INTRODUCTION
Disease control directed at vectors such as ticks, flies and mosquitoes has proved to be a cost-efficient and successful approach (Michalakis and Renaud, Reference Michalakis and Renaud2009), but effective vector control may sometimes suffer setbacks from a lack of reliable identification of vectors (Monteiro et al. Reference Monteiro, Escalante and Beard2001; Marquardt, Reference Marquardt2004). Correct identification of vectors directly relies on systematics research, such as research on species delimitation and reconstructions of evolutionary history (Manguin et al. Reference Manguin, Garros, Dusfour, Harbach and Coosemans2008). From a control perspective, it may appear that the systematics of vectors is well understood, but for a large number of important vector species there is actually a lack of basic systematics knowledge. Among systematists, it is for instance known that several of the lower taxonomic levels of the genus Anopheles, such as Series, Groups, Subgroups and complexes, may not always represent the actual evolutionary history of the organisms (Foley et al. Reference Foley, Bryan, Yeates and Saul1998; Krzywinski et al. Reference Krzywinski, Wilkerson and Besansky2001; Sallum et al. Reference Sallum, Schultz, Foster, Aronstein, Wirtz and Wilkerson2002; Harbach, Reference Harbach2004, Reference Harbach2007). Furthermore, for a large number of genetic species-level studies, one morphological taxon has turned out to be a species complex, where the species may differ in traits that are of importance for vector control, such as vectorial capacity, behaviour or ecology (Collins and Paskewitz, Reference Collins and Paskewitz1996; Manguin et al. Reference Manguin, Garros, Dusfour, Harbach and Coosemans2008). Therefore, we suggest that vector control programmes could benefit greatly from an improved knowledge of vector evolution and systematics, and from a better understanding of the uncertainties associated with systematics and methods of species delimitation. We outline some of the recent major advances in the systematics discipline and give examples of the recent research on cryptic species. We also discuss how recent development of systematics enables it to be more informative for targeted vector control strategies.
UNDERSTANDING SPECIES DIFFERENCES
Recent whole-genome scans of hundreds of thousands of genetic markers (Emelianov et al. Reference Emelianov, Marec and Mallet2004; Scotti-Saintagne et al. Reference Scotti-Saintagne, Mariette, Porth, Goicoechea, Barreneche, Bodénès, Burg and Kremer2004; Nosil et al. Reference Nosil, Egan and Funk2007) have led to a better insight into how the genomes of two closely related species differ. The picture is emerging of a ‘mosaic genome’, with gradual accumulation of reproductive isolation across limited genomic regions (della Torre et al. Reference della Torre, Costantini, Besansky, Caccone, Petrarca, Powell and Coluzzi2002; Coyne and Orr, Reference Coyne and Orr2004; Wu and Ting, Reference Wu and Ting2004; Via, Reference Via2009). If these cases represent common patterns of genome-wide differences, it seems possible that reproductive isolation can occur through incompatibilities in a small number of genes, while gene flow (Table 1) may still occur along other parts of the genome. Complex gene flow boundaries have already been found among the An. gambiae chromosomal forms (della Torre et al. Reference della Torre, Costantini, Besansky, Caccone, Petrarca, Powell and Coluzzi1997; Turner et al. Reference Turner, Hahn and Nuzhdin2005; Slotman et al. Reference Slotman, Mendez, della Torre, Dolo, Toure and Caccone2006; Lawniczak et al. Reference Lawniczak, Emrich, Holloway, Regier, Olson, White, Redmond, Fulton, Appelbaum, Godfrey, Farmer, Chinwalla, Yang, Minx, Nelson, Kyung, Walenz, Garcia-Hernandez, Aguiar, Viswanathan, Rogers, Strausberg, Saski, Lawson, Collins, Kafatos, Christophides, Clifton, Kirkness and Besansky2010; Neafsey et al. Reference Neafsey, Lawniczak, Park, Redmond, Coulibaly, Traore, Sagnon, Costantini, Johnson, Wiegand, Collins, Lander, Wirth, Kafatos, Besansky, Christophides and Muskavitch2010). For vector control, this may mean that there is a potential for transgenic elements or biocide resistance to spread across species boundaries (Djogbenou et al. Reference Djogbenou, Chandre, Berthomieu, Dabire, Koffi, Alout and Weill2008), so future monitoring programmes will perhaps need to extend monitoring of vectors across species boundaries in species complexes.
Table 1. Definitions of terms used for the speciation process
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20160921022351353-0566:S003118201100062X:S003118201100062X_tab1.gif?pub-status=live)
TOWARDS A SYNTHESIS OF SYSTEMATICS AND EVOLUTIONARY HISTORY
Genomic research has advanced our understanding of how species are formed, but it has also highlighted strong discrepancies between what we understand a species to be, and how species have been described and delimited – a rift that current systematics is trying to suture. A formal species description requires that the new species is given a valid name following the International Code of Zoological Nomenclature (ICZN), and a holotype specimen is designated (ICZN, 1999). Information must also be included that distinguishes a new species from related species, but the Code imposes no limit on the quality or nature of the information (morphological characters, DNA sequences, chromosome banding patterns, or a combination of different types of information) (ICZN, 1999). The quality of evidence presented in vector species designations therefore varies widely and has evolved with technological advancement; early 20th century species delimitation was conducted through a brief comparison of key morphological differences to similar species (Wilson, Reference Wilson2004) (e.g. Theobald, 1901–1910; Dönitz, Reference Dönitz1903). For species descriptions of later date, the delimitation may also include a wider geographical sampling, ecological information and chromosome, allozyme or molecular evidence (e.g. White, Reference White1985; Baimai et al. Reference Baimai, Andre, Harrison, Kijchalao and Panthusiri1987; Wilkerson et al. Reference Wilkerson, Reinert and Li2004; Linton et al. Reference Linton, Dusfour, Howard, Ruiz, Manh, Trung, Sochanta, Coosemans and Harbach2005). Regardless of the evidence, a species description remains a hypothesis, which may or may not be correct.
The problem of species delimitation using pattern description
Many of these attempts at species delimitation are, as in Linnaean times, implicitly essentialist; regarding species as fixed entities, preserved through time and space, and they may neglect biologically relevant within-species heterogeneity and gene flow across species boundaries. Pattern-based species descriptions are simple comparisons of similarities and differences in morphology or DNA markers, rather than on in-depth study of evolutionary history. Some possible artefacts can be avoided by integrating results from several sources of data to reach an overall conclusion, a so-called ‘total evidence approach’ or ‘integrative taxonomy’ (Puillandre et al. Reference Puillandre, Baylac, Boisselier, Cruaud and Samadi2009; Roy et al. Reference Roy, Dowling, Chauve and Buronfosse2009). However, the problem remains that many such studies do not state a priori criteria for identifying species boundaries, so they interpret each clade as a separate species, irrespective of the biogeographical history of the clade (Sites and Crandall, Reference Sites and Crandall1997; Wiens, Reference Wiens2007). This has the effect that the hypothesis of ‘new lineage equals new species’ is rarely rejected, even when it is unclear if the lineages are real or a result of short-term allopatric fragmentation or incomplete geographical sampling (Zarowiecki et al. Reference Zarowiecki, Walton, Torres, McAlister, Htun, Sumrandee, Sochanta, Dinh, Ng and Linton2011). This may lead to taxonomic instability, which makes it more difficult to use the designated species in other studies (Godfray, Reference Godfray2002; Wiens, Reference Wiens2007), including those for vector control purposes.
Enhanced species delimitation using a process-oriented approach
A better fit between species definition and description could perhaps be achieved by conducting mating experiments, which investigates postzygotic reproductive isolation. Unfortunately, mating experiments are not feasible for many vector taxa, and do not consider geographical, pre-zygotic or ecological barriers to reproduction (Table 1). A promising alternative to create more coherence between species and species delimitation is to use a process-oriented approach, which evaluates current and historical gene flow (Table 1). By thoroughly reconstructing the entire evolutionary history of a species; determining historical gene flow, population structure and biogeography, systematics research provides a much more comprehensive understanding of the species, and more information than just vector identification, some of which may be relevant for vector control. For this purpose, DNA sequence markers have an advantage over cross-mating experiments, because they can be used to detect long-term gene flow and biologically relevant sub-populations in the vector species, not only current post-zygotic mating barriers between individual populations. This might be not only of theoretical value, but also ubiquitously useful, as seen over time and space, one could argue that all species in fact are ‘in speciation’, i.e. they may contain several independently evolving lineages, which may become completely reproductively isolated sometime in the future – or indeed may not be maintained when the ecology, geography or climate changes (Lamont et al. Reference Lamont, He, Enright, Krauss and Miller2003; Seehausen; Reference Seehausen2006), as long as complete post-zygotic reproductive isolation has not occurred (Nosil et al. Reference Nosil, Harmon and Seehausen2009).
IMPROVED PRACTICAL METHODS FOR SPECIES DELIMITATION
Species delineation has not only moved forward theoretically, but technology and statistics have developed as well. Newly introduced methods for species delimitation are often based on statistical testing of hypotheses, and an assessment of historical or current gene flow (Pons et al. Reference Pons, Barraclough, Gomez-Zurita, Cardoso, Duran, Hazell, Kamoun, Sumlin and Vogler2006; Stockman et al. Reference Stockman, Beamer and Bond2006; Stockman and Bond, Reference Stockman and Bond2007; Cummings et al. Reference Cummings, Neel and Shaw2008). They have evolved along two different lines: the rapid large-scale methods that aim at charting mostly unknown territory (Monaghan et al. Reference Monaghan, Balke, Gregory and Vogler2005; Pons et al. Reference Pons, Barraclough, Gomez-Zurita, Cardoso, Duran, Hazell, Kamoun, Sumlin and Vogler2006) and the more specific methods that deal in depth with a small number of taxa (Templeton, Reference Templeton1998, Reference Templeton2001; Posada et al. Reference Posada, Crandall and Templeton2000). These rapid methods are often based on information from a small number of genetic markers and are not expected to give completely correct species delimitation in every case. They provide instead statistical species delimitation, by trying to recover monophyletic clades in phylogenetic trees (Pons et al. Reference Pons, Barraclough, Gomez-Zurita, Cardoso, Duran, Hazell, Kamoun, Sumlin and Vogler2006; Cummings et al. Reference Cummings, Neel and Shaw2008), similarity above a certain threshold value (Hebert et al. Reference Hebert, Stoeckle, Zemlak and Francis2004) or statistical clustering (Schloss and Handelsman, Reference Schloss and Handelsman2005, Reference Schloss and Handelsman2006). Although some statistical methods have been criticised for oversimplifying species delimitation (DeSalle et al. Reference DeSalle, Egan and Siddall2005; Ebach and Holdrege, Reference Ebach and Holdrege2005), such methods have proved useful when applied to real problems, especially in combination with morphology and biogeography (DeSalle et al. Reference DeSalle, Egan and Siddall2005; Vogler and Monaghan, Reference Vogler and Monaghan2007). They can, for instance, rapidly reveal if there is a discrepancy between species delimitation resulting from different data (e.g. DNA, morphology, polytene chromosomes), and indicate clades of apparent taxonomic inflation or cryptic diversity. One caveat is that they are mostly pattern based rather than process orientated, so they can only be seen as a first step towards a more complete understanding of the evolutionary history of a species (Vogler and Monaghan, Reference Vogler and Monaghan2007).
Biogeographical methods for species delimitation
In studies of vector systematics and species delimitation, a much more detailed understanding is often needed than can be provided by rapid methods. These methods nevertheless need to be logical, robust and amenable to scientific validation, e.g. through statistical testing. Several such methods are available, and usually depend on one or more properties one would expect species to acquire during a speciation process, such as monophyletic lineages (Baum and Shaw, Reference Baum, Shaw, Hoch and Stephenson1995), significantly higher genetic divergence compared with intraspecific polymorphism (Highton, Reference Highton1990; Good and Wake, Reference Good and Wake1992), lack of gene flow (Porter, Reference Porter1990), lack of shared polymorphism (Sneath and Sokal, Reference Sneath and Sokal1962; Davis and Nixon, Reference Davis and Nixon1992) and/or heterozygote deficiency (Doyle, Reference Doyle1995). Some tests combine two or more of these categories: morphological differentiation and genetic monophyly (Wiens and Penkrot, Reference Wiens and Penkrot2002), monophyly and ecological differentiation (Stockman and Bond, Reference Stockman and Bond2007) or monophyly, gene flow and ecological differentiation (Templeton, Reference Templeton1998; Posada et al. Reference Posada, Crandall and Templeton2000; Templeton, Reference Templeton2001). These fine-scale methods are primarily developed from classic population genetics approaches, but make use of recently introduced biogeographical methods for estimation of population expansion, time of divergence (Drummond and Rambaut, Reference Drummond and Rambaut2007), levels of differentiation (Pritchard et al. Reference Pritchard, Stephens and Donnelly2000, Reference Pritchard, Falush and Stephens2002; Falush et al. Reference Falush, Stephens and Pritchard2003) and population structure (Dupanloup et al. Reference Dupanloup, Schneider and Excoffier2002). The software LAMARC simultaneously estimates gene flow and population growth (Kuhner et al. Reference Kuhner, Yamato, Beerli, Smith, Rynes, Walkup, Li, Sloan, Colacurcio and Felsenstein2005), and the ‘isolation-with-migration’ model (IM) simultaneously estimates gene flow and divergence time between two populations or species, using multi-locus data (Wakeley and Hey, Reference Wakeley and Hey1997; Hey and Nielsen, Reference Hey and Nielsen2004). As these methods need multi-locus data to give accurate estimates (Hey and Nielsen, Reference Hey and Nielsen2004), they may be costly and technically challenging for non-model organisms (Knowles, Reference Knowles2004), but there is a growing number of such studies available for vector species (Loaiza et al. Reference Loaiza, Scott, Bermingham, Rovira and Conn2010; Morgan et al. Reference Morgan, Linton, Somboon, Saikia, Dev, Socheat and Walton2010; Zarowiecki et al. Reference Zarowiecki, Walton, Torres, McAlister, Htun, Sumrandee, Sochanta, Dinh, Ng and Linton2011). These studies have an advantage over previous systematics research because they can give information about the evolutionary process, e.g. estimates of times of divergence, species origin, population structure, selection on particular genes, population growth and gene flow heterogeneity across genomes – all of which may be important for understanding how the vectors evolved or how insecticide resistance can spread between incipient vector species.
WHY VECTOR EVOLUTIONARY HISTORY MATTERS FOR PARASITE CONTROL
Many otherwise well-studied parasite vectors have, through recent genetic studies, been revealed as cryptic species complexes (Collins and Paskewitz, Reference Collins and Paskewitz1996; Manguin et al. Reference Manguin, Garros, Dusfour, Harbach and Coosemans2008). The species in a complex are morphologically indistinguishable under field conditions, but they are reproductively isolated (Table 1) and may differ significantly in ecology or behaviour. One of the most intensively researched vector species, the African malaria mosquito Anopheles gambiae, belongs to a well-known cryptic species complex, where some species differ widely in feeding preferences and their importance as malaria vectors (Gillies and Coetzee, Reference Gillies and Coetzee1987), but still new genetic forms are being discovered (Riehle et al. Reference Riehle, Guelbeogo, Gneme, Eiglmeier, Holm, Bischoff, Garnier, Snyder, Li, Markianos, Sagnon and Vernick2011). There are also, perhaps less well known, examples of where better species identification has been achieved through biogeographical studies: Anopheles subpictus is widely distributed across southern and eastern Asia, and is considered a major vector of malaria (Abhayawardana et al. Reference Abhayawardana, Wijesuriya and Dilrukshi1996). In India, there are at least two species within this nominal taxon that differ in many traits, including vectorial capacity (Suguna et al. Reference Suguna, Rathinam, Rajavel and Dhanda1994; Sahu, Reference Sahu1998), but a recent study revealed several additional species in Southeast Asia, some of which separated several million years ago (Zarowiecki, Reference Zarowiecki2009). The number of species that are actually malaria vectors is still unknown, but as several of the species in the complex are sympatric, it is certain that for a large number of previous studies on vectorial capacity, biting behaviour and insecticide resistance, it is unclear whether one or several species were studied. Rapid molecular identification methods for the An. subpictus complex are urgently needed to identify the vectors, so that their bionomics can be investigated and vector control methods properly focused. This is important, as mainland Southeast Asia has a relatively high incidence of P. falciparum malaria, confirmed parasite resistance to malaria drugs (Van Bortel et al. Reference Van Bortel, Trung, Thuan, Sochantha, Socheat, Sumrandee, Baimai, Keokenchanh, Samlane, Roelants, Denis, Verhaeghen, Obsomer and Coosemans2008), and restricted resources for vector control.
The An. dirus complex includes seven species (Harbach, Reference Harbach2004; Sallum et al. Reference Sallum, Peyton and Wilkerson2005). Of these, An. baimaii, An. dirus and An. scanloni are all major vectors of malaria in Southeast Asian tropical forests (Sallum et al. Reference Sallum, Peyton and Wilkerson2005). These three species are largely allopatric, but where they coexist there is a deficiency of heterozygotes in both polytene chromosomes and allozymes (Baimai et al. Reference Baimai, Andre, Harrison, Kijchalao and Panthusiri1987; Green et al. Reference Green, Munstermann, Tan, Panyim and Baimai1992). Post-mating reproductive barriers have been identified for all species, but some fertile offspring can be found between An. dirus and An. scanloni (Baimai et al. Reference Baimai, Andre, Harrison, Kijchalao and Panthusiri1987). There are discrepancies between the phylogenies of these species constructed from nuclear and mitochondrial markers, and it is believed that mitochondrial introgression between An. baimaii and An. dirus could explain these observations (Sallum et al. Reference Sallum, Foster, Li, Sithiprasasna and Wilkerson2007; O'Loughlin et al. Reference O'Loughlin, Okabayashi, Honda, Kitazoe, Kishino, Somboon, Sochantha, Nambanya, Saikia, Dev and Walton2008; Morgan et al. Reference Morgan, Linton, Somboon, Saikia, Dev, Socheat and Walton2010). As a consequence, there is a concern that genes, such as those that confer insecticide resistance, could disperse across species boundaries (Morgan et al. Reference Morgan, Linton, Somboon, Saikia, Dev, Socheat and Walton2010).
The An. albitarsis complex is broadly distributed in the Central and South America. Six species are currently recognised (Motoki et al. Reference Motoki, Wilkerson and Sallum2009), only three of which have been demonstrated to be vectors of malaria: An. deaneorum, An. marajoara and An. janconnae (Povoa et al. Reference Povoa, de Souza, Lacerda, Santa Rosa, Galiza, de Souza, Wirtz, Schlichting and Conn2006). Post-mating barriers have been investigated only between An. deaneorum and the non-vectors An. oryzalimnetes (Klein et al. Reference Klein, Lima, Tada and Miller1991) and An. albitarsis (Lima et al. Reference Lima, Valle and Peixoto2004). As several members of the complex are sympatric (Brochero et al. Reference Brochero, Li and Wilkerson2007), behavioural, ecological and malaria transmission studies have been confounded by identification issues, but due to improved systematic techniques five species can now be reliably separated using PCR-RFLP of ITS2 (Li and Wilkerson, Reference Li and Wilkerson2005; Brochero et al. Reference Brochero, Li and Wilkerson2007).
These examples reveal a range of genetic and behavioural diversity within species complexes. The basic systematics provides species-diagnostic tools, but also insights into the evolution of traits that are important for mosquito vectorial capacity and vector control, and allows for further research into the biology of vector species (Powell et al. Reference Powell, Petrarca, della Torre, Caccone and Coluzzi1999; Manguin et al. Reference Manguin, Garros, Dusfour, Harbach and Coosemans2008).
CONCLUSIONS AND FURTHER DIRECTIONS
Some parts of more general vector control programmes (such as utilizing non-selective insecticidal spraying) can lead to biocide resistance and have adverse side effects on human health and ecosystems. This review has illustrated how understanding the evolutionary history of vector species is important for correct identification of vectors – which in turn is essential for developing efficient vector control programmes targeted at specific vectors. We have highlighted the difference between species descriptions, species delimitation and the current understanding of how species are formed, and suggest that correct species identification and delimitation should be underpinned by information on the evolutionary process of the vector species. For instance, systematics studies can provide information about genetic heterogeneity within the vector (Ramsey et al. Reference Ramsey, Salinas, Rodriguez and Beaudoin1994; Habtewold et al. Reference Habtewold, Povelones, Blagborough and Christophides2008), allowing vector control studies to incorporate vector genetic diversity. Detailed systematics data on gene flow of the vector are also crucial for transgenic vector control projects (Michalakis and Renaud, Reference Michalakis and Renaud2009). There are still significant challenges ahead for systematics, such as – in a standardised way – being able to incorporate evolutionary history in species descriptions, and allowing for differences in gene flow across different parts of the genomes of hybridising species. Nevertheless, it is clear that recent understanding of speciation and species delimitation renders systematics research increasingly useful for vector control studies.
FINANCIAL SUPPORT
This work was supported by the Natural History Museum (M.Z., Prize Studentship).