Phylogeography, Vectors and Transmission in Latin America

The overall focus of this chapter is the impact of phylogeographic studies on information pertinent to vector control, and an update on the relative importance and taxonomic status of five malaria vectors, some of which are species complexes, in the subgenus Nyssorhynchus: Anopheles albimanus Wiedmann, Albitarsis Complex, Anopheles aquasalis Curry, Anopheles darlingi Root, and Anopheles nuneztovari s.l. Gabaldon, considering literature predominantly since 2000. This cut-off date is to avoid repetition or overlap with some of the same subjects that have been covered in other places [1-4].


Biology and vector status 2.1. An. albimanus
Throughout its broad, mostly coastal distribution (Figure 1), Anopheles albimanus is an important local vector and is considered to be ecologically adaptable (18). In general this species is crepuscular, zoophilic, exophagic, exophilic and seasonally abundant (19)(20). Despite heteogeneity of several attributes, such as host-feeding behaviour, longevity, insecticide resistance and susceptibility to Plasmodium species, thoughout its distribution, it has maintained single species status (20).

Colombia
An. albimanus is distributed in Colombia along the Atlantic and Pacific coasts ( Figure 1). It is the main malaria vector on the Pacific coast, but its presence is considered a risk factor in other regions, even where malaria transmission is low [22]. Adult abundance of An. albimanus is associated with malaria transmission. The El Niño-Southern Oscillation Event (ENSO), that affects global climatic conditions every 2 to 7 years, has been strongly associated with increases in malaria cases, particularly in areas where An. albimanus is the main malaria vector, such as the Pacific coast [23]. An. albimanus breeding sites are very diverse, ranging from temporary small ponds to lagoons, and even include artificial containers. Its human biting rates can range between a few specimens per night night up to thousands, depending on the availablity of breeding sites in and around villages. Despite its considerable distribution on both coasts, An. albimanus has been found naturally infected with P. vivax only along the Pacific [24] (Table 1). In this region, its biting activity shows at least two peaks, one around midnight and a second one of less intensity before dawn, both indoors and outdoors [4]. Consistent use of insecticidetreated nets (ITNs) in this region could potentially reduce malaria transmission risk. In the Buenaventura peri-urban area, around 20% of the bites occur indoors and the main biting activity is outside houses between 18-21h. Then, at midnight, there is a second peak inside houses. Finally, between 05-06h, activity increases again outdoors [4].  [21].

Peru
An. albimanus in Peru is considered the main malaria vector along the Pacific coast, particularly in the north, where it is seasonal, and linked to agriculture [50]. Due to high insecticide application, mainly in rice fields, An. albimanus is resistant to all insecticides used in public health in this area [51]. Flooded rice fields provide ideal mosquito breeding habitat and An. albimanus density is associated with rice crops. Since 2005, the Peruvian Minister of Health, together with the Agricultural sector, implemented a modified irrigation system, so that the fields are dry for a week, and then intermittently irrigated, resulting in a decrease in mosquito larvae by 87% [52]. An important follow-up question would be whether this An. albimanus larval control has actually resulted in a decrease in malaria incidence rate (MIR), or in local An. albimanus adult female abundance, as measured by human biting rate (HBR) and entomological inoculation rate (EIR).
In several other South American countries (Ecuador, Panama, Venezuela) where An. albimanus is a malaria vector, as determined by sporozoite detection or other comparable information in earlier studies, data based on newer techniques are not available. However, some recent investigations have drawn attention to new distributions or larval habitat characterizations that pinpoint areas of fruitful potential research and possible targets for control measures [53][54][55][56].

Albitarsis Complex
Presently, there are eight recognized species (An. albitarsis s.s., An. albitarsis F, An. albitarsis G, An. albitarsis I, An. deaneorum, An. janconnae, An. marajoara, An. oryzalimnetes) and one lineage (An. albitarsis H) in the Albitarsis Complex [57]. The species described as near An. janconnae from Colombia [58] is now considered to be An. albitarsis I [57]. The overall distribution of members of this complex is wide-ranging, including both Central and South America, as well as some Caribbean Islands [59] (Figure 2).

Brazil
Six species of the Albitarsis Complex are known from Brazil to date: An. albitarsis s.s., An. albitarsis G, An. deaneorum, An. janconnae, An. marajoara, and An. oryzalimnetes, [57,60] ( Figure  3). The most broadly distributed member of this complex is An. marajoara Galvão and Damasceno [57,59]. It can be very abundant locally [12], and its breeding site types vary from swampy shores of lakes and ponds to small road puddles; it is generally associated with sunlight and often with aquatic or semi-aquatic vegetation [59]. It has been found infected by Plasmodium falciparum (Welsh), Plasmodium vivax (Grassi & Feletti) 210 and P. vivax 247, and Plasmodium malariae (Grassi & Feletti), and is a peri-urban as well as a rural vector, depending on locality, availability of breeding sites and hosts [12,25]. It is also associated with deforested areas of the Vale do Ribeira in the southeastern Atlantic Forest of Brazil [61]. Although the EIR in Amapá state, Brazil was found to be lower than that of An. darlingi, it is an important local vector, at least in lowland rainforest in parts of the eastern Amazon [12,62].
The distribution of An. janconnae Wilkerson and Sallum [classified previously as An. albitarsis E; [27,60] appears to be limited to northern Amazonian Brazil, including along the Amazon River [57,60]. Larval habitat types in several localities in Roraima state ranged from marsh to seepages to stream margins, and, based on analyses of several environmental variables, An. janconnae could be classified as a habitat specialist [McKeon, Conn & Povoa, unpublished data, 2012]. An. janconnae was incriminated as a local malaria vector around Boa Vista, the capital of Roraima state [27]. It is likely that the infected specimens identified as An. albitarsis s.l. from this region [26,28] ( Table 1), are An. janconnae, at least according to the geographic distribution [57].
An.oryzalimnetes (Wilkerson and Motoki), previously An. albitarsis B [60] has a broad distribution in Brazil that includes the Amazon region and southern Brazil [57]. It is frequently associated with rice fields, and is anthropophilic [60]. It was determined to be positive by ELISA for Plasmodium in Pará state [M.M. Povoa 2010, unpublished data], and may play a role in local transmission.

Colombia
At least three members of An. albitarsis s.l. are present in Colombia. Anopheles marajoara (some collections of which, according to the map [57], may be An. albitarsis I) is widely distributed [22], and its biology is similar to that described above under An. marajoara in Brazil. It is conisidered to be a regional vector in Colombia [22,39,63]. In the municipality of Puerto Carreno in eastern Colombia near the Venezuelan border, it was detected infected with P. falciparum [29] at a surprisingly high rate (1.92%; 3/152 specimens infected). Here, its peak biting time was 18-19h, with a minor peak from 20-21h, and it was collected both indoors and outdoors. It is suggested that together with An. darlingi, the dominant vector in the area, it is responsible for maintaining local malaria transmission in this municipality [29].
The second member of the complex is sympatric with An. darlingi in the east, and probably involved in malaria transmission. This species has been identified by various names, including An. allopha, An. marajoara, near An. janconnae [58], and most recently as An. albitarsis I [57]. Its known distribution thus far is restricted to Colombia. Relatively little is known about its biology, because of species identification issues, and there is no direct evidence yet for its involvement in malaria transmission. It appears that the specimens from Vichada, Colombia, identified as An. marajoara [64], are An. albitarsis I [57], so the distribution of An. marajoara in Colombia, and its involvement in malaria transmission, need re-evaluation.
The third species, An. albitarsis F, was first described from Puerto Carreño, near the Venezuelan border, [65]. In this locality it was found in sympatry with An. darlingi and a species in the Albitaris Complex now defined as An. albitarsis I. Its distribution is hypothesized to include Colombia, Venezuela and Trinidad [57]. Because it can easily be confused morphologically with An. marajoara, and it is found in regions of malaria endemnicity, this species is of some epidemiological importance.

Venezuela
It now appears that at least An. albitarsis I and possibly also An. albitarsis F are present is Venezuela [57]. Furthermore, An. albitarsis I could be sympatric in some regions with An. marajoara (the identification of which needs to be confirmed in Venezuela using molecular techniques), a local vector of P. vivax in western Venezuela [66]. An. marajoara also plays a significant role, together with An. darlingi, in malaria transmission in five localities in southern Venezuela [30][31]. In this gold mining region, where transmission is yearlong, the peak biting time, (19-21h, mostly before midnight), was comparable with most other reports of An. marajoara. Although An. marajoara was feeding both indoors and outdoors, it was significantly exophagic.

Anopheles aquasalis
This brackish-water breeder is found along the Pacific as far south as Ecuador and along the Atlantic to southern Brazil [1,14] (Figures 3,4). It is rarely found far from the ocean, but it can tolerate quite low salt ion concentrations, and has been detected in freshwater springs. It can be present in enormous numbers in marshy coastal areas, so that even if it is not extremely susceptible to Plasmodium, it can maintain malaria transmission when its abundance is high, especially during the rainy season [67]. It persists as an important local vector in Sucre state, eastern Venezuela, where a series of pioneering studies have identified hotspots of local transmission that are very useful for prevention and control efforts [68][69]. It has also been incriminated as a vector of P. falciparum and [or] P. vivax in Maranhão [32] and Pará states in Brazil [33,70]. In Linden and Madia, Guyana [34], An. aquasalis was detected infected with P. vivax using VecTest, but the total sample size of anophelines collected was very small (n=45). In three towns in Suriname, Paramaribo, Brokopondo and Galibi, An. aquasalis was also detected infected with Plasmodium by ELISA [35], and is likely responsible for local, coastal transmission in this region.

Anopheles darlingi
The species considered to be the most important vector in the Amazon basin is Anopheles darlingi [2,3] ( Figure 5). It is anthropophilic, adaptive and it has been incriminated in many localities in many countries, where it is often labeled a national vector ( Table 1). As such, it has been the focus of a very wide range of research, monitoring and control efforts, and the publication for the first time of its complete genome is an exciting new development (GenBank accession number ADMH00000000).

Bolivia
An. darlingi is distributed in the northeastern Bolivian Amazon, in the departments of Pando, Beni and Santacruz de la Sierra, along the border with Brazil [71]. It shows a biting peak between 19-21h, with 83% of the bites occurring before 22h, when most local people go to bed.
After this time, numbers decline, with little or no activity between 02:30-05h [72]. There have been relatively few studies on this species in Bolivia, and data are very scarce.

Brazil
Most of the newest incriminations of the continued involvement of An. darlingi in malaria transmission originate in Amazonian Brazil (Table 1, Figure 3). Rather than summarizing each new investigation, this section focuses on the findings on An. darlingi in a longitudinal study that investigated bloodmeal hosts, transmission, and seasonal abundance in three riverine villages along the Matapí River in Amapa state, northern Amazonian Brazil [12,62,73]. An ELISA analysis for IgG of common vertebrates found that the highest human blood indices (HBI) were in An. darlingi and An. marajoara. What was unexpected was that the HBIs of An. darlingi varied significantly among the three villages, which are only 1.5-7.0 km apart, likely because of host availability. It was found to be important to conduct a census of animals in each locality to be able to interpret the HBI results correctly. Even though An. darlingi was the most abundant species collected at human landing catches in each village, the HBI of An. darlingi resting collections, from under houses or in vegetation, ranged from 0.017-0.405, demonstrating how opportunistic this species can be, despite its anthropophily [12]. From the same study sites 113,117 mosquitoes collected from 2003-2005 were analyzed by ELISA. For this part of the study, An. darlingi and An. marajoara had the highest proportion of positives and also the highest EIRs, and thus the highest human-vector contact. Nevertheless, An. darlingi is still considered to be more important in this study area than An. marajoara because of its higher EIR [12]. Seasonal abundance was measured for 32 consecutive months of collection and showed that An. darlingi was most abundant during the wet-dry transition period between June and August, and that a strong positive correlation of An. darlingi abundance with rainfall lagged by several months. The latter finding may indicate that rainfall could be an important factor in predicting vector abundance, at least locally.

Colombia
The distribution of An. darlingi in Colombia is widespread but heterogeneous, and hypothesized to be interrupted by the Andes. It is found mainly south and east of the Andes, including the Amazon region, bordering Brazil and Peru, but also north and west of the Andes, along the main Colombian rivers (such as Magdalena, Cauca and Atrato) [22,74]. An. darlingi is the main malaria vector throughout its distribution. Although associated with forest environments, it was also detected in the peri-urban area of the cities of Quibdó (Chocó) and Villavicencio (Meta), where malaria transmission occurs [40]. Breeding sites are the typical streams with slow water movement, but stationary water bodies such as natural and constructed fish ponds also provide good habitats. Similar to Peru (see below), the biting behaviour in Colombia is mostly before midnight, from 18-24h, with a smaller peak at sunrise (05-06h), but there is also persistant biting activity throughout the night [39].

French Guiana
In French Guiana, there has been renewed activity on malaria vectors, with most findings incriminating An. darlingi as the primary vector (

Peru
An. darlingi is the main malaria vector species in eastern Amazonian Peru, the area with the highest malaria transmission in the country. It invaded this region in the 1990s [75], and its distribution now includes peri-urban settlements around the city of Iquitos, Loreto province. This change has been attributed to logging, agriculture and urban expansion, associated with deforestation [76][77]. To date, its greatest abundance is associated with areas of at least partial deforestation [78]. The main breeding sites in the Peruvian Amazon are streams and river margins in forested areas [77], however, the density of fish ponds has shown a positive association with malaria cases along roads in Loreto, suggesting that such ponds could be important local sources of this species [79]. An. darlingi was found naturally infected with P. vivax and P. falciparum in Loreto [46]. The human biting activity, which is similar indoors and outdoors, peaks two hours after sunset [77]. Because of the early evening biting peak, personal protection may be necessary to supplement bed-net use.

Panama
An. darlingi was recently detected in Panamá for the first time, in the eastern Darien region, near the border with Colombia, associated with the highest prevalence of drug-resistant P. falciparum [80]. This species was only collected by human landing catch, and not recorded in light traps, confirming its high anthropophily. Despite the extension and increased frequency of deforestation in Panama, An. darlingi has been detected only in the east. Its discovery suggests that unplanned deforestation should be avoided to prevent further expansion of this very anthropophilic species, and hence potential P. falciparum transmission, to other regions in the country.

Suriname
A timely and important new development in Suriname is the apparent collapse of populations of An. darlingi in the sparsely inhabited interior, in and around three study communities, correlated with two main factors: the introduction of ITNs and climatic events, i.e., unusual flooding which coincided with the beginning of the control activities in 2006 [47]. However, it should be noted that indoor residual spraying (IRS), active case detection (ACD), and a public awareness campaign also were implemented throughout the interior in 2006. As the authors point out, for Suriname, the next challenge is to try to find ways to use these methods to reduce or eliminate transmission among the gold-mining communities, where people are very mobile, and often active (not using ITNs) during potential biting times of An. darlingi. The latter are notoriously plastic, and vary locally and regionally. An important determination will be whether these results [47] can serve as a model for some communities where An. darlingi and malaria transmission are endemic in other countries.

Venezuela
Investigations along the Upper Orinoco River, southern Venezuela from 1994-1995 confirmed that An. darlingi was responsible for most, if not all of the local transmission of P. falciparum, P. vivax and P. malariae [48], that children under the age of 10 were at greatest risk, and that the EIR of An. darlingi was 129 positive bites/person/year. In a gold-mining region in southern Venezuela, studies from 1999-2000 [30][31] also determined that An. darlingi was one of two main vectors (the other was An. marajoara but see above under Albitarsis Complex). Surprisingly, many of the bionomic aspects of the two species in the five localities studied were quite similar (both more abundant during the rainy season, both biting indoors and outdoors with pronounced endophagic behavior), although An. marajoara was more abundant overall. The most striking bionomics difference between the two species was the peak biting time: An. darlingi bit throughout the night with two minor peaks (23-0h and 03-04h), whereas An. marajoara had a peak from 19-21h [30][31].

Anopheles nuneztovari s.l.
Anopheles nuneztovari s.l. is restricted to northern and Amazonian South America ( Figure 6) and has been considered to be two genetically, ecologically and epidemiologically distinc-tive geographic populations, with the perception that the Colombian/western Venezuelan population was a regional vector (anthrophophilic and endo-exophagic) and the Amazonian population, mostly zoophilic and exophagic, was not [1,5,11]. Initially, evidence for malaria transmission by An. nuneztovari s.l. was found predominantly in Colombia and western Venezuela [22,49]. However, a series of positive ELISA results and incriminations of malaria transmission involvement from localities in the Brazilian Amazon (Table 1; Figure 3) since 2000 soon undermined this relatively simple view. An. goeldii, which had been synonymized with An. nuneztovari, was resurrected as a valid species [81]. This work proposed different geographical distributions for each species, with An. goeldii in the Amazon region and An. nuneztovari more restricted to Colombia and Venezuela. The report of the discovery of An. nuneztovari infected with Plasmodium from Saint Georges de l'Oyapock, French Guiana, using results from a longitudinal study (2006)(2007)(2008)(2009)(2010)(2011) is of at least local relevance, but it will be taxonomically important to determine whether this species is actually An. nuneztovari, or might possibly be An. goeldii, since susceptibility of An. goeldii to Plasmodium has not yet been tested [42]. An. nuneztovari s.l. is widely distributed in Colombia [22], particularly in the east, along the Venezuelan frontier, in the northwest region (Departments of Córdoba and Antioquia), where approximately 50% of the malaria cases occur, and in some areas along the Pacific Coast in the west, notably along the San Juan river (Chocó), and in the Buenaventura area (Valle). Specimens from Tierralta, Córdoba, a region of crop and livestock production, where An. nuneztovari s.l. was the most abundant species collected by human landing catches, were infected by P. vivax [38]. The breeding sites include small, permanent ponds, sunlit flooded pastures, and it has been determined that aquaculture ponds are one of this species' most frequent breeding places. In the west, in Cimitarra (Santander), such ponds, characterized as permanent, completely exposed to sun and containing emerging vegetation, particularly grasses, represent approximately 81% of the breeding sites [63]. In Colombia, An. nuneztovari s.l. shows differing biting behavior by region. An exophagic tendency has been described in the northwest (Córdoba) [82], whereas in the east (Santander), a more endophagic behavior has been described [63]. The endophagic-exophilic variability makes control by residual insecticides very difficult.

Anopheles albimanus
An exemplary study, based on large sample sizes using microsatellite markers and a mtDNA ND5 gene fragment, laid the groundwork for several Anopheles albimanus phylogeographic ideas [83]. These researchers detected restricted gene flow that they hypothesized to be the result of the physical barrier of the Central American Cordillera. Recent work, more geographically focused on one country or a region, with additional local sampling, provides additional insights into phylogeography in Central America [20,85] and Colombia [84].
A mitochondrial DNA COI gene fragment and microsatellites were used to test for congruence with biogeographical provinces [86] in Colombia. In this case [84], one population, Turbo, was from Magdalena (Caribbean), three were from Maracaibo (Caribbean), and the four were from Choco (Pacific). The eight populations tested were clearly differentiated into two coastal regions, Caribbean and Pacific, with evidence for a late Pleistocene expansion (estimated to 21,994 years ago) or a selective sweep. Even though there was evidence for historical restrictions to gene flow (COI data), the microsatellites detected contemporary gene flow between the regions. Interestingly, a SAMOVA analysis found an unusual division. Only the three most easterly populations along the Caribbean coast grouped together. The fourth and most western Caribbean population, Turbo, was consistently more closely related to the four Pacific populations. Taken together, these data suggest possible semi-permeable boundaries among the three biogeographical provinces tested. Most relevant to malaria is the fact that the evidence for contemporary gene flow indicates that insecticide resistance genes, for example, could spread readily in these Colombian regions [84].
An. albimanus from Central America was examined using a fragment of the mtDNA COI gene to test the original hypothesis [83]. Physical barriers to gene flow were not detected (i.e., the Central America Cordillera was porous for An. albimanus) and contemporary isolation by distance was not supported [20]. Three divergent, co-occuring haplotype groups were detected using a statistical parsimony network, and these were not evenly distributed across Costa Rica and Panama. A new hypothesis suggested that they could be the result of multiple introductions into the region, probably caused by historical fragmentation and subsequent secondary contact. A more wide-ranging study incorporated the samples from Colombia [84], Ecuador, and Nicaragua with those from Costa Rica and Panama [20] and added two molecular fragments: the nuclear white gene and the ITS2. A SAMOVA analysis defined three large population demes, one from Nicaragua, Costa Rica and the Atlantic coast of western Panama; a second one incorporating the Pacific coast of western Panama, central-eastern Panama and the Caribbean Colombian coast; and a third one restricted to the Pacific coast of Colombia and Ecuador [85]. There were also four haplogroups, based on the COI fragment, which differed little from those found in the earlier Panamanian study [20] except for the addition of a fourth, restricted to the Pacific coast of Colombia and Ecuador, and separated by 18 mutation steps from its nearest haplogroup. Interestingly, because it tracks an earlier history, the white gene network showed much less divergence, supporting the overall conclusion that the primary time-frame for anopheline divergence at the species level is Pleistocene [15]. In summary, the combined An. albimanus data set strongly supported the presence of a single species in this region, which was expected, but also found very robust evidence for Pleistocene geographic fragmentation followed by range expansion across southern Central America [85].

The Albitarsis Complex
Following the newest revelations about the number of species (eight) plus a novel lineage (An. albitarsis H) in An. albitarsis s.l. [57], parts of an earlier study on the biogeography and population genetics of this complex [17] need to be reconsidered and modified. This is particularly the case for An. janconnae, which is more restricted than thought (under the taxonomic name of An. albitarsis E), the expanded distribution of An. albitarsis F (which now includes Venezuela and Trinidad as well as Colombia), the complexities of the distribution of An. marajoara, which really may have a very broad range, newly described An. albitarsis G, distributed along the Brazilian Amazon, and An. albitarsis I, restricted to northwestern Colombia [57]. Despite these problems, one recent study can be used to illustrate the phylogeography of at least An. albitarsis G [as far as it is known; 57] and part of the range of An. marajoara [87]. An. albitarsis G [lineage 2 in reference 87] may be restricted to localities near the Amazon River or its tributaries. It has little population structure and the small subdivisions that were detected in haplotype networks were unrelated to geographic locality. The evidence from the mtDNA COI fragment used in this study indicates that this lineage is older than An. marajoara [lineage 1 in reference 87]. On the other hand, the white gene and ITS2 data detected a single network between An. albitarsis G and An. marajoara, indicating that the divergence is recent. The most compelling result in this study concerning An. marajoara is that SAMOVA defined two population demes along the Amazon River, splitting this species into western and eastern entities with differing genetic characteristics. The boundary is located near Rio Jari in Amapá state, not far from one detected in An. darlingi [88] and an earlier one seen in a study of An. nuneztovari s.l. using restriction fragment length analysis of the mtDNA genome [89]. For An. marajoara, this boundary is permeable, since there were shared haplotypes on either side [87].
A denser sampling of all three species could more rigorously test whether this is the result of underlying geological boundaries or perhaps more recent climatic events.

An. darlingi
Several studies on the phylogeography of An. darlingi have been undertaken. The earliest one [90] used the mtDNA COI fragment and detected a significant genetic division between Central America/northwestern Colombia, and the rest of South America. According to the statistical parsimony network, the more widespread and ancestral haplotypes were in Amazonian and southern South America, suggesting that the Central American/Colombian haplotypes may have originated there. This division was also supported by sequences of the white gene, which found two genotypes, genotype I, restricted to the Amazon, and genotype II, in northwestern Colombia and Venezuela, and Central America [91]. A microsatellite analysis of 1,376 samples also strongly supported the initial COI genetic division, and found substantial structure within the Amazon Basin [91]. The conclusion was that there were two main drivers for this division: differences in effective population size among the divisions, and physical distances between the populations. A more sophisticated analysis of the mtDNA COI fragment included additional Brazilian samples and excluded the Central American samples [88]. These researchers detected six main population groups in South America, and found ancestral distribution to be central Amazonia. They proposed that populations became isolated by three barriers: the Amazon River, the Andes and the southeastern Brazilian coastal ranges. They also found that limited dispersal across some landscape types has promoted differentiation between other proximate populations. A local study of An. darlingi in Córdoba and Antioquia, Colombia, using mtDNA COI, microsatellites and the white gene [74] supported the earlier geographic hypothesis [90], discovering that the five populations tested were more closely related to the Central American populations of An. darlngi that they were to South American An. darlingi. Because of local high gene flow among the five populations, similar control strategies could be implemented in these two contiguous Colombian states. Similarly, newly detected An. darlingi from Panama were most closely related to Colombian and Central American An. darlingi [80]. Concordant phylogeographies were determined for the two neotropical vectors An. darlingi and An. triannulatus [92]. With the mtDNA COI fragment, SAMOVA detected four similar population subdivisions: one in southern coastal Brazil, two in central Brazil and one northeast of the Amazon. Both species originated south of the Amazon River and seem to have followed a similar expansion pathway to their present-day distributions. Other neotropical anophelines with similar distributions may share a common spatial and demographic history with these species, and remain to be evaluated.

An. aquasalis
The only study that attempted to analyse An. aquasalis within a phylogeographic framework was conducted using a fragment of the mtDNA COI gene with specimens from five localities on either side of the Amazon, in Amapá and Pará states, Brazil [93]. The most important findings from this study inferred that despite the width of the mouth of the Amazon, this freshwater delta was not a barrier for the salt-water tolerant An. aquasalis, likely because of so much tidal mixing, and the numerous islands and channels in the region. However, gene flow was restricted, based on isolation by distance that was detected using a Nested Clade Analysis [94]. The relative regional importance of An. aquasalis as a malaria vector has waned since the earlier publications [1,14], so there may be fewer opportunities to pursue phylogeographic questions, especially because the distribution is relatively limited. However, no one has compared specimens from the Atlantic and Pacific coasts, and it is possible that population structure similar to that found for An. albimanus [85], could be detected in An. aquasalis, considering that both species share a relatively narrow coastal distribution in South America, and were subjected to the same kinds of Pleistocene environmental changes.
The revision of the taxonomic status of An. nuneztovari s.l., that now includes An. nuneztovari s.s. and An. goeldii [81] has implications for the interpretation of the first study of An. nuneztovari phylogeography, which focused on the nuclear white gene [95]. The simplest hypothesis to explain the five lineages taxonomically is that the two in Colombia/ Venezuela are An. nuneztovari s.s. and the three in Brazil are An. goeldii. The sharing of haplotypes across the Andes, between eastern and western Venezuela [95], is congruent with and supports findings for An. albimanus [83], The Albitarsis Complex [57] and An. darlingi (Conn, unpublished data) that have hypothesized that the eastern Andean Cordillera is only a partial barrier for anopheline mosquitoes.
A second phylogeographic study was undertaken with some of the same samples, plus new ones from Amazonian Brazil, using a mtDNA COI fragment [97], which charts a more recent history of divergence, all within the Pleistocene, compared with the white gene fragment. In this work, there were two major monophyletic clades, I and II. Specimens from Bolivia/ Colombia/Venezuela represent the most basal subclade, IIC; whereas the Amazonian specimens were found in clades I and II-A and II-B. There were also several localities of sympatry among the clades: five in Amapá, Amazonas and Pará states, Amazonian Brazil, and one in Suriname. None of these are the same as the ones detected by the white gene study, perhaps suggesting that these were later areas of sympatry. There was an intriguing connection detected between the specimens from Colombia/Venezuela and those from Amazonian Bolivia, which had previously been seen when sequences of the rDNA ITS2 were used [98]. This may be the signature of the marine incursion hypothesis [95]. One of clades I, II-A or II-B likely represents An. goeldii, but additional analyses are needed to determine which one, and also to test the hypothesis of multiple species in the Amazon.

Conclusions
There have been many changes in the incrimination, identification and several new insights into the phylogeography of the species discussed in this chapter. The most important taxonomic changes are those in the Albitarsis Complex, with the discovery of two new species, An. albitarsis G and I, and a new lineage, An. albitarsis H. Nothing is known about their involvement in malaria transmission, although their ranges all include malaria endemic areas, or their local contribution to diversity or to a better understanding of the complex patterns of Amazonian biogeography and phylogeography. The relative paucity of new work on An. aquasalis is a reminder that its relative importance appears to be lessening, although it is still likely to be important locally, particularly when in high abundance. Obviously, the importance of An. darlingi in still on the rise in several localities in many countries, attributable mainly to its remarkable adaptability and association with landscape changes. The resurrection of An. goeldii from synonymy is also a milestone, because it provides a first step toward resolving a longstanding discussion about the possible importance of An. nuneztovari s.l. in local transmission in Amazonian Brazil. It may also clarify some aspects of the recent phylogeographic inferences based on white and the mtDNA COI genes. Lastly, the detection of concordant phylogeographies, one of which is An. darlingi in Brazil, depict a clear path towards future research which will have important epidemiological consequences.