriate to estimate intra-population genetic diversity, fine-scale population structure, and also for taxonomic and evolutionary genetic studies of very recently evolved species. In the present study, the samples analysed were named
A. nuneztovari
s.l.
Discussion
Approximately half of the anopheline malaria vector species belong to sibling/cryptic species complexes [
65]. In general, cryptic species [
66] are of recent evolutionary origin and may, therefore, be morphologically very similar or identical, making their separation difficult. However, cryptic species can differ genetically, ecologically, behaviourally and epidemiologically. Over time, distinct morphological characters can also evolve; however, morphological differentiation tends to take longer, because changes in morphological traits require changes in multiple genes [
67,
68]. In such a situation, integrated taxonomy with multiple markers is required for an accurate identification.
The two markers used in the present study were sensitive and concordant; both revealed three distinct genetic lineages for
A. nuneztovari
s.l. from the Brazilian Amazon region. These findings are strong enough to indicate that the populations of
A. nuneztovari
s.l. of this region do not belong to a single panmictic species, confirming previous evidence [
17,
18,
21‐
24,
31].
In the phylogenetic analysis made using only sequences of this study, the ML tree retrieved five monophyletic groups, three of which (I, II, III) were represented by the five samples of this study. Group I was represented by the three samples from the central Amazon region and few specimens from AP. Group II comprised most of the specimens from the northeastern region (AP). Group III consisted of specimens from the eastern region (TU). These groups were confirmed by the Bayesian analyses (structure and BAPS) and FCA with microsatellite data.
The groups represented by lineages I and II (
A. goeldii) were not well resolved (Fig.
3), likely due to the small number of informative sites between them for this marker (Additional file
3). Furthemore, there were shared haplotypes between AU and AP. The Bayesian analysis with microsatellite data (Fig.
4) showed very similar results; although this analysis clearly separated lineages I and II, five specimens from AU had a larger proportion of their genomes assigned to the AP sample. Taken together, these findings could reflect shared polymorphism or introgression caused by incomplete lineage sorting [
65], a phenomenon often observed between young lineages and closely related species [
69,
70], as may be the case of lineages I and II of this study.
Comparing the dataset of this study with the sequences from GenBank, the ML tree retrieved six groups, four of which (I, III, IV, V) were represented by specimens of the five localities of this study plus groups II (
A. nuneztovari s.s.) and VI (
A. dunhami) (Additional file
1). All sequences from GenBank identified as
A. goeldii clustered with lineages I and II of this study, suggesting that they represent
A. goeldii. In this case,
A. goeldii was paraphyletic. If this is correct, the three groups could represent two or more species within
A. goeldii. Intriguingly, these data suggest the groups I and III (
A. goeldii) and II (
A. nuneztovari
s.s.) may be same species or incipient species. Supporting these findings, they also presented low genetic distances (1.60–2.0 %). A similar topology (BI tree) was obtained by Scarpassa and Conn [
31], who reported that both samples from Bolivia/Colombia/Venezuela and the central region of the Brazilian Amazon clustered to the same major clade II. In contrast, previous studies [
22,
24,
25] suggested that they are distinct species. Thus, these results could be explained by incomplete lineage sorting caused by retention of ancestral polymorphism due to the extremely recent divergence of this complex.
On the other hand, lineage III had high support in both ML analyses. Interestingly, no sequence from GenBank identified as
A. goeldii clustered to lineage III (group IV), suggesting that it does not belong to
A. goeldii. Furthermore, although all haplotypes were connected in the haplotype network (Fig.
2), there were no shared haplotypes between this lineage and lineage I, which could indicate some barrier to gene flow. Lineage III can be separated from lineages I and II by five to seven fixed differences, respectively. In the Bayesian analysis (Fig.
4), this lineage was considered biologically pure (
Q > 0.95 of belonging to the red group), and in the FCA analysis (Fig.
5) it was the most distant one. Taken together, the lineage III may represent a new species within the Nuneztovari complex and may be following an evolutionary trajectory independent from the other lineages.
In the present study, the barcode region data indicated very high genetic structure and absence of gene flow (Nm < 1) between the samples of the three regions. In contrast, genetic homogeneity was observed among the three samples from the State of Amazonas (lineage I). Therefore, specimens of these locations may belong to a unique species that is undergoing a differentiation process from lineage II, and distinct from lineage III.
The genetic distances (K-2P) among the three lineages were small (1.60–2.32 %) and very similar to those reported by Calado et al. [
25], who also used the barcode region. However, the distance observed between lineages I and III was similar to those between
A. albitarsis
s.s. and
A. oryzalimnetes (2.64 %) of the
A. albitarsis complex [
36] and between
A. dunhami and
A. nuneztovari
s.s. (2.55–2.61 %) in this study. Ruiz-Lopes et al. [
36] observed a threshold of 2.0 % for separating
A. albitarsis H from
A. marajoara, two sister taxa of the
A. albitarsis complex, whereas the genetic distance between
A. triannulatus
s.s. and
A. halophylus/
A. triannulatus C varied from 1.7 to 2.30 % [
69]. As observed, values >2.0 % have consistently been reported between sister taxa of species complexes in the
Nyssorhynchus subgenus [
31,
36,
37,
69‐
71], whereas the intra-specific divergence is rarely >2 %. These results suggest that the members of
A. nuneztovari complex are of recent evolutionary origin, confirming previous studies [
18,
31]. An example of recent divergence was reported in another vector insect,
Lutzomyia umbratilis, that likely consists of two cryptic species which showed genetic distances from 0.8 to 1.4 % and moderately supported clades [
72], suggesting recent diversification.
Most of the lineages observed in the present study were undetected in the isozymes study [
22], except lineage III, likely because of their very recent divergence and the slow evolution rate of this marker for detecting incipient or recently diverged species. However, the sample of TU was the most divergent [
22], indicating that its diversification may have started earlier. The results of this study are partially consistent with those obtained by ITS2 [
17] and mtDNA-RFLP [
18] that identified two groups in the Brazilian Amazon region. This partial disagreement is mainly attributable to differences in sampling strategies between these studies. Our findings, however, match those obtained with the
white gene [
24]. Lineages I and II of this study correspond to lineage 1 of Mirabello and Conn [
24], whereas lineage III corresponds to lineage 4 [
24]. This lineage is represented by samples from Altamira (State of Pará) and Areia Branca (State of Rondônia). Altamira is situated near to TU. The authors [
24] reported two sympatric lineages, with no heterozygotes observed in either Altamira or Areia Branca. Similarly, the samples from TU and Areia Branca (Rondônia) shared haplotypes and clustered together, both in the haplotype network and in the BI tree analyses [
31]. In the present study, an identical situation was observed: most of the specimens from TU clustered in lineage III, whereas the three others clustered in lineage II. Two of these three specimens shared haplotype (H19) with AP, suggesting that two sympatric species might exist in TU: the
A. goeldii group (lineage II) and a new species (lineage III). The occurrence of two distinct genetic pools in TU could explain the highest number of pair-loci (18) in
LD for the microsatellites data, which remained significant after the Bonferroni correction.
The diversification time estimated among the lineages falls in the Pleistocene Epoch (ranging from 0.34 to 0.50 myr), as previously observed [
31], implying divergence within the last one million years. The most plausible hypothesis to explain the diversification among the lineages appears to be climatic changes, such as temperature fluctuations, reduced atmospheric CO
2 and precipitation, occurred during the Pleistocene and which may have influenced the isolation of these populations in refuge areas, causing the differentiation between them by allopatry. Therefore, the two lineages or species sympatric observed in TU could represent secondary contact zones.
However, previous reports have suggested the Amazon river to be a significant barrier to dispersal for several species [
73‐
75], including anophelines [
76,
77]. The sampling of this study was not designed to test the predictions of the riverine barrier hypothesis, but some association may be possible. In this study, the largest differentiation was observed between lineages I and III. MN (lineage I) is situated on the north bank of the Amazon river, whereas CS, AU (both included in lineage I) and TU (lineage III) are located on the south bank. Curiously, mitochondrial and microsatellites markers revealed extensive gene flow, historical and contemporary, between MN and CS/AU, situated in opposite banks. One explanation for this finding is that in these localities the width of the river is not enough to prevent gene flow between populations. Alternatively, the dispersal of these mosquitoes may occur via passive transport, because around these localities, including the MN area, there is intense river traffic. In contrast, there was no gene flow between these locations and TU. Two interfluves (Xingu and Tapajós rivers) separate AU from TU, whereas three interfluves (Xingu, Tapajós and Madeira rivers) separate CS from TU, and these three and the Amazon river separate MN from TU. Therefore, it is possible that these interfluves may be acting as dispersal barriers for these anophelines.
Appreciable and significant mitochondrial and microsatellite differentiation was also observed between AP and TU, situated on opposite sides of the Amazon river delta (mouth). In this region, the Amazon river is widest and can reach up to 50 km in the rainy season. The Amazon river together with the Xingu, Araguaia and Tocantins and other smaller rivers form a large network in this region. This network may act as a dispersal barrier, even porous, restricting the contact between anophelines from the north and south banks, promoting genetic differentiation. Previous studies have reported differentiation between populations of
A. darlingi [
76,
77] and
A. marajoara [
71] in this region. On the other hand, the Mantel test showed that ~90 % of the genetic differentiation found among the five localities is explained by IBD, as observed for the
white gene [
24]. The localities sampled in this study may have influenced these results.
Taken together, the data clearly show that the three genetic lineages studied may be evolving independently in the Brazilian Amazon region. Lineages I and II may represent genetically distinct groups or species within A. goeldii, whereas lineage III may represent a new species and could be the most ancestral one in the Brazilian Amazon region.
Higher levels of intra-population genetic variability, estimated by the microsatellite loci, were detected for the central region (samples from the State of Amazonas) as compared to the samples from TU and AP, supporting previous findings of Scarpassa and Conn [
31]. Based on these results, the authors proposed that the central Amazon region is likely to be the ancestral area of this species complex. In the present study, however, the higher level of genetic variability observed may be a consequence of an extensive contemporary gene flow among specimens of the localities of MN, CS and AU, which represent a panmictic population.