| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Malaria remains one of the world’s major health problems even after years of research and public health efforts. Malaria kills more than one million people each year and its affects cost sub-Saharan Africa an estimated US$12 billion in gross domestic product.1 Human malaria is caused by four separate species of parasitic protozoan within the genus Plasmodium. A number of anopheline mosquito species serve as disease vectors and infect humans with Plasmodium sporozoites during blood meals. One major problem in controlling malaria has been the enormous variation of transmission patterns.2 Transmission rates may vary widely in entomologic inoculation rates (0.01–1,000 infected bites per person per year) with contact occurring throughout the entire year or only during selected months.3 Additionally, these factors may vary from year to year and between geographically proximate localities.3 The most lethal form of malaria is caused by Plasmodium falciparum, which is particularly dangerous when carried by anthropophilic species of mosquitoes.
Five species of anopheline mosquitoes serve as the major vectors for P. falciparum in continental sub-Saharan Africa. All five species belong to the subgenus Cellia3,4 with the two most widely studied and most notorious being Anopheles gambiae and An. arabiensis. The three remaining species, An. funestus, An. nili, and An. moucheti, have also received significant recent attention as major vectors.5,6 These five species are the only anophelines in sub-Saharan Africa that frequently transmit P. falciparum malaria. However, it is not necessarily the case that other species of anophelines cannot transmit P. falciparum, but no other species are so anthropophilic and therefore encounter P. falciparum on a regular, if not constant, basis.
Prolonged, consistent contact between P. falciparum and these anthropophilic mosquitoes over long periods of time gives rise to distinct selective pressures on the organism’s immune systems. To use this in the fight against malaria, an accurate understanding of this association requires some knowledge of their shared history. For instance, if all five anthropophilic African mosquito species were each other’s closest relatives (i.e., form a monophyletic clade), one might assume speciation occurred from a highly anthropophilic ancestor, and thus the close association between mosquito, P. falciparum, and humans arose only once and was retained through the several speciation events. Conversely, one could infer that this close, prolonged association arose independently several times if phylogenetic studies show these mosquito species to be non-sister taxa separated by less anthropophilic and/or non-P. falciparum vectors. Thus, phylogenies can help us to understand the historic patterns of association between mosquito vector, P. falciparum, and humans.
The subgenus Cellia contains five taxonomic series with more than 20 species in each (except the Cellia Series).4 Some work has been done in investigating the relationships between and within series; however, much work is needed. For instance, most phylogenetic analyses are characterized by insufficient sampling within or between many series7–12 or poorly supported phylogenetic trees at the nodes of interest.9,12 We do not criticize these studies because their main objectives were generally not to test the monophyly and relationships of the taxonomic series within Cellia. In one of the more extensive studies, Sallum and others estimated phylogenetic relationships among 32 species of mosquitoes within the subfamily Anophelinae based on portions of the mitochondrial genes COI and COII, the nuclear 18S small subunit ribosomal DNA (rDNA), and the expansion D2 region of the nuclear large subunit 28S rDNA gene.12 In that study, they sampled nine species within the subgenus Cellia, the Asiatic species An. stephensi, An. dirus, An. farauti, An. minimus A, An. subpictus, and An. sundaicus, as well as the African species An. arabiensis, An. gambiae, and An. funestus. Similar to the Africa species, Asiatic species are known to be found infective with P. falciparum.13–17 However, unlike the African species, few would be considered highly anthropophilic18 and regular carriers of P. falciparum3,19 and thus not subjected to a constant association with P. falciparum over evolutionary time. As one example, the anthropophilic behavior exhibited by An. minimus A is dependent on the availability of cattle and thus the contact with P. falciparum is sporadic at best.20
In our study, we have compiled a new data set by generating sequence data for two additional anthropophilic African mosquitoes, An. moucheti and An. nili, which we added to a subset of species from the study by Sallum and others.12 Our data set consists of all nine Cellia species from the study by Sallum and others,12 two additional species, An. moucheti and An. nili, and four outgroup species, Uranotaenia lowii, Aedeomyia squamipennis, An. marajoara, and An. atropos, which were selected from various regions of the best estimated phylogenetic tree of Sallum and others.12 GenBank numbers for these taxa are found in Sallum et al.12 Although the addition of only two species from separate taxonomic series (Figure 1
) presents only a slightly more complete sampling from other studies, our study will be the first phylogenetic analysis of all five African P. falciparum vector species, as well as the first phylogenetic analysis specifically aimed at a rigorous test of monophyly and association patterns of these vectors through alternate topology tests and ancestral state reconstruction.
|
Figure 1
shows to which taxonomic series within the subgenus Cellia each species belongs, as well as summarizes the most important phylogenetic results of our study. We believe that the phylogenetic tree based on a combined 18S and 12S data set (Figure 1B) represents the best estimate of relationships. Figure 1B shows An. nili, An. farauti, and An. dirus forming a monophyletic Neomyzomyia clade (posterior probability [PP] = 72), as well as An. funestus, An. moucheti, and An. minimus forming a monophyletic Myzomyia clade (PP = 100, bootstrap [BP] value = 71). Similar to Sallum and others,12 we found little overall phylogenetic signal within the COI and COII mitochondrial genes (Figure 1C), with exception of some weak support (BS value = 54) for the monophyly of the Pyretophorus Series (which was paraphyletic in the rDNA analysis), and strong support (BS value = 100) for sister-taxon status of An. arabiensis with An. gambiae and An. subpictus with An. sundaicus. The mitochondrial DNA (mtDNA) data also supported (BS value = 100) the monophyly of all anopheline taxa. However, we should also note the surprising sister relationship recovered between An. marajoara (an outgroup taxon) and An. dirus. This result and lack of resolution call into question the utility of the COI and COII data sets for estimating the phylogenetic relationships within the subgenus Cellia.
Combining data sets can recover hidden signal because, unlike phylogenetic noise, true signal is not randomly distributed across data sets.26 We tested for hidden signal by performing a combined analysis with all available data. Figure 1D shows the results of our combined analysis. Not surprisingly, the addition of 1,269 characters (from a total of 1941 characters) from the COI and COII data sets caused the Pyretophorus Series to form a monophyletic clade. However, the addition of these data broke apart the monophyly of both the Myzomyia and Neomyzomyia Series. The mtDNA sequence makes up the majority (65%) of the combined data set and given the poor overall signal contained in mitochondrial sequences, we tend to give more weight to the relationships estimated with the rDNA data set. The exception may be the monophyly of the Pyretophorus Series because this relationship was strongly supported by the mtDNA and further investigation, possibly with different genes, is needed to clarify this question. To date, several studies, including this one, have supported the monophyly of most taxonomic series,7,8,12 whereas other studies have supported the paraphyly of various series.7,9 A much better sampling of both taxa and genes is needed to truly test the monophyly of these taxonomic series.
We have presented best phylogenetic estimates for the relationships between some species within the subgenus Cellia (Figure 1). However, a best phylogenetic estimate does not always equate to a significantly better phylogenetic estimate when compared with an alternative topology.27 We used maximum likelihood methods to test our best estimate (Figure 1B) against some possible alternative topologies. We generated three alternative topologies by estimating likelihood trees (using the TrN + I + G model of evolution) from the combined rDNA data sets with the following constraints: 1) all highly anthropophilic African species form a monophyletic clade, 2) all species from the Pyretophorus Series form a monophyletic clade, and 3) all relationships illustrated by Sallum and others12 (their Figure 5) are maintained. Likelihood scores for our best estimate tree and the three constrained trees were as follows: our best estimate tree (which was identical to the tree estimated by Bayesian methods) = –1,839.29, the African monophyly tree = –1,857.05, the Pyretophorus monophyly tree = –1841.90, and the relationship tree of Sallum and others = –1845.1121. We performed the conservative Shimodaira-Hasegawa test28 to compare each of the alternative constrained trees with our best estimate. This test showed that our best estimate tree was preferred to the alternative at the following P values: African monophyly tree, P < 0.05; Pyretophorus monophyly, P = 0.20; and relationships of Sallum and others, P = 0.12. Due to the conservative nature of this test, we are confident that our best estimate tree is significantly better than the African monophyly tree and slightly better than the estimate of Sallum and others. However, our analysis only provides minor evidence against the monophyly of the Pyretophorus Series and as such further studies are needed.
This study shows that the constant association between P. falciparum and its African mosquito vectors probably arose independently several different times in the past. However, it is also possible that this association arose in a common ancestral species and was then lost or altered (a shift toward more zoophilic or exophagic behavior) in some of the later evolved species. To test between these two alternative hypotheses, we reconstructed the most parsimonious character states (highly anthropophilic and encounter P. falciparum on a constant basis versus limited P. falciparum contact) for ancestral taxa based on our best estimate tree using Mesquite 1.05.29 Weighting both character states equally, we found that the regular association between mosquito and P. falciparum has arisen two or three times in the past with equal probability. However, if we assume that gaining the association is even slightly more probable than escaping it, it may have occurred as many as three or four times independently. Also, An. funestus belongs to a group of at least five morphologically similar species (not included in this study) of which An. funestus in the only human Plasmodium vector,3 indicating that P. falciparum acquisition in An. funestus and An. moucheti were probably independent events. Figure 1B shows the preferred scenario (four acquisitions) and the likely positions on the tree branches where long-term associations with P. falciparum likely began.
The conclusion that the five African vectors began regular, constant encounters with P. falciparum at least three times independently is not surprising. Most researchers are aware these vectors are member of taxonomic series and species complexes with non-vector species. Our study provides addition evidence, within a more rigorous framework of statistical tests, that the close association of P. falciparum and several of its mosquito vectors has occurred multiple times in the evolutionary past. This is important because as such, each instance will serve as a natural replicate experiment. Since each African vector species has gone through thousands of years of co-evolution with P. falciparum independently, researchers can contrast immune responses in the vectors and determine which immune genes have responded most to this association and possibly which are most closely related to resistance. This knowledge could be very important in the construction of genetically modified mosquitoes and in the battle against malaria.
Received February 5, 2005. Accepted for publication May 24, 2005.
Acknowledgments: We thank Ralph E. Harbach for useful discussions on phylogenetic relationships among anopheline mosquito species; Vincent Medjibe for supplying the An. nili and An. moucheti samples collected in Bayanga, Central African Republic; and Carlo Costantini (Centre National de Recherche et Formation Sur le Paludisme, Ougadouguo, Burkina Faso) for identifying them. We also thank three anonymous reviewers for their very useful and insightful comments on an earlier draft of this report.
Financial support: Jonathon C. Marshall is supported by a National Institutes of Health (NIH) multidisciplinary parasitology training grant (PI: Dr. Diane McMahon-Pratt, Yale School of Public Health #2T32AI07404). Jeffrey R. Powell is supported by NIH grant RO1 AI46018.
* Address correspondence to Jeffrey R. Powell, Department of Ecology and Evolutionary Biology, Yale University, PO Box 8105, New Haven, CT 06520-8105. E-mail: jeffrey.powell{at}yale.edu 
Authors’ addresses: Jonathon C. Marshall, Department of Ecology and Evolutionary Biology, Yale University, PO Box 8105, New Haven, CT 96520-8105, Telephone: 203-432-3886, Fax: 203-432-6066, E-mail: jonathan.marshall{at}yale.edu. Jeffrey R. Powell, ESC 170, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT 96520-8105, Telephone: 203-432-3887, Fax: 203-432-6066, E-mail: jeffrey.powell{at}yale.edu. Adalgisa Caccone, ESC 140, Yale Institute for Biospherics Studies–Molecular Systematics and Conservation Biology Laboratory, Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT 96520-8105, Telephone: 203-432-5259, Fax: 203-432-7394, E-mail: adalgisa.caccone{at}yale.edu.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
50% are shown. OG = outgroup taxa; An. = Anopheles; Aed. = Aedes; Ur. = Uranotaenia; * = close, continual association between members of the mosquito species and Plasmodium falciparum; 
= likely origin of close mosquito-P. falciparum association. a, Taxonomic series within the subgenus Cellia with corresponding species used in this study. b, Phylogram representing the best phylogenetic estimate and branch lengths from a Bayesian analysis of the combined rDNA (18S and 28S) sequences, c, Parsimony bootstrap consensus tree illustrating the low phylogenetic signal for the mtDNA sequence. d, Bayesian tree estimated from a data set containing all mtDNA and rDNA sequence data.