Morphometrics Studies on Females Anopheles arabiensis Patton (Diptera: Culicidae) from Kassala State, Eastern Sudan

Anopheles arabiensis Patton (1905) is the most important malaria vector in Sudan. A morphometric analysis was carried out to characterize the morphology of adult females An. arabiensis and to test the existence of morphological variations among local populations of females An. arabiensis found in eastern Sudan. Adult females An. arabiensis were collected from four sites in Kassala State, eastern Sudan (Kassala, New Halfa, Aroma and Wager). In addition, An. arabiensis specimens, obtained from Sennar laboratory colony were also used in the study. Collection of females An. arabiensis mosquitoes was done by hand capture during the rainy season 2008. Thirty seven morphological characters were examined on samples representing each of the collection sites. One Way ANOVA test showed significant differences in most of the measured characters on females An. arabiensis. Principal component analysis showed that the populations studied differed mainly in the body size of mosquito. A discriminant function analysis was used, based on the new variables generated by principal component analysis, to select an aggregate of morphological characters which collectively could differentiate local populations of females An. arabiensis and to assess the reliability associated with multivariate statistics. Using the body size measurements, the analysis revealed that geographical clustering of field collected females An. arabiensis populations were not clear and the body size characters had little discrimination values. The cluster analysis summarized the phylogenetic relationships among the different populations of An. arabiensis according to their mean body sizes. Laboratory colony samples had a smaller body size compared to the field collected ones. The morphometric analysis confirmed the existence of some geographical variations in the mosquito body size among local populations of An. arabiensis in eastern Sudan.


Background
In Sudan, malaria is a major health problem heading the list of the diseases afflicting the country. Of the seven recognized species of the Anopheles gambiae complex, An. arabiensis Patton is the most abundant and most important vector of human malaria in Sudan. It has been regarded from many localities in Sudan, from the extreme south up to the northern borders with Egypt. It was reported from many localities in Kassala State, eastern Sudan by (Lewis, 1959;Haridi, 1972;Petrarca et al., 1986;Himeidan et al., 2004;Hamza et al., 2005Hamza et al., , 2014. The morphological description of this mosquito has been given by Evans (1938), cited in Gillies and De-Meillon (1968) and Gillies and Coetzee (1987). Morphological characters appear to be in many cases a very useful tool for anopheline taxonomy.
Morphometrics has proved useful in studying some species complexes, when used in conjunction with statistical analysis (Coluzzi, 1964;White, 1977;Lambert and Coetzee, 1982;Service, 1988). Morphometrics concerned with variations and changes in the body form of an organism. It transforms the complex body forms into quantitative series of numbers that can be analyzed and used for comparisons of different forms (Daly, 1985).
Since the discovery that An. gambiae is not a single species but a complex of cryptic species, many workers attempted to find morphological differences between the members of the complex including An. arabiensis (Coluzzi, 1964;Ramsdale and Leport, 1967). Subsequently, many workers have attempted to find morphological variations between the three fresh water species-An. arabiensis, An. gambiae and An.
quadriannulatus (Ismail and Hammoud, 1968;Zahar et al., 1970;Clarke, 1971;White and Muniss, 1972) without success. Petrarca et al. (1998) carried out a morphometric analysis on An. arabiensis and An. gambiae from different sites of Sub-Saharan Africa and he found that all the measures were significantly larger for An. arabiensis. Ribeiro et al. (1979) described An. quadriannulatus davidsoni, a member of the An. gambiae group based on morphological criteria only. Schmidt et al. (2003) proposed taxonomic identification keys for the members of the An. punctulatus complex based on morphological features of the adults and immature stages. The separation between the species of the An. gambiae complex is much less reliable owing to the existence of a considerable overlap between the different character distributions (Bryan, 1980;Bushrod, 1981;Coetzee, 1989).
The present study was conducted to characterize the morphology of adult females An. arabiensis found in eastern Sudan and to investigate the existence of morphometric variations among local populations of this important malaria vector. Relatively few studies have addressed this important vector species in eastern Sudan. However, field based studies such as that have much to add to our understanding of malaria transmission in the area. The results of the study can help in the correct identification of this malaria vector. Correct identification of malaria vectors is essential for the correct evaluation of malaria vector ecology studies and control programs in the area. The results of the morphological measurements of An. arabiensis revealed in this study can be used with the published morphological data of the members of An. gambiae for proposing morphological identification key for the complex.  (Figure 1) under arid and semi-arid climate with rainfall of varying intensity and duration. The climate in the state is characterized by three seasons, the short rainy season (July-October), cool dry season (November-February) and hot dry season (March-June). The topography of the state is characterized by high-lands to the East and South and flat plains to the West and North. Atbara and Gash rivers represent the most important water courses in the state. The Atbara River is flowing out of Ethiopia and it is the largest seasonal river in the state, and the only one which has a discharge during the greatest part of the year (July-December). The Gash River originates in Eritrean high-lands and flows across the flat plain from late June to October. About 10 Km North of Kassala town, the Gash River fans out in an inland delta, the Gash Delta.
Kassala town (15°: 28˝ N & 36°: 24˝ E) is located in an area characterized by the presence of the back-bones of mountains surrounding the eastern part in addition to presence of the horticultural farms of vegetables and fruits along the Gash river banks. New Halfa town (15°: 19 ˝N & 35°: 36˝ E) is located in the western bank of the Atbara River in the middle of New Halfa an agricultural scheme. Aroma (15°: 55˝ N & 36°: 11˝E) and Wagar (16°: 09˝ N & 36°: 12˝E) localities are located in flat area with a moderate vegetation cover in the northern part of Gash Delta. The major consideration in the selection of the study sites is that they represent the different environmental conditions of the state. The selection is also based on the easy accessibility of the collection sites.

Samples collection and preservation
In door resting wild females Anopheles mosquitoes were caught from rooms by hand capture (WHO, 1975) during the rainy season 2008. Collected mosquitoes were fixed alive in the field in 70% ethanol and preserved individually at -20 °C for subsequent processing in the laboratory. The processing of the materials for this study was performed at the Department of Zoology, Faculty of Science, and University of Khartoum, Sudan.

Identification and mounting of females An. arabiensis
Members of the An. gambiae complex were sorted out of other anopheline mosquitoes using morphological identification keys described by Gillies and De-Mellion (1968) and Gillies and Coetzee (1987) with the aid of the dissecting microscope. Subsequently, specimens were dissected out, the head, wings and legs were carefully detached from the body and mounted individually in Puri's mounting medium on glass microscopic slides as described by WHO (1975) with minor modifications. Mosquito body parts were mounted directly into Puri's mounting medium without prior clearing because gum-chloral mountant continues to clear samples after mounting. The remaining of the dissected parts were preserved in 70% alcohol and stored at -20 °C for molecular identification. Females An. arabiensis were identified by molecular polymerase chain reaction (PCR) techniques using the protocol developed by Scott et al. (1993). For morphometric analysis, the permanent slides were used for detailed examination of the external body structures of An. arabiensis.

Morphological characters measurements
For terminology and morphological characters of palps, antennae, wings (appendix 1) and legs (appendix 2) the nomenclature adopted by Evans (1938) cited in Gillies and De-Mellion (1968) were followed, except for wing spots, the nomenclature of Gillies and Coetzee (1987) was followed.
First, a preliminary list was prepared from all morphological characters of the adult female An.
arabiensis. Out of this list, all characters that fixation could not have changed were selected. Thirty seven metric characters were measured on 270 females An. arabiensis represented all the collection sites. The characters consisted of one character associated with antennae, 10 characters with palps, 14 characters with wings and 12 characters of the legs (Table 1).
Morphological measurements were done by the projection method adopted by (Zahar et al., 1970). Selected characters were measured using a Wild MII binocular calibrated compound microscope fitted with and 1.25X phototube. The slide mounted specimens were projected with a phototube on a microscope field using 3.5X objective and 8X eyepiece. After an excellent view of the projected image was obtained, the projected image was drawn and subsequently measured to the nearest half millimeter. Then the measures obtained were calibrated to the real lengths using a micrometer stage (1mm-Erma, Tokxc).

Data analysis
All statistical analysis of the morphometric data were conducted using SPSS Version 16.0 for Windows. The morphological measurements were not transformed to ratios so as to preserve the possible influence of differences in the body sizes of the species.
The measured data were subjected to univariate and multivariate statistical analysis. Univariate statistics involved calculation of descriptions, One-Way ANOVA test and Box -and whisker plots. Descriptions involved calculation of mean size measurements plus or minus their standard error and sample sizes. A One-Way ANOVA test was carried out to test the significant differences of the measured characters among different populations of An. arabiensis. Box -and whisker plots were used to check for the presence of outliers and to summarize the univariate data.
Multivariate statistics such as, principal component analysis, discriminant function analysis and cluster analysis were used.
Principal component analysis is a method of data reduction and it was used to simplify subsequent analysis of the morphological data of field specimens. The analysis aims to produce a small number of derived variables that explain most of the variance and can be used in place of the large numbers of original variables. It obtains a set of principal components by weighting all the available variables. The first component explains the most variation; the second explains the next most variation, and so on. Investigation of the first few components will show which variables contribute most to the variations between individuals (Dythan, 2003;Pimentel, 1992

Discriminant function analysis
The analysis was applied on the characters linked with the mosquito body size to discriminate between the four field populations of An. Arabiensis (Figure 4). These characters comprised the eight most influential characters for component 1, derived by the principal component analysis ( Table 2). The characters used were: mid leg length, hind leg length, wing length excluding the fringe, wing width excluding the fringe, fore leg length, length of palpal segment III, antenna flagellum length and length of palpal segment IV.
When these characters were subjected to discriminant function analysis, three significant functions were derived.  (Table 4). However, this discrimination was not a complete one, since only 37.9% of the original grouped specimens, and 64.5% of the cross-validated ones were correctly classified (Table 5).

Cluster analysis
Cluster analysis based on a matrix of distance for 37 morphological characters (Table 1) and complete linkage between groups was used to produce hierarchical clusters of the field and colony populations in a distance dendogram ( Figure 5).
The cluster shows that field samples of An. arabiensis clustered together, with 2 subclusters, one containing population Kassala and Aroma and the other containing population New Halfa and Wagar. The laboratory colony specimens clustered alone and well apart because of their smaller body sizes. The dendogram confirms the results obtained by principal component analysis and showed that the clustering of all the populations based mainly on the measurements of the mosquito body size.

Discussion
Measurements of various morphological characters of females An. arabiensis from Sudan were in accordance with the published data of the same species from other sites along the distribution range of the species (Coluzzi, 1964;White et al.;1972;Petrarca et al., 1998;Adeieke et al., 2008).
8   Principal component analysis showed that the populations studied differed mainly in the mosquito body size. Kassala and Aroma populations had bigger body size followed by New Halfa. Wagar population had the smallest body sizes. The results confirmed the report of (Dythan, 2003) on the output of principal component analysis. In this morphometric analysis, it is usually the case that individual specimens will vary in size. The first principal component will always account for size and it is often employed as a method for removing size from the analysis. The second and the rest of the principal components are usually interpreted as shape component (Dythan, 2003). Unlike the first component, which is positively contributed to all dimensions, the second principal component as well as other components are positively correlated to some variables, and negatively correlated to others.
Based on the results of the morphometrics analysis, the geographical features characterizing the different collection sites may influence the microclimatic conditions of the area which had a direct effect on the mosquito body sizes. The results drawn from this study indicate the presence of some geographical variations among local An. arabiensis populations. This suggests the presence of some degree of biogeographical structuring within the species range which may be correlated with the environmental conditions persisting in these localities. Wasserberg et al. (2003) mentioned that difference in morphometrics measurements may be attributed to the effect of microclimatic conditions within the microhabitat niche of the species that affect the species abundance, adult life span and larval development. Thompson (1970) and Petrarca et al. (1998) reported obvious variations on the mosquito body sizes among field collected females An. arabiensis and An. gambiae. They stated that the larger body size of females An. arabiensis could be caused by the cooler climate of the collection sites in comparison to the other warm collection sites. Moreover, the influence of temperature on the body size (expressed as an inverse, temperature-size correlation) has been repeatedly reported in mosquitoes as An. merus (Ie Sueur et al., 1992) and An. crucians (Hu et al., 1993).
Results of various biological data on adult mosquitoes over the years had revealed differences in characteristics of some species depending on the ecotypes and the season (Lehane, 1991). The host seeking behavior and vectorial capacity of some members of An. gambiae differ significantly in different geographical zones of the world (Adeleke, 2008).
A discriminant function analysis was carried out to select morphological characters which could differentiate local populations of females An. arabiensis. Using the body size measurements, the analysis revealed that geographical clustering of field females An. arabiensis populations was not clear and the body size characters had little discrimination values.
Cluster analysis was carried out using the complete set of measured characters which were subjected to principal component analysis. The analysis summarized the relationships among all the populations of An. arabiensis according to their mean body sizes. As shown in the cluster analysis results, An. arabiensis individuals obtained from Sennar laboratory colony became an out group and well apart because of their smaller body sizes. Kassala and Aroma populations appeared as sisters group because of their larger body sizes, while New Halfa and Wagar populations clustered together.
An. arabiensis females obtained from colony materials had a smaller mean body size compared to the field collected ones. This noticeable decrease in the body size of the laboratory specimens may be a result of stressful environmental influences, such as larval crowding or high mean temperature or both, which can lead to a decrease of the body size (Bock and Milby, 1981). Moreover, Coluzzi (1964) investigated the morphological characters of members of An. gambiae complex obtained from colony materials. He stated that, there may have been changes in the mosquito phenotypes as a result of forced gene exchange or because of the selection pressure by artificial conditions. Munstermann (1994) and Petrarca et al. (1998) mentioned that the colonization process can modify the organism phenotype, which can make conclusions drawn from observations on laboratory populations' unreliable.
The results of the study can aid in the description of detailed morphology of adult female An. arabiensis. The morphometric analysis confirmed the existence of some geographical variations in the mosquito body size among local populations of An. arabiensis in eastern Sudan. These variations may be effected with the environmental conditions characterized the localities.