The Anopheles (Anopheles) maculipennis complex (Diptera: Culicidae) in Greece

Mosquitoes belonging to the Maculipennis Complex of Anopheles subgenus Anopheles were collected in 10 prefectures across Greece: Attiki, Evros, Florina, Fthiotida, Ioannina, Lakonia, Magnesia, Rodopi, Serres, and Xanthi. DNA was extracted from 276 specimens and sequences for the nuclear rDNA ITS2 region were obtained from 257 of these (93.1%). Four members of the An. maculipennis Meigen complex were identified: An. maculipennis, An. sacharovi Favre, An. melanoon Hackett, and An. messeae Falleroni. Species were identified based on correlation of their sequences with those available in GenBank. All four species have been incriminated as primary or secondary vectors of malaria. Species distributions in relation to potential threats of reintroduction of malaria in Greece are discussed. This study comprises the most complete study of the Maculipennis Complex in Greece to date, and contributes substantially to the current knowledge of the genetics and distributions of the four species.


Introduction
Malaria was prevalent across Europe as late as the mid-20th century. Being abundant and widespread, Anopheles maculipennis Meigen was circumstantially incriminated as the primary vector. Following studies into the phenomenon of ''anophelism without malaria'', this taxon was the first mosquito sibling species complex to be recognized when two biologically/reproductively distinct species were distinguished by means of egg and wing morphology, correlated with behavioural and ecological differences (Falleroni 1926;van Thiel 1927). Following these early works, extensive efforts were made to elucidate all Macedonia (Shannon and Hadjinicolaou 1941). Anopheles maculipennis and An. sacharovi were reported from the island of Corfu (Stephanides 1937(Stephanides , 1938 and the Copias district of mainland Greece (Sakellariou and Lane 1977). In the early literature, An. sacharovi was widely reported in Greece (Cardamatis 1931;Barber 1935;Shannon 1935;Rice and Barber 1937;Weyer 1942;Eichler 1944;Peus 1954), although populations were decimated and the species believed to be eliminated following the activities of the Malaria Eradication Programme in the mid-1950s. The reintroduction of An. sacharovi was reported in the central region of Lamia (Fthiotida Prefecture) (Hadjinicolaou and Betzios 1973) and, more recently, it was found in sympatry with An. maculipennis in the wetlands of northeastern Greece (Ouzounis and Samanidou-Voyadjoglou 1993).
Following his synonomy of An. subalpinus with An. melanoon, White (1978) suggested that An. maculipennis, An. sacharovi, and An. melanoon were the three members of the Maculipennis Complex present in Greece. The proposed distribution of An. atroparvus (Dahl and White 1978;White 1978) suggested that this species may also be present in Greece. For these reasons, An. atroparvus was also included in the most recent species list (Samanidou-Voyadjoglou and Darsie 1993) and morphological keys (Darsie and Samanidou-Voyadjoglou 1997;Samanidou-Voyadjoglou and Harbach 2001) of Greek mosquitoes.
Three species of the Maculipennis Complex, An. atroparvus, An. sacharovi, and An. labranchiae, are known to be efficient current or historical malaria vectors in Europe (Jaenson et al. 1986;Ribeiro et al. 1988;Kasap 1990;Jetten and Takken 1994;Romi et al. 1997Romi et al. , 2001Romi et al. , 2002Romi 1999;Alten et al. 2000). Climatic changes, including global warming and associated increase of precipitation, are expected to extend vector ranges and population sizes of some species, potentially increasing malaria transmission rates (Snow 2000). Indicative of possible increased involvement in malaria transmission, An. maculipennis and An. melanoon (as An. subalpinus) were recently incriminated as secondary vectors in the Biga Plains of Turkey (Alten et al. 2000), and most recently An. sacharovi has re-emerged as a vector of malaria in Armenia (Romi et al. 2002). Anopheles messeae, a reportedly efficient vector in western Asia but not previously in Europe (Bruce-Chwatt and de Zulueta 1980), has been identified as the principal vector of resurgent malaria in the Ukraine and Russia (Nikolaeva 1996).
Despite the eradication of malaria in Europe following extensive vector control programmes after the Second World War, increasing numbers of malaria cases have been reported in southern European countries over the last decade (Sartori et al. 1989;Nikolaeva 1996;Baldari et al. 1998;Linton et al. 2001bLinton et al. , 2002aRomi et al. 2001). These cases, along with Plasmodium carriers in the form of exotic travellers (Steffen and Brethens 1992) and increasing immigrant labour from endemic areas, have heightened concern for the reintroduction of malaria into southern European countries. This is especially true in Greece and Italy, where malaria was once highly endemic and competent mosquito vectors still exist (Jetten et al. 1996;Romi et al. 1997Romi et al. , 2001Lindsay and Birley 1996;Romi 1999;Linton et al. 2001b).
Correct vector identification is essential to assess the potential risk of malaria in these border regions, and to devise appropriate control or monitoring strategies. In view of the heightened risk of reintroducing malaria, unreliable historical records, and incomplete knowledge about the ecology and disease relations of the mosquitoes, it is important to characterize the species composition and distributions of members of the Maculipennis Complex in Greece. These studies are essential for the development of effective mosquito control programmes and will serve to identify those areas of highest risk for the reintroduction of malaria. In this study, we investigated the composition and distribution of members of the Maculipennis Complex from 10 prefectures in Greece and determined their identity by correlation of their nuclear ITS2 sequences with those available in GenBank (Marinucci et al. 1999;Proft et al. 1999;Djadid, direct GenBank submissions 2001;Linton et al. 2001bLinton et al. , 2002aLinton et al. , 2002bLinton et al. , 2002cLinton et al. , 2003Linton et al. , 2005Sedaghat et al. 2003aSedaghat et al. , 2003bNicolescu et al. 2004;Kampen 2005aKampen , 2005b.

Materials and methods
Mosquitoes were collected in 10 prefectures of Greece, namely Evros, Rodopi, Serres, and Xanthi in the northeast, Ioannina and Florina in the northwest, Attiki, Fthiotida, and Magnesia in the heart of the country, and Lakonia in the south, from June 1996 to August 2001 (Table I). Ioannina borders Albania, Florina borders the Former Yugoslav Republic of Macedonia and Albania, and Rodopi, Serres, and Xanthi border Bulgaria. Evros is unique in that it shares borders with both Bulgaria in the northwest and Turkey in the east. Collection data, along with the numbers of mosquitoes collected and sequenced from each collection site are given in Table I. Larval collections were carried out in all prefectures, except Florina and Attiki; resting adults were collected in Attiki, Florina, Salino (Xanthi), and Monastiraki (Evros), and one specimen was captured on human bait (Table I). Specimens were collected from a light trap in Didymoticho (Evros) and Roditsa (Fthiotida) ( Table I).
In the case of resting collections, captured females were held for 2 days before being induced to lay eggs, which were then reared as progeny broods to obtain adults with associated larval and pupal exuviae. At least 10 eggs from each brood were stored in Bouin's solution (BDH, Poole, UK) for future study. Link-reared mosquitoes from progeny broods and larval collections serve as voucher specimens for this work, and are retained in the mosquito collections of The Natural History Museum, London (BMNH), the Benaki Museum, Athens, and The National School of Public Health, Athens. For the purposes of this study, and to reflect the relative species composition in each area, the DNA sequence of only one individual per progeny brood was used in the data analysis.
DNA was extracted from 276 mosquitoes following the phenol-chloroform extraction detailed by Linton et al. (2001a). Amplification of the ITS2 nuclear ribosomal spacer was carried out using the 5.8SF and 28SR primers listed by Collins and Paskewitz (1996), using the PCR amplification conditions outlined by Linton et al. (2001a). Following the manufacturer's instructions, products were cleaned using the QIAgen PCR purification kit (QIAgen, Sussex, UK). Purified products were diluted to 1 ng ml 21 per 200 bp of product prior to undergoing the cycle sequencing reaction, using the Big Dye Terminator Kit (PE Applied Biosystems, Warrington, UK). Sequences were obtained in both directions. An ABI 377 automated sequencer (PE Applied Biosystems) was used to read the sequences, and the data were edited and aligned using Sequencher TM version 3.1.1 (Genes Codes Corporation, Ann Arbor, MI, USA) and CLUSTAL X (Thompson et al. 1997). The FASTA search engine (http://www.ebi.ac.uk/fasta33/) was used to assess the similarity of sequences with those in GenBank. Sequence analyses were conducted using MEGA version 2.1 (Kumar et al. 2001). Following sequencing, the template DNA was dried and retained at 270uC in the mosquito DNA bank of the Molecular Systematics Laboratory, Department of Entomology, Natural History Museum for future reference.
Sequences for ITS2 were used for species identification based on comparisons with ITS2 sequences available in GenBank. Restriction enzyme cutting sites were located in the ITS2 sequences to facilitate rapid species diagnosis of members of the Maculipennis Complex in future studies.

Results
DNA was extracted from 276 specimens identified as An. maculipennis s.l. on the basis of morphology. Sequences for the nuclear ITS2 region were obtained from 257 of these (93.1%) ( Table I). The sequences generated in this study are available in GenBank under the accession numbers listed in Table II.

Bionomics and species distribution
Species identification and exact locality data of specimens collected in Greece are given in Table I, along with numbers of specimens for which DNA was extracted and sequenced. Relative species composition in relation to specimens sequenced per prefecture is shown in Figure 1. In this study, An. maculipennis was the most abundant species, followed by An. sacharovi, with the latter species having the widest distribution, occurring along the eastern seaboard, from northern to southern Greece.
Anopheles maculipennis was identified among specimens from Ioannina (70), and Florina (six) in the northwest, Xanthi (23), Rodopi (11), Serres (two), and Evros (34) in the northeast, and in the southern prefecture of Lakonia (one). Anopheles sacharovi was collected in highest densities in the northeastern prefecture of Xanthi (64), where it comprised 71.9% of the samples analysed. It was also found among samples from Evros (seven) and the central and southern prefectures of Attiki (one), Magnesia (one), Fthiotida  (eight), and Lakonia (three). Anopheles melanoon was prominent in the northeastern prefectures, being collected in highest densities in Evros (18), but samples were also found in Rodopi (two) and Xanthi (two). A single specimen of An. melanoon was also found in Ioannina, where it represented only 1.4% of An. maculipennis s.l. captured there. The presence of An. messeae in northwestern Greece was confirmed by the detection of two specimens from Lake Petron and Lake Zazari in the Florina Prefecture, representing a new distribution record for the country (Linton et al. 2001b(Linton et al. , 2002a. Members of the An. maculipennis complex were rarely found in biotic sympatry. However, larvae of An. maculipennis and An. melanoon were found together in Evros (River Tis Mantheas, Itea and River Erithropotomas, Didymoticho) and in Rodopi (Nesti-Krovilli, Maronia and Loutros village). Anopheles maculipennis and An. sacharovi were found in the same larval habitat at Paralia Mandras, northern Kessani in the Xanthi Prefecture. Highest species diversity was recorded from resting collections made in sheep and goat stables in Monistiraki, Evros Prefecture and Salino, Xanthi Prefecture, where An. maculipennis, An. sacharovi, and An. melanoon were collected together as blood-fed adults. Adults of An. maculipennis and An. messeae were also found resting together in sheepfolds near Lake Zazari, Florina Prefecture (Linton et al. 2001b).

Intraspecific variation in the ITS2 sequences
Sequences representing 149 An. maculipennis, 23 An. melanoon, two An. messeae, and 83 An. sacharovi produced an alignment of 496 bases (Figure 2). No intraspecific variation was found either in the length or composition of the sequences in any of the four species (Figure 2). Inclusive of primers (43 bp), the size of the PCR fragment varied according to species, being 472 bp in An. maculipennis, 482 bp in An. melanoon, 485 bp in An. messeae, and 494 bp in An. sacharovi. Interspecific length differences were accounted for in full by length variability in the ITS2 spacer region, not in the 5.8S or 28S flanking genes ( Figure 2 (Porter and Collins 1991) and Quadrimaculatus Complexes (Cornel et al. 1996), and the Palaearctic Maculipennis Complex (Marinucci et al. 1999;Proft et al. 1999;Linton et al. 2001bLinton et al. , 2002aLinton et al. , 2002bLinton et al. , 2002cLinton et al. , 2003Linton et al. , 2005Sedaghat et al. 2003aSedaghat et al. , 2003bKampen 2005aKampen , 2005b.

Interspecific variability in ITS2 sequences
Levels of sequence divergence for the entire fragments of the four Greek species of the Maculipennis Complex are shown in Table III. Over the whole length of the amplicon, excluding gaps, An. messeae and An. maculipennis were shown to be most similar with only 3.3% sequence divergence. Sequence divergence between An. maculipennis and An. melanoon, and An. messeae and An. melanoon were similar, at 3.7 and 3.8%, respectively (Table III). The sequence of An. sacharovi was most distant from all other taxa, showing 13.9 and 14.0% divergence from An. messeae and An. melanoon, respectively, and 15.4% from An. maculipennis (Table III). This is indicative of the basal position of An. sacharovi before the more recent speciation of An. maculipennis, An. messeae, and An. melanoon (White 1978). Indeed, An. sacharovi is the only member of the Maculipennis Complex that can reliably be differentiated from other members of the Maculipennis Complex on the basis of adult morphology alone (Shahgudian 1960;White 1978;Sedaghat et al. 2003a).

Discussion
Concerted evolution of multigene families (Zimmer et al. 1980), such as ITS2 rDNA, within species has resulted in the wide utilization of these fast evolving spacer regions to determine the species composition of Anopheles mosquito complexes (Porter and Collins 1991;Fritz et al. 1994;Severini et al. 1996;Xu and Qu 1997;Beebe et al. 1999;Marelli et al. 1999;Torres et al. 2000;Hackett et al. 2000;Sharpe et al. 2000;Mukabayire et al. 2001;Linton et al. 2001bLinton et al. , 2002aLinton et al. , 2002bLinton et al. , 2002cLinton et al. , 2003. The low level of intraspecific variation in the ITS2 region has proven useful as the basis for both species-specific PCR assays (Scott et al. 1993;Crabtree et al. 1995;Cornel et al. 1996;Proft et al. 1999) and RFLP assays (Van Bortel et al. 2000;Manonmani et al. 2001) to differentiate members of mosquito species complexes. Previous studies of ITS2 sequences have shown no intraspecific variation in sibling species of Anopheles, including species A of the Dirus Complex (Xu and Qu 1997), members of the Maculatus Complex (Torres et al. 2000), and the Sundaicus Complex (Linton et al. 2001a) of Anopheles (Cellia). Intraspecific variation has been reported to be negligible in An. (C.) funestus Giles (Mukabayire et al. 1999), species D of the Dirus Complex (Xu and Qu 1997), and in genotype C of the Bancroftii Group (Beebe et al. 2001) of Anopheles (Anopheles). Intraspecific variation in five colony strains of species of the Gambiae Complex was reported to range between 0 and 0.43% (Paskewitz et al. 1993). Contrary to all other studies, Beebe et al. (2001) reported population-specific ITS2 sequences in the Bancroftii Group between Queensland, Australia and the western province of Papua New Guinea, but it remains to be seen  (496) whether unique haplotypes may indeed represent incipient species. No intraspecific variation was noted in this or previous studies of the ITS2 sequences of An. atroparvus (Linton et al. 2002c), An. maculipennis (Linton et al. 2001b(Linton et al. , 2002a(Linton et al. , 2003, An. melanoon (Linton et al. 2002b), An. messeae (Linton et al. 2001b(Linton et al. , 2002a(Linton et al. , 2002c, or An. sacharovi (Sedaghat et al. 2003a). Analysis of the ITS2 sequences of An. freeborni Aitken and An. hermsi Barr and Guptavnij, two Nearctic members of the Maculipennis Group, showed intraspecific variability to be negligible (Porter and Collins 1991). The multiple copies per genome and absolute intraspecific homogeneity of the ITS2 region indicate that this region would be well suited to rapid diagnostic assays. Based on the ITS2 sequences generated for the four species in this study, restriction enzyme sites were isolated which yield species-diagnostic patterns. Sequences were screened with 113 commercially available enzymes (Roche Molecular Biochemicals, Sussex, UK) to provide a simple, single enzyme assay that could be used to differentiate An. maculipennis, An. melanoon, An. messeae, and An. sacharovi. Enzyme Hsp92 II (with the cutting site CATG) proved to be the only enzyme capable of distinguishing these four species (Figure 3). Recent studies in our laboratory have shown that ample products can be obtained when a single leg of insects, captured and dried directly over silica gel, is placed directly in the ITS2 PCR (as described for extracted DNA: see ''Materials and methods''). The PCR products were found to be sufficiently clean to be used in subsequent RFLP assays. This method provides a cheap, rapid means of identifying the four species in Greece.
In this study, ITS2 sequences generated from 257 wild-caught specimens of An. maculipennis s.l. from 10 prefectures across Greece were used to identity the mosquitoes by comparison with those sequences available in GenBank. Samples were identified as An. maculipennis (58.0%), An. sacharovi (32.3%), An. melanoon (8.9%), and An. messeae (0.8%). These species distributions, with the addition of the new country record for An. messeae (Linton et al. 2001b(Linton et al. , 2002a, agreed with the species list of White (1978). The predicted presence of An. atroparvus (Dahl and White 1978;White 1978) was not confirmed, despite extensive collections, suggesting that this species is not present in Greece. Anopheles daciae, the cryptic sibling species of An. messeae, was also not detected.
All four species of the Maculipennis Complex found in Greece, An. maculipennis, An. melanoon, An. messeae, and An. sacharovi, have been implicated in malaria transmission, but An. sacharovi is the most efficient vector. An. sacharovi is the primary vector in endemic zones in Turkey (Kasap 1990;Alten et al. 2000), Syria (Berberian 1946;Abdel-Malek 1958), Armenia (Romi et al. 2002), and elsewhere in the Middle East (Christophers 1920;Manouchehri et al. 1974;Zahar 1974). It is also regarded as the principal potential vector in regions of southern Europe (Romi 1999;Ramsdale and Snow 2000). In this study, An. sacharovi was found on the eastern coastal seaboard of Greece, with highest densities in Xanthi and Evros Prefectures in northeastern Greece, near the border of Turkey. The species was also found in the drier central and southern prefectures of Attiki, Fthiotida, Magnisia, and Lakonia ( Figure 1). Low densities of the species in this region may reflect the active insecticide-spraying campaigns in these regions or our restricted sampling in these areas.
Anopheles maculipennis, An. melanoon (as An. subalpinus), and An. messeae have recently been incriminated as primary or secondary vectors in restricted geographical regions (Bruce-Chwatt and de Zulueta 1980;Nikolaeva 1996;Alten et al. 2000). Anopheles maculipennis and An. melanoon are found predominantly in the north of Greece along with An. sacharovi, and An. messeae has only been found in the northwestern prefecture of Florina. Currently, An. maculipennis and An. melanoon (as An. subalpinus) are not regarded as efficient vectors of malaria, and, in the Biga Plains of Turkey (Alten et al. 2000), vectorial competence is only noted where these species are present in high densities. Indeed in Portugal, An. maculipennis is known to be predominately zoophilic (Ribeiro et al. 1988), but in certain situations, i.e. at high densities, or where non-human hosts are limited, the species will be anthropophagic and in these cases exhibits malarial vector capacity (sensu Garrett-Jones 1964). The distributions of the highest mosquito population densities were found in the northeastern corner of Greece, comprising Xanthi, Rodopi, and Evros Prefectures, where An. sacharovi, An. maculipennis, and An. melanoon are often sympatric. This region encompasses the delta of the Evros and Nestos Rivers, and is surrounded by the borders of Turkey and Bulgaria. With the endemic status of malaria in Turkey (Kasap 1990;Alten et al. 2000) and the economic difficulties experienced in Bulgaria at present, it seems increasingly important that mosquito control programmes are established and maintained in these northeastern prefectures of Greece. Given the vector competence and population density of the mosquitoes, and the passage of potentially infected persons through this border zone, mosquito control is needed to reduce the risk of malaria being reintroduced and re-established in Greece.