SEARCH

SEARCH BY CITATION

Keywords:

  • Anopheles gambiae;
  • larval ecology;
  • mosquito larval habitat;
  • malaria vector;
  • mosquito control;
  • Kenya

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

Control of aquatic-stage Anopheles is one of the oldest and most historically successful interventions to prevent malaria, but it has seen little application in Africa. Consequently, the ecology of immature afrotropical Anopheles has received insufficient attention. We therefore examined the population dynamics of African anopheline and culicine mosquitoes using operationally practicable techniques to examine the relative importance and availability of different larval habitats in an area of perennial malaria transmission in preparation for a pilot-scale larval control programme. The study was conducted in Mbita, a rural town on the shores of Lake Victoria in Western Kenya, over 20 months. Weekly larval surveys were conducted to identify the availability of stagnant water, habitat characteristics and larval densities. Adult mosquitoes were collected indoors at fortnightly intervals. Availability of aquatic habitats and abundance of mosquito larvae were directly correlated with rainfall. Adult mosquito densities followed similar patterns but with a time-lag of approximately 1 month. About 70% of all available habitats were man-made, half of them representing cement-lined pits. On average, 67% of all aquatic habitats on a given sampling date were colonized by Anopheles larvae, of which all identified morphologically were A. gambiae sensu lato. Natural and artificial habitats were equally productive over the study period and larval densities were positively correlated with presence of tufts of low vegetation and negatively with non-matted algal content. The permanence of a habitat had no significant influence on larval productivity. We conclude that A. gambiae is broadly distributed across a variety of habitat types, regardless of permanence. All potential breeding sites need to be considered as sources of malaria risk at any time of the year and exhaustively targeted in any larval control intervention.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

In Suba District, western Kenya, malaria transmission is meso- to holoendemic, with an average parasite prevalence in the human population of 60% (Gouagna et al. 2004). Malaria in this area is associated with ideal climatic conditions for transmission (Bayoh & Lindsay 2003) and the most efficient vector species in the world (Beier et al. 1999; Hay et al. 2000): Anopheles gambiae Giles, A. arabiensis Patton and A. funestus Giles (Mutero et al. 1998; Minakawa et al. 1999, 2002). National malaria control efforts in Kenya, as in most African countries, focus on case management and the use of insecticide-treated bednets (MOH 2003). Domestic vector control interventions against adult mosquitoes in the form of insecticide-treated bednets (Kazura 2003) or indoor residual spraying (Kouznetsov 1977), combined with improved access to effective diagnosis and treatment, have enormous potential to Roll Back Malaria and constitute the primary means to achieve the Abuja targets (WHO/UNICEF 2003). However, even these highly effective interventions are insufficient to eliminate malaria transmission from most endemic parts of Africa (Killeen et al. 2000b; Gu et al. 2003). Additional interventions are clearly required to build truly integrated malaria control programmes that can make further inroads to the enormous burden of malaria in African communities (Beier et al. 1999; Killeen et al. 2000a,b; Shiff 2002). The potential of integrated vector control for malaria prevention in Africa is now being reappraised (Castro et al. 2004; Keiser et al. 2004; Killeen et al. 2004) including underused interventions such as routine application of larvicides (Fillinger et al. 2003) and source reduction through removal of Anopheles larval habitats. Such approaches have received little attention in recent decades despite proven successes in diverse settings (Watson 1911; Soper & Wilson 1943; Muirhead-Thomson 1945; Shousha & Pasha 1948; Watson 1953; Killeen et al. 2002a). Since the advent of dichlorodiphenyltrichloroethane (DDT), most efforts on vector control have focused on mosquito adults and research on the larval ecology of African malaria vectors has been largely neglected. As a consequence, little is known about the habitats, abundance and distribution of the larvae of the main African malaria vectors, with most information in relation to practical larval control dating back to publications in the first half of the previous century (Hopkins 1940; Muirhead-Thomson 1945, 1951; Watson 1953; Holstein 1954; Clyde 1967).

To control mosquitoes, whether adults or larvae, it is crucial to understand the relevant ecology of the target species. This requires the study of not only the fluctuations of the adult populations, but also the factors affecting larval abundance and distribution. While considerable progress towards understanding the aquatic stages of A. gambiae sensu lato has been made in recent years (Chinery 1984; Minakawa et al. 1999; Gimnig et al. 2001, 2002; Bøgh et al. 2003; Keating et al. 2003; Klinkenberg et al. 2003; Koenraadt et al. 2003; Koenraadt & Takken 2003; Ye-Ebiyo et al. 2003), systematic research on the larval ecology of the main malaria vectors in Africa is still limited and often represents short data collection periods, frequently considering mosquito-infested habitats only. In contrast, more intensive larval ecology studies evaluating and acknowledging options for larval vector control have been conducted in central and southern America on various malaria vectors in the last decade (Berti et al. 1993; Rodriguez et al. 1993; Fernandez-Salas et al. 1994; Manguin et al. 1996a,b; Rejmankova et al. 1999; Grillet 2000).

We report our observations on the population dynamics of African anophelines and associated culicine mosquitoes in order to evaluate the importance and availability of different larval habitats in an area of moderate and perennial malaria transmission. This study was a baseline survey to inform an operational larval control programme. Thus, the methods applied are those that are considered practical and useful under operational conditions in health programmes and generalizeable to larger, sustainable programmatic scales (Gubler & Clark 1994; Habicht et al. 1999; Durrheim et al. 2002; Booman et al. 2003). This survey was conducted intensively over a period of 20 months from October 2000 to May 2002. It is the first detailed longitudinal study describing the nature of the aquatic habitats of malaria vectors in Africa.

Study area

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

The study was carried out in the area surrounding the International Centre of Insect Physiology and Ecology's (ICIPE) Mbita Point Research and Training Centre in Mbita (0°30′ S; 34°15′ E), a rural town on the shores of Lake Victoria, in Suba District, western Kenya. About 8000 people live in Mbita and most are resident fishermen with a few traditional farmers. Mbita is a peninsula bordered by the lake from the east and west and consists of several small housing groups. In the north Rusinga Island is connected with the mainland via a causeway. The data presented here were collected in an area of approximately 1.5 km2, along the eastern shores of the lake and in an inland area known as ‘Hillside’ because of its closeness to a small central hill (Figure 1). During the study period, mosquito larval and adult surveys were conducted and the relative availability, changes and colonization of aquatic habitats recorded.

image

Figure 1. Map of the study area, Mbita town, Suba district, western Kenya. Black triangles (bsl00066) indicate sentinel houses for mosquito resting catches, black circles (•) indicate surveyed larval habitats with semipermanent character, grey areas indicate larval habitats with temporary character, transparent squares indicate potential habitats outside the investigation area. The investigation area is enclosed between dotted lines and main roads.

Download figure to PowerPoint

The ICIPE operates a meteorological station where daily climate data are recorded. Mbita's average annual rainfall between 2000 and 2002 was 1423 mm, and temperatures varied between an average minimum of 18 and maximum of 29 °C. Two rainy seasons occur annually from approximately March to June and October to November. The peak rainfall is generally expected between March and May, but the seasons are usually not well defined with some years of more or less regular rains and others with prolonged dry periods.

Total annual rainfall in the year 2000 was 1597 mm, with more than a third of this falling in the last 3 months of the year. The year 2001 was comparatively dry with an annual rainfall of 1139 mm, and only 438 mm during the long rains (March–May). This rainy season was thus much drier than the one in 2000 with 683 mm, and the one in 2002 with 513 mm from March to May.

Mosquito adult survey

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

Adult mosquitoes were collected indoors (resting catches) from 12 houses in close vicinity to the surveyed larval habitats. The adult survey was conducted at fortnightly intervals. All selected houses were traditional mud huts with iron roofing representing the most common houses in the area. The distance of the surveyed houses to the nearest, surveyed breeding habitat varied between 15 and 91 m. Every sampling day, the number of people that slept in the house the previous night was recorded.

Indoor-resting mosquito adults were sampled using the pyrethroid spray catch (PSC) technique following WHO guidelines (WHO 1992). Before spraying inside the hut, large furniture was carefully removed and white sheets placed on the floor. Ten minutes after spraying, all dead mosquitoes found on sheets were collected and transported to the laboratory where A. gambiae s.l. and A. funestus were separated from the other mosquito species using morphological criteria. Because of operational resource allocation priorities, the A. gambiae complex was not identified to species level using the polymerase chain reaction (PCR) methodology. According to earlier investigations in Mbita location (Minakawa et al. 1999, 2002), A. gambiae s.s. accounts for an average of 80–90% of both the larval and adult population of the species complex, with the rest being A. arabiensis. Other, far less abundant Anopheles species identified in the area are A. funestus and A. coustani; especially the latter is only found sporadically (Minakawa et al. 2002) and not at all during our survey.

The number of males and females, and the number of bloodfed and unfed females were recorded for all Anopheles and culicine mosquitoes. Culicine mosquitoes were not identified to species level, but their general inclusion in the study was considered important as they represent a significant nuisance in the area and need to be considered in larval control operations to gain community support.

Relative availability of habitats.

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

During a preliminary survey in September 2000, most aquatic habitats were found at the eastern shores and in the hillside inland area (Figure 1). These areas were therefore selected for detailed study. The study area was examined twice a week for stagnant water bodies, with the presence or absence of the aquatic stages of Anopheles and culicine mosquitoes being noted in each water body. All 21 mosquito-containing habitats identified in the preliminary September 2000 survey were selected for biweekly determination of larval densities from October 2000 to December 2001, and further weekly surveys from January to May 2002. Thirteen of these sites were located between the lake shore and the main road at Uyoga Beach and East Shore, with the remaining eight sites being located inland (farther than 100 m away from the lake shores), in the hillside area enclosed by the two major roads in Mbita (Figure 1). We estimated the proportion of habitats that actually contained water at a given time as an index of relative availability (Killeen et al. 2001). For the purposes of simplicity of terminology, we define habitats as being available when they contain water.

Larval densities and habitat characteristics.

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

Larvae were sampled using a WHO-standard 250 ml capacity mosquito dipper (WHO 1992) in order to (i) score the presence or absence of larvae from (water containing) habitats, (ii) to determine the preferred breeding sites of vector and other mosquito species, in relation to larval densities and continuous colonization of a site and (iii) to measure changes in larval density over time. Sampling was conducted in a deliberately non-random fashion to maximize sensitivity of collections using standard procedures (Bøgh et al. 2003).

To measure the average number of larvae per dip per breeding site per sampling day, constant number of dips (six in sites smaller and 10 in sites larger than 6 m2) were taken. Larvae were classified into three categories: early instars (stages I and II), late instars (stages III and IV) and pupae. Larval instars were classified separately for Anopheles and culicines, but pupae were not distinguished in the field and summarized in the data collection since their species-dependent morphological differences cannot easily be distinguished in the field. After quantification, larvae and pupae were returned to the water.

Anopheles larvae were identified according to morphological criteria in the field. Furthermore, to separate A. gambiae s.l. and A. funestus, samples of five to 10 Anopheles larvae per habitat were identified morphologically in the laboratory in approximately 2-monthly intervals.

To investigate habitat characteristics associated with Anopheles and culicine development, we classified habitats as being

  • in their origin man-made or natural,
  • in their appearance natural (man-made and natural habitats can contain soil and natural vegetation) or an artificial pit structure (cemented), and we recorded
  • the surface area in m2,
  • water depth in cm,
  • turbid or clear water,
  • non-matted algal content (according to the intensity of the green colour of the entire water surface area we categorized four levels: absent, low, medium and high), and
  • vegetation present or absent (recorded vegetation included grass, Eichhornia crassipes, Lemna sp. and algal mats).

Statistical methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

General linear models (GLM) have been applied to analyse statistical differences between treatment groups in Table 1 and Figure 3. Proportions were arcsine transformed to normalize the data. Spearman's non-parametric rank correlation of Williams mean values and Mann–Whitney U-tests were used to identify correlations between abundances of aquatic stage mosquitoes or between habitat characteristics and mosquito abundance. All correlation analyses present summary statistics of one value per habitat over the entire study period. The specific tests applied are mentioned in the text.

Table 1.  Larval habitat characteristics in Mbita between October 2000 and May 2002 (20 months, dipping twice a week up to December 2001, once a week, i.e. January–May 2002)
HabitatTotal number of larvae/ dip (sum of 20 months)Proportion of Anopheles (%)Time habitat flooded (%)Proportion of time flooded habitat was colonized (%)Proportion of late instars dipped (%)Proportion of time species colonized habitats together (%)Average number of larvae/dip per sampling date with water
NumberLocation Type*All larvaeAnophelesAnophelesCulicinesAll larvaeAnopheles
  1. * Habitat sizes varied between 2 m2 for habitat no. 9 and 75 m2 for habitat no. 18, water depths of all habitats varied between 1 and 106 cm (average 27.3, SD = 20.1).

1LakesideConcrete441.5209367453022494.200.84
2LakesideNatural315.6686774682111894.152.83
3LakesideConcrete58.386438684282291.191.02
4LakesideNatural73.788327575243222.051.80
5LakesideConcrete23.18361001003454863.302.74
6LakesideConcrete395.3657369663230494.763.09
7LakesideConcrete46.181091641326604.190.35
8LakesideConcrete190.4414283753130653.971.64
9LakesideNatural312.8214180632722786.801.42
10HillsideConcrete401.8569293883315573.862.18
11HillsideConcrete238.1318851362416492.380.75
12HillsideConcrete514.7131973437373514.710.16
13HillsideConcrete422.986599442825425.790.47
14HillsideConcrete576.6198188502525466.341.23
15HillsideConcrete199.1253678591311664.861.22
16HillsideConcrete1196.9175932519332514.080.17
17HillsideConcrete287.4246273473316554.110.99
18LakesideNatural525.4419995862922674.691.92
19LakesideNatural417.53435907325277510.443.57
20LakesideNatural372.71128915332465911.651.25
21LakesideNatural149.148161001001845948.280.70
image

Figure 3. Aquatic habitat availability and colonization by mosquito larvae during study period from October 2000 to May 2002. ○, number of available aquatic habitats in study area per sampling date; bsl00001, number of habitats colonized by any mosquito larvae; inline image, number of habitats colonized by Anopheles larvae.

Download figure to PowerPoint

The relationship between the abundance of immature or adult mosquitoes and rainfall or temperature, as well as with each other, was evaluated by autoregressive time-series analysis fitted by the exact maximum likelihood method. Monthly total rainfall, and 1 month lags thereof, were treated as independent variables in autoregression models which were selected manually beginning with a full set of potential determinants and sequentially removing the least significant variables, including the constant, until only significant variables remained. All analyses were carried out using spss 11 for Windows.

Mosquito adult densities

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

During the 20 months of the study, 42 sampling events took place to collect adult mosquitoes in 12 sentinel houses in the investigation area. Altogether 14 226 mosquitoes were collected, 29% (4182) of them A. gambiae s.l., 1% (153) A. funestus and 70% (9891) culicines of various species. A considerable proportion of males could be found resting inside houses as well. Males represented 50% of all culicines, 32% of all A. gambiae s.l. and 10% of all A. funestus adults. About 92% of all A. gambiae s.l., 75% of all A. funestus and 77% of all culicine females were bloodfed.

The average number of bloodfed Anopheles mosquitoes per house ranged from a high of 17.4 in December 2000 to 0.2 in September 2001. Taking into account the number of persons that slept in the house the night before any adult PSC took place, the number of bloodfed Anopheles per person ranged from 6.5 in December 2000 to 0.1 in September 2001. The entomological inoculation rate (EIR) could not be directly calculated since no data were collected on sporozoite incidence and human biting index. Assuming that all bloodfed Anopheles females fed on humans [there are very few cattle or other alternative hosts in Mbita and we would expect human blood indices to exceed 90% (Killeen et al. 2001)] the night before and taking into consideration a sporozoite incidence of 3.5% as previously identified for all three Anopheles species in a location nearby (Mathenge et al. 2004), we can approximate a relatively modest EIR of 20 infective bites per person per year which is comparable with previous rapid assessments for the township (Shililu et al. 2003) [for standard methods of EIR calculations see Hay et al. (2000), Drakeley et al. (2003)].

Particularly high numbers of Anopheles adults were observed from November 2000 to February 2001 (Figure 2c) following extended rain fall (Figure 2a) after which habitats were available in greatest numbers (Figure 3). After February 2001, bloodfed, female densities per person varied between 1.2 and 0.1 with a peak abundance in July 2001, decrease of densities up to October 2001 and a slow and steady increase since November 2001 up to the end of the study (Figure 2c).

image

Figure 2. The dependence of mosquito abundance upon rainfall from October 2000 to May 2002. (a) Monthly rainfall in mm. (b) Immature abundance: ○, proportion of habitats containing water; bsl00063, Anopheles larvae of all stages; inline image, culicine larvae of all stages; ×, combined Anopheles and culicine pupae. All points represent the monthly Williams mean of mean counts per dip for all samples from surveyed habitats. (c) Biting adult abundance: bsl00063, A. gambiae sensu lato; inline imageA. funestus; inline image culicine species. All points represent the monthly Williams mean of total catches of bloodfed mosquitoes per person.

Download figure to PowerPoint

Dynamics of relative availability of larval habitats

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

Over the study period, 53 stagnant water bodies representing potential mosquito larval habitats were identified, including the 21 habitats where weekly density measures took place. Most of the habitats were man-made (70%). The man-made habitats had either a natural character containing soil, grass or other vegetation (57%) or were cement-lined pits (43%). The latter (Figure 4) are the most common habitats in the entire location of Mbita town, constituting 60–80% of all larval habitats found in the rainy and dry seasons (U. Fillinger, unpublished data). Most of the possible natural habitats identified were extremely ephemeral, not keeping water long enough to allow a complete life cycle of mosquito development. Only during periods of long and heavy rain, as from October 2000 to January 2001 (Figure 2a), did the water level of the lake increase, resulting in a high water table, water saturated soil and presence of natural puddles and pools for several weeks (Figures 3 and 5).

image

Figure 4. A cement-lined pit – the most common larval habitat in Mbita. This artificial habitat is created any time a modern brick house is built in the area. The cement-lined pit will be filled with water that is then used during construction work, for example, to mix cement.

Download figure to PowerPoint

image

Figure 5. Puddles are natural larval habitats that occur in Mbita only temporary in larger numbers after extended rainfall. (a) Puddle in overview. (b) Puddle close up: high density of Anopheles larvae and pupae, freshly emerged adults at the edges.

Download figure to PowerPoint

On average, 89% of all available aquatic habitats on any given sampling date contained mosquito larvae, with an average of 67% being colonized by Anopheles per sampling date (Figure 3). There is evidence of a difference in the occupation of possible habitats between the periods from October 2000 to March 2001 (where the Anopheles colonization is closely associated with the occurrence of potential habitats), and the time after March 2001, when Anopheles colonization followed the occurrence of habitats to a much lesser extent. For example, between October 2000 and March 2001 an average of 83% of all available habitats were colonized by Anopheles larvae compared with an average of 60% after March 2001. The difference between these two time periods was significant (F1,109 = 57.4, P < 0.01).

Larval identification and distribution

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

The characteristics of the 21 surveyed mosquito larval habitats are summarized in Table 1. Anopheles funestus was not found in those habitats at any time. Species within the A. gambiae complex (later referred to as Anopheles) were not further identified and we assumed that A. gambiae s.s. represented the majority of the specimens (Minakawa et al. 1999, 2002). Previous work has shown no differences in A. gambiae s.s. and A. arabiensis habitat preferences in western Kenya, other than proximity to humans (Minakawa et al. 1999, 2002; Gimnig et al. 2001, 2002).

Variation in larval densities over space and time was substantial. Considering only sampling dates when habitats contained water, the average number of Anopheles larvae per dip varied between 0.16 and 3.57 (average 1.44, SD = 0.98). Culicine mosquitoes occurred in higher densities, with their average density being three times that of Anopheles (average 4.55. SD = 4.12). Maximum numbers of Anopheles and culicine larvae dipped in a single habitat at a single sampling date were 38 and 52 larvae per dip, respectively.

The mosquito larval productivity varied substantially between the different sites with a total of 23.1 larvae/dip in habitat no. 5 and 1196.9 larvae/dip in habitat no. 16 for the entire study period, respectively. The proportion of Anopheles larvae in all available habitats varied between 1 and 88%. The proportion of time that single habitats contained water and were therefore available for oviposition varied from 6 to 99% of the entire study period. Anopheles larvae were detected in all habitats, at one time or another, being present at 25–100% of all occasions when habitats contained water. Late larval instars accounted for only 25% of all mosquito larvae dipped, indicating high larval mortality. There was no significant difference between the proportions of late instars of Anopheles and culicine larvae (F1,40 = 0.30, P = 0.58) in the studied habitats. This proportion remained constant for Anopheles and culicine development even when different micro- and macrohabitat types where compared (proportion of late instars in natural vs. cement-lined pit structures: F1,19 = 0.01, P = 0.91, in habitats situated at the lake vs. those at the hillside: F1,19 < 0.01, P = 0.99).

All surveyed habitats were shared by Anopheles and culicine larvae for at least a certain percentage of the time the habitat was available (co-existence in wet habitats 57% of the time). A few of the cement-lined pit structures were characterized by very high culicine densities and few Anopheles larvae (Table 1). In these sites, these species co-occurred less frequently (51% of time) than in natural sites (69% of time, F1,19 = 4.35, P = 0.05).

The area in which habitats were located (e.g. hillside vs. lakeside) appeared to be a more important determinant of vector potential than microhabitat variation. For example, comparing lake and hillside habitats, the proportion of Anopheles larvae differed significantly (47% and 20%, respectively, F1,19 = 5.53, P = 0.03). Also the average number of Anopheles larvae per dip was significantly higher in lakeside habitats (1.8 larvae per dip) than in hillside habitats (0.9 larvae per dip) (F1,19 = 4.86, P = 0.04). Furthermore, Anopheles was found in lakeside habitats more often (present 73% of the time in lakeside habitats vs. 48% in hillside habitats, F1,19 = 10.4, P < 0.01). As a consequence of the apparently lower preference of Anopheles mosquitoes for hillside habitats, Anopheles and culicine larvae co-existed only 47% of the time in contrast to 63% of the time in lakeside habitats (F1,19 = 3.6, P = 0.07).

As all hillside habitats are cemented (Table 1) pits, one could argue that these observations are not spatial effects but dependent on the habitat type. However, a comparison between natural and cemented habitats within the lakeside area showed that the proportion of Anopheles larvae recorded is on average 47% and does not significantly differ between the two habitat types (F1,11 = 0.14, P = 0.72). Furthermore, the average number of Anopheles larvae per dip for all lakeside habitats was 1.8 and no significant difference could be detected between natural sites and cement-lined pits (F1,11 = 0.3, P = 0.6).

Seasonality of mosquito population dynamics

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

The bulk of rainfall occurring in Mbita during the study period was concentrated outside of what is traditionally considered ‘the long rains’ from March to June (Figure 2a). Months with the heaviest total rainfall typically included a handful of days with torrential downpours of >50 mm. The presence of water in the 21 surveyed habitats, as well as the abundance of Anopheles larvae, culicine larvae and combined pupae (see Methods) were directly correlated with rainfall in that particular month (Figure 2b and Table 2) and in the case of culicine larvae and adults of A. gambiae and culicines, to rainfall the previous month (Table 2). The fact that no constant term was significant or included in any of the models indicates that these habitats are more or less directly fed by rainfall through either direct precipitation, runoff or basin recharge, suggesting more stable water sources such as groundwater do not support many of these habitats through dry periods. The constant in these models represents the mosquito densities expected if no rain fell. Thus, the set of larval habitats surveyed seem largely dependent on direct rainfall. The adult mosquito populations are similarly dependent upon rain suggesting that more stable habitats maintained by the groundwater reservoir through the dry season do not contribute substantially to those populations. Interestingly, the total number of aquatic stage mosquitoes were more closely related to rainfall than the proportion of habitats containing water, suggesting that fresh water, and possibly fresh nutrients picked up by runoff, promote proliferation of both A. gambiae s.l. and culicine species. Adult densities of A. gambiae s.l. and culicine species both generally followed a similar pattern but with a clear lag of approximately 1 month (Figure 2c and Table 2). Although not detected in the larval habitats surveyed, A. funestus adults appeared in small numbers during and after the heavy rains of late-2000 and mid-2001. Notably, the heavy rains of mid-2001 stimulated proliferation of both Anopheles and culicine aquatic stages, resulting in a surge of culicine adults but relatively few Anopheles adults (Figure 2).

Table 2.  Time series autoregression analysis of the dependence of mosquito abundance on rainfall
Dependent variabled.f.LLIndependent variablesParameter estimateP-value
  1. Degree of freedom (d.f.), log likelihood (LL) and statistical significance (P-value) are presented.

Aquatic stages
 Anopheles larvae18−16.8Autoregressive0.781 ± 0.144<0.0001
Rainfall that month (mm)0.0042 ± 0.00150.0097
 Culicine larvae17−20.2Autoregressive−0.472 ± 0.2120.0399
Rainfall that month (mm)0.0099 ± 0.0017<0.0001
Rainfall previous month (mm)0.0092 ± 0.00180.0001
 Pupae182.65Autoregressive−0.237 ± 0.2320.3199
Rainfall that month (mm)0.0034 ± 0.0003<0.0001
Adults
 Anopheles gambiae17−40.1Autoregressive0.856 ± 0.116<0.0001
Rainfall that month (mm)0.0112 ± 0.00480.0316
Rainfall previous month (mm)0.0160 ± 0.00480.0042
 Anopheles funestus18−6.4Autoregressive0.380 ± 0.2300.1154
Rainfall that month (mm)0.0022 ± 0.00070.0084
 Culicine17−61.1Autoregressive0.351 ± 0.2430.1676
Rainfall that month (mm)0.0337 ± 0.01360.0236
Rainfall previous month (mm)0.0308 ± 0.01420.0440

Figures 2 and 3 reveal that these peaks of heavy rainfall in late-2000 and early 2001 resulted in the majority of the habitats containing both water and aquatic stage mosquitoes. Non-matted algal content was positively associated with habitat stability in the 21 surveyed habitats and highest after long periods of consistently holding water. Anopheles larvae were more evenly distributed across habitats and time than culicines (Figure 6). While Anopheles were found at varying densities in habitats of all stability levels, culicines were most abundant and commonplace in semipermanent and temporary habitats. Notably, the three most stable habitats contained A. gambiae s.l. larvae throughout the study period, whereas culicine larvae were conspicuously and almost completely absent from these essentially permanent water bodies (Figure 6). When we examined the relationship between the characteristics and productivities of the individual habitats over the duration of the study we found that the overall abundance of pupae was more closely correlated with that of culicine larvae than with Anopheles larvae, in terms of both overall productivity throughout the study period and productivity when water was present (Table 3). This suggests that the bulk of pupae collected were in fact culicines, although these were not distinguished from Anopheles in the field. A negative association between the abundance of Anopheles larvae and those of both culicine larvae and pupae approached significance when the habitats contained water (Table 3), possibly as a result of differential habitat preference as discussed above. However, it might also reflect competition between these species in certain habitats.

image

Figure 6. Permanence, depth, non-matted algal index and productivity of individual habitats over time. The permanence of individual habitats was ranked and plotted according to the proportion of the study period during which they contained water. All productivity values for aquatic stage mosquitoes reflect Williams mean values of monthly mean counts per dip.

Download figure to PowerPoint

Table 3.  Correlations between abundances of aquatic-stage mosquitoes
 Culicine larvaePupae
rP-valuerP-value
  1. * Samples from periods when the habitat were dry considered to be zero.

  2. Correlation coefficients (r) and statistical significance (P-value) are presented for Spearman's non-parametric rank correlation of Williams mean values of mean counts per dip. Bold values represent P < 0.05.

Throughout study period*
 Anopheles larvae0.0050.9820.0760.736
 Culicine larvaeNANA0.869<0.001
Only when water present
 Anopheles larvae−0.3690.091−0.4000.065
 Culicine larvaeNANA0.808<0.001

Correlation of habitat characteristics and mosquito abundance

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

Table 4 gives associations between the abundances of aquatic-stage mosquitoes and characteristics of individual habitats. Mean depth over the study period was positively correlated with the abundance of Anopheles larvae, pupae and, to a lesser extent, culicine larvae (the latter only approached significance). Similarly, the overall productivity of habitats for culicine larvae was positively associated with mean algal content over the entire period. However, depth and algal content had no association with the productivity of all types of immature mosquitoes when only sampling occasions on which the habitat contained water were considered. This indicates that these correlations are artefacts arising from the covariance between depth or algal content and the presence of water.

Table 4.  Correlations between habitat characteristics and aquatic-stage mosquito abundance
 All samples from habitatWhen habitat held water
U† or r*P-valueU† or r*P-value
  1. Correlation coefficients (r) and statistical significance (P-value) are presented for Spearman's non-parametric rank correlation* of Williams mean values of mean counts per dip vs. the mean values for each habitat characteristic for either the entire study period or only when the habitat contained water. The U statistics are presented for Mann–Whitney U-test† of Williams mean values of mean counts per dip vs. habitat type (number of habitats, n = 21). Bold values represent P < 0.05.

Anopheles larvae
 Habitat type†440.581210.026
 Depth*0.4520.035−0.2090.349
 Grass*0.0290.9000.4270.047
 Non-matted algae*−0.0430.8500.5490.008
 Proportion of time containing water*0.5790.005NANA
 Permanence rank*0.5840.004−0.2080.352
Culicine larvae
 Habitat type†510.945420.490
 Depth*0.3790.082−0.2780.210
 Grass*−0.0460.8380.1240.582
 Non-matted algae*0.5750.0120.1170.603
 Proportion of time containing water*0.5720.005NANA
 Permanence rank*0.5740.005−0.2670.230
Pupae
 Habitat type†470.731420.490
 Depth*0.4970.018−0.0160.942
 Grass*−0.0300.894−0.0370.869
 Non-matted algae*0.6280.0020.2390.283
 Proportion of time containing water*0.5630.006NANA
 Permanence rank*0.5650.006−0.0900.691

In contrast, the correlation of Anopheles larval density with the presence of grass and low algal content appeared to be genuine because it was only apparent when counts from periods, when the habitats were wet only, were considered (Table 4). The clear negative association between Anopheles larval density and non-matted algal content (Table 4) may help to explain the low densities of Anopheles larvae and adults in Mbita during the wet period of mid-2001 (Figures 2 and 6c) because algal content was very high in those habitats containing water during this period (Figure 6a,b).

Total productivity for larvae of both Anopheles and culicine and for pupae over the full course of the study period was clearly and positively correlated with the proportion of that time during which the habitats contained water. We ranked habitats according to an index of permanence so that we could assess the importance of stability in terms of simply containing water. The right hand side of Table 4 reveals that the permanence of a habitat has no significant influence on its productivity so long as water is present. This confirms that the limiting factor of the abundance of mosquitoes is the availability of water and thus rainfall.

Although both natural and artificial habitats appeared equally productive over the course of the entire study period, natural habitats had higher densities of Anopheles larvae than artificial habitats when they contained water (1.41 vs. 0.58 larvae per dip, respectively; Table 4). As Anopheles larval density was also positively correlated with the presence of grass or other low vegetation and negatively correlated with non-matted algal content, and parametric Pearson's correlation analysis yielded essentially identical results to those presented in Table 4, we carried out a parametric partial correlation analysis to determine whether such factors could account for the higher productivity of natural habitats for Anopheles larvae. Partial correlation analysis, controlling for the presence of grass and algal index, confirmed that when these factors were taken into account, natural habitats were no more productive than artificial ones when they contained water (correlation coefficient = −0.168, P = 0.48). Thus, the seven surveyed natural habitats were more productive when containing water than were the 14 artificial habitats because they more commonly contain grass (90%vs. 27% of all sampling occasions, respectively; Mann–Whitney U = 7.5, P < 0.01) and have lower mean algal indices (0.29 vs. 1.10, respectively; Mann–Whitney U = 15, P < 0.01). Although E. crassipes and Lemna sp. were also observed in one and three habitats, respectively, and when present seemed to result in lower counts of aquatic stage mosquitoes, they were not common enough to enable meaningful statistical analysis.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

The generalized and frequently cited view most likely dating back to Muirhead-Thomson's (1951) observations in West Africa that A. gambiae s.l. larvae develop in freshwater habitats that are small, temporary, clean and sun-exposed is correct, but our results suggest it might mislead those interested in working on larval control of African malaria vectors. Muirhead-Thomson (1951) himself emphasized, ‘unfortunately gambiae does not lend itself to generalization’. This is supported by Holstein's (1954) observation that ‘A. gambiae does not follow the textbook in its habitat selection and it can be found in a wide variety of water bodies independent of size’. He furthermore highlighted ‘that it is difficult to attribute a definite type of breeding place to gambiae and that this anopheline is likely to breed in almost any water that happens to be available; admittedly, it can be said to show a slight preference for small sunlit pools, but it does not follow that these represent the characteristic breeding-place of this mosquito’. Furthermore, work in larger urban settings has indicated that A. gambiae can adapt to new habitat types over a relatively short number of years (Chinery 1984). So the flexibility of this species complex should never be underestimated in operational larval control programmes where high effective coverage are necessitated by high levels of endemicity (MacDonald 1957; Beier et al. 1999; Smith et al. 2001b; Killeen et al. 2002b; Gu et al. 2003).

Our study further quantifies the ubiquitous nature of A. gambiae breeding habitats showing that semipermanent and permanent habitats can be as suitable for the proliferation of both Anopheles and culicines as temporary habitats. For example, of all available habitats, the majority were colonized by mosquito larvae at any sampling date, regardless of permanence rank.

However, small-scale habitat variation in the characteristics of water-containing habitats also influenced Anopheles abundance, with their larval density being negatively associated with non-matted algal content and positively with the occurrence of tufts of low vegetation-like grass. The latter observation was described by Muirhead-Thomson (1945) from breeding sites in Sierra Leone. However, the negative association we report between Anopheles density and non-matted algal content contradicts the conclusion of Gimnig et al. (2002) that algal food plays a key role in Anopheles habitats. Since in our study, high non-matted algal content was associated with older habitats, this combination was probably responsible for abiotic (e.g. chemical water quality) and biotic (predatory fauna) conditions that reduced oviposition and/or larval survival. The study of Gimnig et al. (2002) specifically investigated artificial habitats of very small-size (35 cm in diameter, 2–3 l of water) and habitat conditions are therefore not comparable with the habitats surveyed in this study. Thus, we conclude that there does not as yet appear to be any straightforward association between larval density and algal productivity and further investigations are needed.

Man-made larval habitats in close vicinity to human habitation are known to play an important role in Anopheles proliferation (Holstein 1954; Minakawa et al. 2002; Girardin et al. 2004), especially in the dry season when supported by human activity. The very common cement-lined pits in Mbita provide excellent conditions for both Anopheles and culicine larvae as they keep water for considerably longer time than natural habitats. Furthermore, they are often re-filled with fresh lake water so that the number of available habitats is artificially kept higher than rain alone would support.

While anopheline and culicine larvae co-existed in all investigated habitats, there appears to be some evidence of differential habitat use. Anopheles gambiae s.l. was found in habitats more frequently and in higher densities than in hillside habitats, indicating that macrohabitat variations because of complex microhabitat features play an important but not yet well understood role in Anopheles larval distribution.

Despite some evidence of habitat separation, even habitats with extremely high culicine larval proliferation held consistent numbers of Anopheles throughout the year, indicating that larval control of selected Anopheles habitats will not be sufficient for malaria control in endemic areas of Africa because very large reductions of transmission are required to achieve useful alleviations of clinical disease burden (Beier et al. 1999; Smith et al. 2001a; Gu et al. 2003).

The long-term nature of our study has revealed that many of the direct associations concluded for Anopheles larval distribution (e.g. algal content and mosquitoes, habitat preference) may be the result of selective, cross-sectional sampling and may be more complicated than previously thought. Short-term studies and spot-check observations on mosquito behaviour, larval ecology and larval habitats bear the risk of overlooking important aspects and to generalize conclusions. Anopheles gambiae's ability to use a great variety of larval habitats and to colonize even those considered less favourable must be taken into account in the course of antimalarial campaigns against larval instars. When Watson (1953) describes the larval habitats of A. gambiae in the copperbelt of Zambia he concluded for some sites ‘larvae were never found there but it was a potential breeding place for this species and was most probably occupied by them at certain times’. As presented here, all monitored habitats did contain Anopheles larvae at least once during the study period of 20 months. We conclude that previous, short-term investigations probably greatly underestimated the range and importance of different potential Anopheles habitats. We therefore suggest that there is a great need for longitudinal larval ecology studies over extended time periods in preparation of antilarval measures in various African settings.

From our involvement in recent larval control projects in a variety of settings across East Africa, we found that an excessively generalized and dogmatic view of the nature of A. gambiae larval habitats has found its way in practical applications. Our work challenges this perception, and highlights the risk that important breeding habitats or those representing dry season refugia may be overlooked in control operations. Recent reports on urban malaria (Chinery 1984; MacIntyre et al. 2002; Keating et al. 2003) suggest that A. gambiae adapted to urban environments by changing their behaviour and colonizing dirty and polluted habitats. It is possible, however, that these habitats have always been capable of supporting larvae, if not to the same frequency as ‘cleaner’ sites. Indeed, the definition of ‘clean’, ‘dirty’ and ‘polluted’ water itself is unclear when examining the original literature (Muirhead-Thomson 1945, 1951; Holstein 1954). For example, Muirhead-Thomson (1951) concluded that ‘pools and puddles in which gambiae breeds become unsuitable very quickly when polluted with cut and rotting vegetation, but animal pollution does not seem to have such an effect’. The latter was often observed during our study, and suggests distinctions between clean and dirty water are not obvious. Anopheles larval development has been described previously for essentially all conceivable water bodies with the potential to harbour them, and exclusions have hardly been made (Holstein 1954).

The term ‘preferred site’ was historically associated with larval densities per site, but several recent studies from southern America (Berti et al. 1993; Fernandez-Salas et al. 1994; Grillet 2000) support the view that sites with highest larval densities per surface area are not necessarily the most important habitats for vector proliferation over space and time. Larval ecology studies of malaria vectors in Venezuela have shown that a large number of low density, but continuously productive anopheline habitats contribute more than single high-density larval habitats to the density of adults and the risk of malaria (Grillet 2000). Furthermore, Berti et al. (1993) concluded for the same study area that one cannot assume that wet season habitats are more favourable than dry season habitats for malaria vector development, but that there is an important interplay among habitat types over space and time. The authors hypothesize that although highest larval densities have been measured in temporary larval habitats, permanent habitats seem to be of more importance for malaria transmission in the area.

The importance of dry season habitats for the perennial transmission of malaria needs to be considered in vector control operations. Dry season habitats are usually well defined and limited in numbers although responsible for continuous production of adult mosquitoes. Antilarval measures are very well suited to targeting these sites, thus reducing the overall mosquito population before the increase of habitat availability during the rainy season (Soper & Wilson 1943; Shousha & Pasha 1948; Fernandez-Salas et al. 1994; Grillet 2000). However, endemic malaria transmission in Africa is often viewed as being associated with a very high number of temporary habitats. Consequently, antilarval measures for malaria control are often considered impossible. While this is surely true for some regions in tropical Africa where intensive and extended rainfall leads to vast areas of breeding habitats inaccessible by foot, in areas-like Suba District, western Kenya, the high number of temporary habitats occurring only over a very short period of time do not represent a major challenge for larval control interventions as they are easily accessible and only partly responsible for the Anopheles production in the area.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

We conclude that A. gambiae s.l. in a setting such as Mbita, a small town in western Kenya, is broadly distributed across a variety of habitat types, regardless of permanence and all potential breeding sites should be considered as sources of malaria risk at any time of the year and exhaustively targeted in any larval control intervention. Man-made habitats that occur in larger numbers in all environments contribute significantly to the transmission of malaria throughout the year. We also conclude that there are various areas in rural Africa with perennial malaria transmission and seasonally larger number of habitats where larval control provides an option to be integrated into malaria control programmes.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References

The authors thank the residents of Mbita for their continuous support and for allowing access to private compounds and houses at any time. We are particularly grateful to Dr Heather Ferguson for her help with the statistical analyses and the valuable comments on the manuscript. The administrative support of ICIPE and the facilitation of the project through Dr John Githure are highly appreciated. Funding for this study was provided by Valent BioSciences Corp., Illinois, USA and GFS, Waldsee, Germany. GK was supported by the Swiss Tropical Institute.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Study area
  6. Mosquito adult survey
  7. Mosquito larval survey
  8. Relative availability of habitats.
  9. Larval densities and habitat characteristics.
  10. Statistical methods
  11. Results
  12. Mosquito adult densities
  13. Dynamics of relative availability of larval habitats
  14. Larval identification and distribution
  15. Seasonality of mosquito population dynamics
  16. Correlation of habitat characteristics and mosquito abundance
  17. Discussion
  18. Conclusion
  19. Acknowledgements
  20. Conflict of interest
  21. References
  • Bøgh C, Clarke SE, Jawara M, Thomas CJ & Lindsay SW (2003) Localized breeding of the Anopheles gambiae complex (Diptera: Culicidae) along the river Gambia, West Africa. Bulletin of Entomological Research 93, 279287.
  • Bayoh MN & Lindsay SW (2003) Effect of temperature on the development of the aquatic stages of Anopheles gambiae sensu stricto (Diptera: Culicidae). Bulletin of Entomological Research 93, 375381.
  • Beier JC, Killeen GF & Githure JI (1999) Short report: Entomologic inoculation rates and Plasmodium falciparum malaria prevalence in Africa. American Journal of Tropical Medicine and Hygiene 61, 109113.
  • Berti J, Zimmerman R & Amarista J (1993) Spatial and temporal distribution of anopheline larvae in two malarious areas in Sucre State, Venezuela. Memorias do Instituto Oswaldo Cruz 88, 353362.
  • Booman M, Sharp BL, Martin CL, Manjate B, La Grange JJ & Durrheim DN (2003) Enhancing malaria control using a computerised management system in Southern Africa. Malaria Journal 2, 13.
  • Castro MC, Yamagata Y, Mtasiwa D et al. (2004)Integrated urban malaria control: a case study in Dar es Salaam, Tanzania. American Journal of Tropical Medicine and Hygiene 71(Suppl. 2) 103117.
  • Chinery WA (1984) Effects of ecological changes on the malaria vectors Anopheles funestus and the Anopheles gambiae complex of mosquitoes in Accra, Ghana. Journal of Tropical Medicine and Hygiene 87, 7581.
  • Clyde DF (1967) Malaria in Tanzania. Oxford University Press, London, UK.
  • Drakeley C, Schellenberg D, Kihonda J et al. (2003) An estimation of the entomological inoculation rate for Ifakara: a semi-urban area in a region of intense malaria transmission in Tanzania. Tropical Medicine and International Health 8, 767774.
  • Durrheim DN, Speare R & Harries AD (2002) Research that influences policy and practice – characteristics of operational research to improve malaria control in Mpumalanga province, South Africa. Malaria Journal 1, 9.
  • Fernandez-Salas I, Roberts DR, Rodriguez MH & Marina-Fernandez CF (1994) Bionomics of larval populations of Anopheles pseudopunctipennis in the Tapachula foothills area, Southern Mexico. Journal of the American Mosquito Control Association 10, 477486.
  • Fillinger U, Knols BGJ & Becker N (2003) Efficacy and efficiency of new Bacillus thuringiensis var. israelensis and Bacillus sphaericus formulations against Afrotropical anophelines in Western Kenya. Tropical Medicine and International Health 8, 3747.
  • Gimnig JE, Ombok M, Kamau L & Hawley WA (2001) Characteristics of larval anopheline (Diptera: Culicidae) habitats in Western Kenya. Journal of Medical Entomology 38, 282288.
  • Gimnig JE, Ombok M, Otieno S, Kaufman MG, Vulule JM & Walker ED (2002) Density-dependent development of Anopheles gambiae (Diptera: Culicidae) larvae in artificial habitats. Journal of Medical Entomology 39, 162172.
  • Girardin O, Dao D, Koudou BG et al. (2004) Opportunities and limiting factors of intensive vegetable farming in malaria endemic Côte d'Ivoire. Acta Tropica 89, 109123.
  • Gouagna LC, Ferguson HM, Okech BA et al. (2004) Plasmodium falciparum malaria disease manifestations in humans and transmission to Anopheles gambiae: a field study in Western Kenya. Parasitology 128, 235243.
  • Grillet ME (2000) Factors associated with distribution of Anopheles aquasalis and Anopheles oswaldoi (Diptera: Culicidae) in a malarious area, northeastern Venezuela. Journal of Medical Entomology 37, 231238.
  • Gu W, Killeen GF, Mbogo CM, Regens JL, Githure JI & Beier JC (2003) An individual-based model of Plasmodium falciparum malaria transmission on the coast of Kenya. Transactions of the Royal Society of Tropical Medicine and Hygiene 97, 4350.
  • Gubler DJ & Clark GG (1994) Community-based integrated control of Aedes aegypti: a brief overview of current programs. American Journal of Tropical Medicine and Hygiene 50, 5060.
  • Habicht JP, Victora CG & Vaughan JP (1999) Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology 28, 1018.
  • Hay SI, Rogers DJ, Toomer JF & Snow RW (2000) Annual Plasmodium falciparum entomological inoculation rates (EIR) across Africa: literature survey, Internet access and review. Transactions of the Royal Society of Tropical Medicine and Hygiene 94, 113127.
  • Holstein MH (1954) Biology of Anopheles gambiae. World Health Organization, Geneva, Switzerland.
  • Hopkins MA (1940) Afforestation as a method of drying up swamps. East African Medical Journal 17, 189194.
  • Kazura JW (ed.) (2003) The western Kenya insecticide-treated bed net trial. American Journal of Tropical Medicine and Hygiene 68 (Suppl. 4), 1173.
  • Keating J, MacIntyre K, Mbogo C et al. (2003) A geographic sampling strategy for studying relationships between human activity and malaria vectors in urban Africa. American Journal of Tropical Medicine and Hygiene 68, 357365.
  • Keiser J, Utzinger J, Castro MC, Smith T, Tanner M & Singer BH (2004) Urbanization in sub-Saharan Africa and implication for malaria control. American Journal of Tropical Medicine and Hygiene 71(Suppl. 2) 118127.
  • Killeen GF, McKenzie FE, Foy BD, Schieffelin C, Billingsley PF & Beier JC (2000a) The potential impact of integrated malaria transmission control on entomologic inoculation rate in highly endemic areas. American Journal of Tropical Medicine and Hygiene 62, 545551.
  • Killeen GF, McKenzie FE, Foy BD, Schieffelin C, Billingsley PF & Beier JC (2000b) A simplified model for predicting malaria entomologic inoculation rates based on entomologic and parasitologic parameters relevant to control. American Journal of Tropical Medicine and Hygiene 62, 535544.
  • Killeen GF, McKenzie FE, Foy BD, Bøgh C & Beier JC (2001) The availability of potential hosts as a determinant of feeding behaviours and malaria transmission by African mosquito populations. Transactions of the Royal Society of Tropical Medicine and Hygiene 95, 469476.
  • Killeen GF, Fillinger U, Kiche I, Gouagna LC & Knols BG (2002a) Eradication of Anopheles gambiae from Brazil: lessons for malaria control in Africa? Lancet Infectious Diseases 2, 618627.
  • Killeen GF, Fillinger U & Knols BG (2002b) Advantages of larval control for African malaria vectors: low mobility and behavioural responsiveness of immature mosquito stages allow high effective coverage. Malaria Journal 1, 8.
  • Killeen GF, Seyoum A & Knols BG (2004) Rationalizing historical successes of malaria control in Africa in terms of mosquito resource availability management. American Journal of Tropical Medicine and Hygiene 71(Suppl. 2) 8793.
  • Klinkenberg E, Takken W, Huibers F & Toure YT (2003) The phenology of malaria mosquitoes in irrigated rice fields in Mali. Acta Tropica 85, 7182.
  • Koenraadt CJ & Takken W (2003) Cannibalism and predation among larvae of the Anopheles gambiae complex. Medical and Veterinary Entomology 17, 6166.
  • Koenraadt CJ, Paaijmans KP, Githeko AK, Knols BG & Takken W (2003) Egg hatching, larval movement and larval survival of the malaria vector Anopheles gambiae in desiccating habitats. Malaria Journal 2, 20.
  • Kouznetsov RL (1977) Malaria control by application of indoor spraying of residual insecticides in tropical Africa and its impact on community health. Tropical Doctors 7, 8191.
  • MacDonald G (1957) The Epidemiology and Control of Malaria. Oxford University Press, London, UK.
  • MacIntyre K, Keating J, Sosler S et al. (2002) Examining the determinants of mosquito-avoidance practices in two Kenyan cities. Malaria Journal 1, 14.
  • Manguin S, Roberts DR, Andre RG, Rejmankova E & Hakre S (1996a) Characterization of Anopheles darlingi (Diptera: Culicidae) larval habitats in Belize, Central America. Journal of Medical Entomology 33, 205211.
  • Manguin S, Roberts DR, Peyton EL, Rejmankova E & Pecor J (1996b) Characterization of Anopheles pseudopunctipennis larval habitats. Journal of the American Mosquito Control Association 12, 619626.
  • Mathenge EM, Omweri GO, Irungu LW et al. (2004) Comparative field evaluation of the Mbita trap, the Centers for Disease Control light trap, and the human landing catch for sampling of malaria vectors in western Kenya. American Journal of Tropical Medicine and Hygiene 70, 3337.
  • Minakawa N, Mutero CM, Githure JI, Beier JC & Yan G (1999) Spatial distribution and habitat characterization of anopheline mosquito larvae in western Kenya. American Journal of Tropical Medicine and Hygiene 61, 10101016.
  • Minakawa N, Seda P & Yan G (2002) Influence of host and larval habitat distribution on the abundance of African malaria vectors in western Kenya. American Journal of Tropical Medicine and Hygiene 67, 3238.
  • MOH (2003) National Business Plan for Malaria Control. Ministry of Health, Nairobi, Kenya.
  • Muirhead-Thomson RC (1945) Studies on the breeding places and control of Anopheles gambiae and Anopheles gambiae var. melas in costal districts of Sierra Leone. Bulletin of Entomological Research 39, 185252.
  • Muirhead-Thomson RC (1951) Mosquito Behaviour in Relation to Malaria Transmission and Control in the Tropics. Edward Arnold & Co., London, UK.
  • Mutero CM, Ouma JH, Agak BK, Wanderi JA & Copeland RS (1998) Malaria prevalence and use of self-protection measures against mosquitoes in Suba district. East African Medical Journal 75, 1116.
  • Rejmankova E, Rubio-Palis Y & Villegas L (1999) Larval habitats of anopheline mosquitoes in the upper Orinoco, Venezuela. Journal of Vector Ecology 24, 130137.
  • Rodriguez AD, Rodriguez MH, Meza RA et al. (1993) Dynamics of population densities and vegetation associations of Anopheles albimanus larvae in a coastal area of southern Chiapas, Mexico. Journal of the American Mosquito Control Association 9, 4658.
  • Shiff C (2002) Integrated approach to malaria control. Clinical Microbiology Reviews 15, 278293.
  • Shililu J, Ghebremeskel T, Mengistu S et al. (2003) Distribution of anopheline mosquitoes in Eritrea. American Journal of Tropical Medicine and Hygiene 69, 295302.
  • Shousha AT & Pasha MD (1948) The eradication of Anopheles gambiae from upper Egypt. Bulletin of the World Health Organization 1, 309333.
  • Smith TA, Hii JL, Genton B et al. (2001a) Associations of peak shifts in age-prevalence for human malarias with bednet coverage. Transactions of the Royal Society of Tropical Medicine and Hygiene 95, 16.
  • Smith TA, Leuenberger R & Lengeler C (2001b) Child mortality and malaria transmission intensity in Africa. Trends in Parasitology 17, 145149.
  • Soper FL & Wilson DB (1943) Anopheles gambiae in Brazil. The Rockefeller Foundation, New York, USA.
  • Watson M (1911) The Prevention of Malaria in the Federated Malay States. Liverpool School of Tropical Medicine, Liverpool, UK.
  • Watson M (1953) African Highway. John Murray, London, UK.
  • WHO (1992) Entomological Field Techniques for Malaria Control. World Health Organization, Geneva, Switzerland.
  • WHO/UNICEF (2003) The African Malaria Report 2003. WHO/UNICEF, Geneva, Switzerland.
  • Ye-Ebiyo Y, Pollack RJ, Kiszewski A & Spielman A (2003) Enhancement of development of larval Anopheles arabiensis by proximity to flowering maize (Zea mays) in turbid water and when crowded. American Journal of Tropical Medicine and Hygiene 68, 748752.