• View in gallery
    Figure 1.

    Study area in Mali showing the irrigation scheme, the agricultural zones, and the study villages.

  • View in gallery
    Figure 2.

    Variation in Anopheles gambiae s.l. (top) and An. funestus (bottom) density (bars), parity rate (PR) (white dots), and human blood index (HBI) (black dots) over the study period. Error bars show 95% confidence intervals.

  • 1

    Ijumba JN, Lindsay SW, 2001. Impact of irrigation on malaria in Africa: paddies paradox. Med Vet Entomol 15 :1–11.

  • 2

    Carnevale P, Guillet P, Robert V, Fontenille D, Doannio J, Coosemans M, Mouchet J, 1999. Diversity of malaria in rice growing areas of the Afrotropical region. Parassitologia 41 :273–276.

    • Search Google Scholar
    • Export Citation
  • 3

    Bambaradeniya CNB, Edirisinghe JP, 2001. The Ecological Role of Spiders in the Rice fields of Sri Lanka. Biodiversity 2 :3–10.

  • 4

    Khush GS, 1984. Terminology for rice growing environments. Terminology for Rice Growing Ecosystems. Manila: International Rice Research Institute, 5–10.

  • 5

    Diuk-Wasser MA, Bagayoko M, Sogoba N, Dolo G, Toure MB, Traore SF, Taylor CE, 2004. Mapping rice field anopheline breeding habitats in Mali, West Africa, using Landsat ETM+ sensor data. Int J Remote Sens 25 :359–376.

    • Search Google Scholar
    • Export Citation
  • 6

    Diuk-Wasser MA, Toure MB, Dolo G, Bagayoko M, Sogoba N, Traore SF, Manoukis N, Taylor CE, 2005. Vector abundance and malaria transmission in rice-growing villages in Mali. Am J Trop Med Hyg 72 :725–731.

    • Search Google Scholar
    • Export Citation
  • 7

    Coluzzi M, Petrarca V, 1973. Aspirator with paper cup for collecting mosquitoes and other insects. Mosq News 33 :249–250.

  • 8

    Detinova TS, 1962. Age-grouping methods in Diptera of medical importance with special reference to some vectors of malaria. Monogr Ser World Health Organ 47 :13–191.

    • Search Google Scholar
    • Export Citation
  • 9

    Beier MS, Schwartz IK, Beier JC, Perkins PV, Onyango F, Koros JK, Campbell GH, Andrysiak PM, Brandling-Bennett AD, 1988. Identification of malaria species by ELISA in sporozoite and oocyst infected Anopheles from western Kenya. Am J Trop Med Hyg 39 :323–327.

    • Search Google Scholar
    • Export Citation
  • 10

    Garrett-Jones C, 1964. Prognosis for interruption of malaria transmission through assessment of the mosquito’s vectorial capacity. Nature 204 :1173–1175.

    • Search Google Scholar
    • Export Citation
  • 11

    Davidson G, 1954. Estimation of the survivalrate of anopheline mosquitoes in nature. Nature 174 :792–793.

  • 12

    Service MW, 1976. Mosquito Ecology: Field Sampling Methods. Essex, United Kingdom: Applied Science Publisher.

  • 13

    Briet OJ, 2002. A simple method for calculating mosquito mortality rates, correcting for seasonal variations in recruitment. Med Vet Entomol 16 :22–27.

    • Search Google Scholar
    • Export Citation
  • 14

    Charlwood JD, Alecrim WA, 1997. Capture-recapture studies with the South American malaria vector Anopheles darlingi, Root. Ann Trop Med Parasitol 83 :569–576.

    • Search Google Scholar
    • Export Citation
  • 15

    Mehugh CP, 1990. Survivorship and gonotrophic cycle length of Culex tarsalis (Diptera, Culicidae) near Sheridan, Placer country, California. J Med Entomol 27 :1027–1030.

    • Search Google Scholar
    • Export Citation
  • 16

    Ecker MD, Gelfand AE, 1997. Bayesian variogram modeling for an isotropic spatial process. J Agric Biol Environ Stat 2 :347–368.

  • 17

    Spiegelhalter DJ, Best NG, Carlin BR, van der Linde A, 2002. A Bayesian measures of model complexity and fit. J R Stat Soc Ser B 64 :583–616.

    • Search Google Scholar
    • Export Citation
  • 18

    Cressie NAC, 1993. Statistics for Spatial Data. New York: John Wiley & Sons, Inc.

  • 19

    Dolo G, Briet OJ, Dao A, Traore SF, Bouare M, Sogoba N, Niare O, Bagayogo M, Sangare D, Teuscher T, Toure YT, 2004. Malaria transmission in relation to rice cultivation in the irrigated Sahel of Mali. Acta Trop 89 :147–159.

    • Search Google Scholar
    • Export Citation
  • 20

    Koudou BG, Tano Y, Doumbia M, Nsanzabana C, Cisse G, Girardin O, Dao D, N’Goran EK, Vounatsou P, Bordmann G, Keiser J, Tanner M, Utzinger J, 2005. Malaria transmission dynamics in central Côte d’Ivoire: the influence of changing patterns of irrigated rice agriculture. Med Vet Entomol 19 :27–37.

    • Search Google Scholar
    • Export Citation
  • 21

    Klinkenberg E, Takken W, Huibers F, Toure YT, 2003. The phenology of malaria mosquitoes in irrigated rice fields in Mali. Acta Trop 85 :71–82.

    • Search Google Scholar
    • Export Citation
  • 22

    Mouchet J, Brengues J, 1990. Agriculture-health interface in the field of epidemiology of vector-borne diseases and the control of vectors. Bull Soc Pathol Exot 83 :376–393.

    • Search Google Scholar
    • Export Citation
  • 23

    Doannio JM, Dossou-Yovo J, Diarrassouba S, Rakotondraibe ME, Chauvancy G, Chandre F, Riviere F, Carnevale P, 2002. Dynamics of malaria transmission in Kafine, a rice growing village in a humid savannah area of Côte d’Ivoire. Bull Soc Pathol Exot 95 :11–16.

    • Search Google Scholar
    • Export Citation
  • 24

    Temel T, 2004. Malaria from the gap: need for cross-sector cooperation in Azerbaijan. Acta Trop 89 :249–259.

Past two years Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 256 61 7
PDF Downloads 18 9 0
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

SPATIAL ANALYSIS OF MALARIA TRANSMISSION PARAMETERS IN THE RICE CULTIVATION AREA OF OFFICE DU NIGER, MALI

NAFOMON SOGOBAMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by NAFOMON SOGOBA in
Current site
Google Scholar
PubMed
Close
,
PENELOPE VOUNATSOUMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by PENELOPE VOUNATSOU in
Current site
Google Scholar
PubMed
Close
,
SEYDOU DOUMBIAMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by SEYDOU DOUMBIA in
Current site
Google Scholar
PubMed
Close
,
MAGARAN BAGAYOKOMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by MAGARAN BAGAYOKO in
Current site
Google Scholar
PubMed
Close
,
MAHAMADOU B. TOURÉMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by MAHAMADOU B. TOURÉ in
Current site
Google Scholar
PubMed
Close
,
IBRAHIM M. SISSOKOMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by IBRAHIM M. SISSOKO in
Current site
Google Scholar
PubMed
Close
,
SEKOU F. TRAOREMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by SEKOU F. TRAORE in
Current site
Google Scholar
PubMed
Close
,
YÉYA T. TOURÉMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by YÉYA T. TOURÉ in
Current site
Google Scholar
PubMed
Close
, and
THOMAS SMITHMalaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako, Mali; Swiss Tropical Institute, Basel, Switzerland; World Health Organization, Libreville, Gabon; World Health Organization, Geneva, Switzerland

Search for other papers by THOMAS SMITH in
Current site
Google Scholar
PubMed
Close

The effects of rice growth environment on malaria transmission, taking into account spatial correlation, were assessed in the Office du Niger, Mali. Between April 1999 to January 2001, 8 quarterly entomologic surveys were conducted in 18 villages in 3 agricultural zones. Vector densities in sleeping houses were related to rice crop, rice development stages, vegetation abundance, water state, and seasons. They were high throughout the rice-growing seasons, increased as the rice crop developed, and decreased as vegetation became abundant. They also showed large spatial correlations (up to 30.6 km). The vectorial capacity exhibited both seasonal and village-to-village variation. Parity and the human blood index were weakly related to adult densities and showed low spatial correlations (up to 3.4 km), which suggested that small area variation in malaria transmission results mainly from variations in vector-human contact. Control strategies in rice cultivation areas should pay attention to this local variation.

INTRODUCTION

Many studies have been carried out in Africa to assess the impact of rice cultivation on malaria. However, no consistent association has been found between irrigated rice fields and malaria transmission measured by classic entomologic methods.1 It has been reported that transmission intensity in irrigated settlements is higher, similar, or lower than in neighboring villages outside the irrigation scheme depending on the malaria situation before the implementation of the irrigated projects.1,2

Little attention has been paid to the spatial variation in malaria transmission in the rice agro-ecosystems because they are generally monocultures and are considered to be homogenous. However, rice-growing environments change during rice development and vary significantly within and between countries.3,4 This variability affects the risk of malaria transmission in large irrigated rice-cultivation areas.

The Office du Niger, in the district of Niono, Mali, represents one such area where two main environments result from an ongoing renovation process: renovated and non-renovated (Figure 1). Our previous study used remote sensing data to map anopheline breeding sites and described the relation between mosquito densities, survival rates, zoophilic rates, and vectorial capacity to explain the low prevalence of malaria.5,6 In the current study, we reanalyze the same data to assess the effects of rice growth environmental features on malaria transmission parameters to get an insight into the spatial variation of malaria risk within a large-scale irrigated rice-cultivation area. This work was complemented by repeated cross-sectional anopheline larval collections in selected rice plots, which will be published elsewhere.

MATERIALS AND METHODS

Study area.

The study was carried out in the Office du Niger area (Figure 1) located in the inner delta of the Niger River, 350 km northeast of Bamako, Mali, in the prefecture of Niono, in the region of Se gou. This area comprises a colonial-era irrigation system that has undergone upgrading/repairing starting in the 1980s. At the time of the study, the Niono and Ndebougou zones were renovated, unlike the Molodo zone, and a surface of 68,000 hectares was used for rice cultivation.7

The district of Niono has approximately 360,000 inhabitants with 180,000 living in the irrigated area. Approximately 44% of the population is less than 15 years of age and only 20% is literate. The production system is based on animals (cow, donkey) that are used for plowing, for producing organic fertilizer, and as pushcarts. Some farmers are also involved in the production of meat and milk.

Depending on the quality of water supply and regimen control, there are three categories of rice plots: controlled plots, shallow water regimen plots that are cropped either once or twice a year, and unbounded plots with maximum sustained water depths. The first two categories have adequate delivery and disposal of excess water, and the last category has a poor draining system. In the renovated zones, all plots are shallow controlled water plots and in the un-renovated zone of Molodo all three plot types are encountered.

Study sites.

Eighteen villages were selected in the three agricultural zones of Niono, Ndebougou, and Molodo (Figure 1). The selection criteria were 1) a minimum distance of 2 km between 2 selected villages, 2) accessibility in all seasons, and 3) village cooperation. Each selected village was geo-referenced using handheld GPS receivers (Trimble® Geo-Explorer II). A population census indicated that the median number of inhabitants per village was 963 (minimum = 600, maximum = 2,080).

Rice growth cycle.

The typical rice cultivation cycle occurs from June to December and includes 1) a sowing-transplanting phase (June–August), 2) a growing phase (August–November), and 3) an after-harvest phase (November–December). A second and shorter cultivation cycle (or off season crop) occurs from January to May. The duration of the rice cycle varies between 120 and 150 days, depending on the rice variety.

Following the practice of the Office du Niger administration, we categorized the growing stages of the rice as follows: 1) fallow/plowing (no rice), 2) early vegetative (tilling), 3) vegetative (elongation), 4) reproductive/flowering (gaining), and 5) maturation (mature grain). In addition, we recorded whether fields were fertilized, pre-irrigated, or undergoing irrigation. We also recorded crop type (rice/vegetable/fallow), vegetation abundance, rice state (sparse/dense), water turbidity, soil type, and rice plot types (Table 1).

Mosquito collections and processing.

Between April 1999 to January 2001, 8 cross-sectional surveys were carried out in 18 villages to determine mosquito adult abundance, human biting rates (MBR or ma) rates, parity rate (PR or P), human blood index (HBI or a), and thus, the vectorial capacity (VC or C). The surveys were scheduled according to rice cropping activities and carried out in March 1999 and 2000 (dry hot season), August 1999 and 2000 (rainy season), October 1999 and 2000 (end of rainy season), and January 2000 and 2001 (dry and cold season). Mosquitoes were collected using pyrethrum spray catches (PSCs) and the human bait catches (HBCs).

The PSCs were carried out during daytime in houses using an aerosol of 0.3% pyrethrum sold under the label of Timor. During each survey, 30 compounds (conglomerate of houses) were randomly selected from the list of the compounds in each village. The collection was performed in one house per compound by two teams of three collectors each during two consecutive days in each village. The total number of mosquito collected and the number of sleepers in the house were recorded.

In each village, HBCs were performed at night by two collectors using a mouth aspirator8 and sitting inside and outside of each 1 of 2 sentinel houses at least 200 meters from each other from 6:00 pm. to 6:00 am.7

In both cases, mosquitoes were morphologically identified and malaria vectors selected (Anopheles gambiae s.l. and An. funestus). Mosquitoes from HBC ovaries were dissected and their tracheoles were examined to determine their physiologic parity.8 Blood meals of blood fed and semi-gravid mosquitoes from PSCs were used to determine the HBI by enzyme-linked immunosorbent assay.9 A polymerase chain reaction (PCR) method was used to determine the species of An. gambiae s.l. (An. gambiae s.s. versus An. arabiensis). The potential malaria transmission was estimated by the vectorial capacity formula C = ma2 Pn/(−log P) of An. gambiae s.l., which is the abundant vector,10 where C represents the expected number of inoculations to human from an infected person per time unit, ma is the human-biting density, a is the product of the human-biting habit (estimated to be two days in Mali) and the human blood index (proportion of mosquitoes fed on human), P is the average daily survival of the female mosquito, and n is the mean extrinsic period of development of the parasite in the mosquito (estimated to be 12 days in the study area). We applied the parity status method to estimate mosquito longevity.11 This approach does not incorporate effects of unstable age structure of mosquito population or irregular feeding pattern.12 However, the large time intervals of three months between our surveys did not allow us to apply alternative methods.1315

Statistical analysis.

The data were entered and cleaned in SPSS version 11.0 (SPSS Inc., Chicago, IL) and analyzed in STATA version 8.0 (Stata Corporation, College Station, TX) and WinBUGS version (Imperial College and Medical Research Council, London, United Kingdom). Mosquito densities and human-biting rates were summarized by geometric means. Poisson regression analyses were performed to assess the bivariate relations between mosquito density and a set of rice-growth related predictors. A Bayesian spatial Poisson model was fitted in WinBUGS on the vector density data with explanatory those variables that appeared significant at a 15% significance level in the bivariate regressions. This model was used to quantify spatial correlation in the mosquito density and to adjust the significance of the predictors under the presence of these correlations. In particular, we assumed that the mosquito density Yit in village i and survey t follow a Poisson distribution Yit ~ P0it). Spatial correlation was modeled by village-specific random effects ϕi, i = 1, . , N(N = 72) that assumed to arise from a multivariate normal distribution ϕ = (ϕ1, . ϕN)T ~ MVN (0, Σ) with covariance matrix Σ. We further assumed that spatial correlation is a function of distance between locations, irrespective of the locations themselves (stationarity) and of the direction (isotropy). We adopt an exponential correlation function Σij = σ2 exp(−ρdij where σ2 is the spatial variance, ρ models the rate of correlation decay and dij the distance between the centroids of villages i and j. For the exponential correlation structure specified above, the minimum distance that correlation becomes less than 5% is given by 3/ρ.16 Temporal correlation was introduced by assuming an autoregressive process AR (1) of order 1 on two-week–specific random effects vt = 1, . , 48. The predictors as well as the spatial and temporal effects were modeled on the log scale of the mean parameter μit of the Poisson distribution that corresponds to the average mosquito density in village i and two weeks t log(μit) = X̃itT β̃ + ϕi + vt where X̃it is the predictors of vector and β̃ are the coefficients of the predictors. A non spatial-temporal Bayesian model was also fitted in WinBUGS. The deviance information criterion (DIC) was used to assess the goodness-of-fit of the models.17 The smaller the DIC is, the better the fit.

In a separate analysis, we linked the larval density data with the vector adult data using a Bayesian spatial Poisson model to assess the relation between larvae and vector-related transmission indicators. In particular, we extracted from the larvae data set those collections made two weeks prior to the adult data collection that allowed a two-week lag for the larvae to become adults. The Pearson’s chi-square test was applied to assess seasonality in the PR and HBI. A Bayesian spatial logistic regression was used to look at the relationship between HBI and mosquito density. The Kruskal-Wallis test was used to compare the median vectorial capacity by season and by agricultural zone. Bayesian spatial Poisson models were fitted in WinBUGS to assess the relation between the adult density and environmental factors, as well as adult density and larval density. Previous studies12 have already shown that An. gambiae complex and An. funestus were responsible for malaria transmission in the area; therefore, we focused only on these species.

RESULTS

Vectors population composition and structure.

A total of 366,657 specimens of malaria vectors (An. gambiae s.l. and An. funestus) were collected. Anopheles gambiae s.l. was the predominant species with a relative frequency of 90.2%. Higher frequencies of An. funestus were observed at the end of the rainy season and during the dry cold season, specifically in villages located in the non-renovated zone of Molodo.

Results from the PCR identification–based method show that An. gambiae s.l. was composed of 93.1% An. gambiae s.s. and 6.9% An. arabiensis (n = 891). The highest relative percentage of An. arabiensis (31.2%, n = 93) was observed at the end of the rainy season.

Malaria transmission parameters.

Figure 2 shows the variation of the geometric mean density per house, the PR, and the HBI of both An. gambiae s.l. and An. funestus. Over the study period, the mean density per house was 69.5 (95% confidence interval [CI] = 52.7–86.3) for An. gambiael s.l. and 5.6 (95% CI = 4.4–6.8) for An. funestus. The mean PR and HBI were 60.3% (95% CI = 59.4–61.3, n = 10,705) and 34.7% (95% CI = 33.9–35.4, n = 15,980) for An. gambiae s.l. and 74.4% (95% CI = 72.9–75.9, n = 3,323) and 32.2% (95% CI = 30.9–33.4, n = 5,854) for An. funestus, respectively. On average, the daily survival rate was 77.7% for An. gambiae s.l. and 86.3% for An. funestus. The highest mosquito density period corresponded to the lower HBI and PR for both An. gambiae s.l. and An. funestus. Particularly in August 2000, when the highest density (252.5, 95% CI = 205.4–299.6) for An. gambiae s.l. was observed, the HBI (17.0%, 95% CI = 15.6–18.5) and the PR (57.8, 95% CI = 56.0–59.5) were also low.

The median vectorial capacity (interquartile range) was 0.33 (0.01–1.03), 0.11(0.01–0.79), 0.00 (0.0–0.10), and 0.01(0.0–0.10) during the dry cold season, the dry hot season, the rainy season, and at the end of the rainy season, respectively. The vectorial capacity differs significantly between the seasons (Kruskal-Wallis value = 21.33, degrees of freedom = 3, P < 0.001). In particular, the highest vectorial capacity was observed in the dry cold season, which showed the lowest mosquito density. The median vectorial capacity was not significantly higher (Kruskal-Wallis value = 4.97, P = 0.083) in the non-rehabilitated agricultural zone of Molodo (0.1, 0.0–0.61) than in the rehabilitated zones of Niono (0.02, 0.0–0.44) and Ndebougou (0.0, 0.0–0.54).

Spatial analysis of malaria transmission parameters.

Bivariate and multiple non-spatial and spatial Poisson models were fitted to assess the association between mosquito density and rice growth–related environmental features (Table 1). The goodness of fit criterion indicates that the spatial multiple model fits the data better (DIC = 4,360.0) than the non-spatial model (DIC = 5,092.8). The good predictors of vector density were rice crop, rice development stages, vegetation abundance, water state, and seasons. Field types, which was a good predictor of mosquito density in the non-spatial model, was no longer significant in the spatial multiple Poisson model. Tilling stage of rice, which was not significantly correlated with mosquito density in the multiple independent model, became negatively related in the spatial model. The association of abundant vegetation category to mosquito density changed from positive in the multiple independent model to negative in the spatial model. This clearly illustrates how the standard statistical method, which assumes independence of observations, can overestimate or underestimate the standard error and the significance of the covariates when they are used to analyze spatially correlated data.18 The data show a spatial correlation up to a distance of 30.0 km (95% CI = 22.2–133.2), which was not accounted for in the non-spatial model.

A separate multiple spatial Poisson model was fitted to assess an association between larval density in rice fields and adult density in human settlements. The model estimated a density ratio of 1.005 (95% CI = 1.0013–1.0016) for every increase of adult density by one mosquito. When adjusted for the environmental covariates, the larval density was no longer significant. Spatial correlation was strong and decreased to less than 5% at 35.5 km (95% CI = 21.1–427.4).

Spatial logistic models showed that seasonality was significantly associated with PR and HBI for An. gambiae s.l. and An. funestus (Table 2). Both species were also less likely to feed on humans during the rainy, end of rainy, and dry hot seasons than during the dry cold season. Anopheles gambiae s.l. was less likely to be parous during the dry hot and rainy seasons than during the dry cold season. The odds of parity in An. funestus were significantly higher during the dry hot season (odds ratio = 8.27, 95% CI = 4.95–13.29) and significantly lower during the rainy season and end of rainy season relative to the dry cold season. Mosquito density was significantly associated with the PR and HBI for An. funestus but not for An. gambiae s.l. The minimum distances at which there was no spatial correlation in PR and HBI were 3.36 km (95% CI = 1.41–21.29) and 3.17 km (95% CI = 1.41–19.96) for An. gambiae s.l. and 2.56 km (95% CI = 1.39–15.13) and 2.17 km (95% CI = 1.39–7.31) for An. funestus.

DISCUSSION

The aim of this study was to assess malaria transmission parameters in a large scale-irrigated rice cultivation area and taking into account the spatial correlation present in the data. The main species were An. gambiae s.l. and An. funestus. These are also the most common species in western African rice cultivation areas.19,20 Among these two species, An. gambiae s.l was predominant, accounting on average for 90% (n = 366,657) and was particularly abundant during the rainy season of 1999 and 2000 and the dry hot season of 1999 (second agricultural cycle). The lowest density of An. gambiae s.l. during the second agricultural cycle period of 2000 was related to restrictions imposed in rice cropping by the agricultural department to clean the draining system. During this period, the remaining stagnant water in the canals was used by An. funestus as breeding habitats.21 At the end of the rainy season and during the dry cold season, the frequencies of both species reached similar levels. This seasonal variation in the frequency ratio of the two species is commonly observed and it is related to their ecology.22 The sun-loving An. gambiae s.l. colonizes rice fields at the transplanting period and is replaced by the shade-loving An. funestus when rice plants cover the fields.

The negative association between the adult density with PR and HBI in the Office du Niger has been already reported and has been also observed in neighboring Burkina Faso.6,19,23 The most likely explanation is that when the mosquito density increases, individuals take more protective measures (i.e., bed net use) that may divert mosquitoes to animals such as cattle. This argument is supported by the exceptionally low HBI of the anthropophilic species of An. funestus in spite of its very high PR. Whereas a negative association between adult density and HBI has been observed in the Office du Niger, Mali6,19 and Burkina Faso,23 a recent study conducted in Côte d’Ivoire suggested a positive association reporting HBI up to 95% during high-density periods.20

The vectorial capacity was relatively low with a seasonal and village-to-village variation. The median vectorial capacity was higher in the non-renovated zone of Molodo than in the renovated zones of Niono and Ndebougou, but the statistical significance was borderline. The inadequate water disposal system of the non-renovated zone may have raised the relative humidity that aids mosquito survival1 and therefore the vectorial capacity. The higher prevalence of An. funestus in this zone may have also contributed to this finding. The deficiency in the draining system of this agricultural zone has created deep, vegetated, and persistent water bodies that are used by An. funestus as breeding habitats. However, it is important to note that in this study our estimate of the vector survival did not take into account the recruitment rate in mosquito population, which can have an impact on the PR and thus on season-specific vectorial capacity estimates. Unfortunately, the large sampling interval of our data did not enable us to use alternative approaches. However, the possible bias in the estimates of the vectorial capacity should not be reflected in the comparison between locations because the same method was applied.

Our data showed that shallow controlled plots used for the two agricultural cycles (twice a year) produced fewer larvae than all other plots types. The better draining system has shortened the time they serve as breeding sites for anophelines. Research on water management in rice plots reported numerous and lasting-breeding habitats even after harvesting in inefficiently drained plots.21,24 However, more studies are required to rigorously support this observation because restriction was made in cropping during the second year of our study period.

Adult densities showed marked seasonality. However, they were large enough to sustain transmission throughout the year. This is almost certainly due to the current cultivation methods, which are characterized by overlaps between several agricultural cycles.20,21 In spite of the high densities during the rainy season, the potential for transmission was lower than in the dry season. This could be explained by the decreases in HBI (a measure of vector-human contact) and PR (a measure of vector longevity) during that period (Figure 2). In the dry season, lower vector densities may lead to relaxation of individual protection. Vector-human contact may also be higher during the dry hot season because people spend longer periods outside.

Spatial correlation in mosquito density data was significant at distances up to 30.6 km, which indicated that the number of mosquitoes per house is related to the number of mosquitoes up to 30.6 km apart. This strong spatial correlation is likely to be related to the rice cultivation environment that is associated with mosquito abundance because of the suitable conditions it creates. In addition, our analysis does not include climate-related parameters such as rainfall and temperature, which are spatially structured and might also explain the residual spatial correlation. Spatial correlation in PR was relatively low (up to 3.36 km and 3.17 km for An. gambiae s.l. and An. funestus, respectively). Similarly spatial correlation in HBI is up to 2.56 km and 2.17 km for An. gambiae s.l. and An. funestus, respectively. This weak spatial correlation suggests that PR and HBI are more related to local conditions such as population behavior and economic status, presence of animals, rather than similar environment over large areas. A spatial analysis performed to assess the effect of mosquito density on PR and HBI did not show any significant association other that between PR of An. funestus and mosquito density (OR = 0.97, 95% CI = 0.96–0.98). The importance of local environment may also explain the difference we observed in the vectorial capacity from village to village and between the agricultural zones. A separate model linking larvae and adult density suggested that larvae density was significantly related to the mosquito density per house. This association disappeared when we adjusted the density for rice growth environmental factors.

This study is the first to quantify the amount of spatial correlation in rice cultivation areas and to assess the effect of rice growing on malaria transmission taking into account this correlation. Our results show that in the Office du Niger, rice cultivation has created environmental conditions favorable to the occurrence of the two major malaria vectors, which with current agricultural practices, is leading to year-round transmission with a marked seasonality.

Local variation was observed in mosquito parity ratio and human blood index, both of which measure the vector-human contact rate and thus the potential for malaria transmission intensity. Attention must be paid to this local variation when implementing control strategies. Similar studies elsewhere in Africa are needed if we are to understand whether these are general features of malaria transmission in large scale irrigated ecosystems.

Table 1

Estimates of the effects of rice growth on adult mosquito densities*

Bivariate independentMultiple independentMultiple spatial
VariablesEstimates95% CIEstimates95% BCIEstimates95% BCI
* Covariant effects are density ratios. Estimates are posterior means. CI = confidence interval; BCI = Bayesian credible interval; DIC = deviance information criteria.
†Two decimal places are given to show non-significance.
‡Excluded because of collinearity.
Rice crop
    No rice1.01.01.0
    Rice1.21.2–1.31.21.1–1.31.41.3–1.5
Rice stages
    No rice1.01.01.0
    Tilling†2.12.1–2.21.040.95–1.130.80.7–0.8
    Elongation1.31.2–1.31.71.5–1.91.61.4–1.9
    Gaining0.80.8–0.93.93.4–4.52.52.1–3.1
    Maturation0.30.3–0.41.31.1–1.60.90.7–1.1
Field types
    Single crop1.01.01.0
    Double crop0.90.9–1.01.11.0–1.11.40.8–2.4
    Independently managed10.9–1.11.31.2–1.41.20.5–3.1
Seasons
    Dry cold1.01.01.0
    Dry hot12.811.3–14.412.210.6–14.114.912.8–17.3
    Rainy29.826.6–33.437.632.7–43.239.434.0–45.7
    End rainy2.82.5–3.22.31.8–2.82.51.9–3.2
Vegetation abundance
    No vegetation1.01.01.0
    Less abundant0.70.6–0.70.30.3–0.30.30.3–0.3
    Abundant3.23.1–3.41.21.1–1.30.80.7–0.8
    Very abundant2.11.9–2.410.9–1.10.90.8–1.1
Rice state
    No rice‡1.0
    Sparse16.913.9–20.5
    Partly covered14.211.6–17.3
    Covered5.64.6–6.9
Agricultural activities
    No rice1.01.01.0
    Pre-irrigation2.11.9–2.40.70.6–0.80.70.6–0.8
    Transplanting2.11.9–2.40.60.5–0.70.70.6–0.8
    Grass removal1.00.0–3.40 × 1081.00.0–3.2 × 108
    Fertilizing1.51.3–1.70.20.1–0.20.10.1–0.2
    Irrigation1.11.0–1.20.20.2–0.30.20.2–0.2
    No activity0.70.6–0.80.60.6–0.70.50.4–0.5
    Protect birds0.10.09–0.20.20.1–0.20.20.1–0.2
    Canal cleaning0.30.2–0.30.30.2–0.30.40.3–0.6
    Water drainage0.30.2–0.40.70.5–1.01.10.8–1.6
    Harvesting0.10.08–0.10.70.6–0.90.80.6–1.0
    Market gardening0.50.4–0.510.0–3.50 × 1081.00.0–3.2 × 108
Water state
    No water1.01.01.0
    Dusty0.50.5–0.60.10.1–0.10.10.1–0.1
    Turbid1.81.7–1.90.60.5–0.70.80.7–0.9
    Clear1.11.0–1.11.11.0–1.110.9–1.1
Soil types
    Clay1.0
    Mixed0.90.9–0.9
Spatial parameters
    Correlation decay (ρ)10.92.4–15.8
    Spatial Variance (σ2)0.70.2–2.1
Goodness of fit
    DIC5,092.814,359.99
Table 2

Multiple spatial logistic regression of parity ratio and human blood index (HBI) on adult mosquito density adjusted for seasonal effects*

Anopheles gambiae s.l.Anopheles funestus
ParametersParous rate OR (95% BCI)HBI OR (95% BCI)Parous rate OR (95% BCI)HBI OR (95% BCI)
* OR = odds ratio; BCI = Bayesian credible interval.
Season
    Dry1.01.01.01.0
    Dry hot0.63 (0.49–0.80)0.41 (0.37–0.46)8.27 (4.95–13.29)0.47 (0.40–0.55)
    Rainy0.32 (0.24–0.41)0.37 (0.31–0.44)0.68 (0.24–1.63)0.69 (0.59–0.80
    End of rainy1.12 (0.80–1.46)0.48 (0.43–0.54)0.75 (0.58–0.96)0.27 (0.23–0.33)
Density1.00 (1.00–1.00)1.00 (1.00–1.00)0.97 (0.96–0.98)1.01 (1.01–1.02)
Correlation decay (ρ)99.03 (15.66–235.50)104.90 (16.70–235.80)130.00 (22.03–238.20)153.50 (45.59–240.00)
Spatial variance (σ2)0.17 (0.08–0.36)0.12 (0.06–0.24)0.23 (0.09–0.52)0.07 (0.03–0.16)
Figure 1.
Figure 1.

Study area in Mali showing the irrigation scheme, the agricultural zones, and the study villages.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 76, 6; 10.4269/ajtmh.2007.76.1009

Figure 2.
Figure 2.

Variation in Anopheles gambiae s.l. (top) and An. funestus (bottom) density (bars), parity rate (PR) (white dots), and human blood index (HBI) (black dots) over the study period. Error bars show 95% confidence intervals.

Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 76, 6; 10.4269/ajtmh.2007.76.1009

*

Address correspondence to Nafomon Sogoba, Malaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako BP 1805, Mali. E-mail: nafomon@mrtcbko.org and Swiss Tropical Institute, Socinstrasse 57, PO Box 4051 Basel, Switzerland. E-mail: n.sogoba@unibas.ch

Authors’ addresses: Nafomon Sogoba, Malaria Research and Training Center, Faculty of Medicine, University of Bamako, Bamako BP 1805, Mali, Telephone: 223-222-5277, Fax: 223-222-4987, E-mail: nafomon@mrtcbko.org and Swiss Tropical Institute, Socinstrasse 57, PO Box 4051, Basel, Switzerland, E-mail: n.sogoba@unibas.ch. Penelope Vounatsou and Thomas Smith, Swiss Tropical Institute, Socinstrasse 57, PO Box 4051, Basel, Switzerland, E-mails: penelope.vounatsou@unibas.ch and Thomas-A.Smith@unibas.ch. Seydou Doumbia, Mahamadou B, Touré, Ibrahim M. Sissoko, and Sekou F. Traore, Malaria Research and Training Center, Faculty of Medicine, University of Bamako, Mali, E-mails: sdoumbi@mrtcbko.org, ibrahim@mrtcbko.org, and cheick@mrtcbko.org. Magaran Bagayoko, Conseiller Sous-Régional pour l’Afrique, Central Biologie des Vecteurs et Lutte Antivectorielle, Organisation Mondiale de la Santé Bureau, Regional pour l’Afrique Bureau Organisation Mondiale de la Santé du Gabon, BP 820, Libreville, Gabon, E-mail: mbagayoko@internetgabon.com. Yéya T. Touré, Special Program for Research and Training in Tropical Diseases, World Health Organization, 1211 Geneva 27, Switzerland, E-mail: tourey@who.int.

Acknowledgments: We thank the Niono Health Center, Office du Niger, the Institute d’Economie Rurale, the 18 villages, and the larvae collectors for their collaboration and participation in the study. We also thank Richard Sakai, Robert Gwatz, and all the members of the Malaria Research and Training Center Entomology Laboratory for their support.

Financial support: This study was supported by a grant from the National Institute of Health, by the National Aeronautic and Space Administration through an Interagency Agreement Y3-AI-5059-03 with the National Institute of Allergy and Infectious Diseases for work at the Malaria Research and Training Center in Mali, and by Project T16/181/476 TDR/WHO. Data analysis was supported by the Swiss National Foundation Project Nr. 3252B0-102136/1.

REFERENCES

  • 1

    Ijumba JN, Lindsay SW, 2001. Impact of irrigation on malaria in Africa: paddies paradox. Med Vet Entomol 15 :1–11.

  • 2

    Carnevale P, Guillet P, Robert V, Fontenille D, Doannio J, Coosemans M, Mouchet J, 1999. Diversity of malaria in rice growing areas of the Afrotropical region. Parassitologia 41 :273–276.

    • Search Google Scholar
    • Export Citation
  • 3

    Bambaradeniya CNB, Edirisinghe JP, 2001. The Ecological Role of Spiders in the Rice fields of Sri Lanka. Biodiversity 2 :3–10.

  • 4

    Khush GS, 1984. Terminology for rice growing environments. Terminology for Rice Growing Ecosystems. Manila: International Rice Research Institute, 5–10.

  • 5

    Diuk-Wasser MA, Bagayoko M, Sogoba N, Dolo G, Toure MB, Traore SF, Taylor CE, 2004. Mapping rice field anopheline breeding habitats in Mali, West Africa, using Landsat ETM+ sensor data. Int J Remote Sens 25 :359–376.

    • Search Google Scholar
    • Export Citation
  • 6

    Diuk-Wasser MA, Toure MB, Dolo G, Bagayoko M, Sogoba N, Traore SF, Manoukis N, Taylor CE, 2005. Vector abundance and malaria transmission in rice-growing villages in Mali. Am J Trop Med Hyg 72 :725–731.

    • Search Google Scholar
    • Export Citation
  • 7

    Coluzzi M, Petrarca V, 1973. Aspirator with paper cup for collecting mosquitoes and other insects. Mosq News 33 :249–250.

  • 8

    Detinova TS, 1962. Age-grouping methods in Diptera of medical importance with special reference to some vectors of malaria. Monogr Ser World Health Organ 47 :13–191.

    • Search Google Scholar
    • Export Citation
  • 9

    Beier MS, Schwartz IK, Beier JC, Perkins PV, Onyango F, Koros JK, Campbell GH, Andrysiak PM, Brandling-Bennett AD, 1988. Identification of malaria species by ELISA in sporozoite and oocyst infected Anopheles from western Kenya. Am J Trop Med Hyg 39 :323–327.

    • Search Google Scholar
    • Export Citation
  • 10

    Garrett-Jones C, 1964. Prognosis for interruption of malaria transmission through assessment of the mosquito’s vectorial capacity. Nature 204 :1173–1175.

    • Search Google Scholar
    • Export Citation
  • 11

    Davidson G, 1954. Estimation of the survivalrate of anopheline mosquitoes in nature. Nature 174 :792–793.

  • 12

    Service MW, 1976. Mosquito Ecology: Field Sampling Methods. Essex, United Kingdom: Applied Science Publisher.

  • 13

    Briet OJ, 2002. A simple method for calculating mosquito mortality rates, correcting for seasonal variations in recruitment. Med Vet Entomol 16 :22–27.

    • Search Google Scholar
    • Export Citation
  • 14

    Charlwood JD, Alecrim WA, 1997. Capture-recapture studies with the South American malaria vector Anopheles darlingi, Root. Ann Trop Med Parasitol 83 :569–576.

    • Search Google Scholar
    • Export Citation
  • 15

    Mehugh CP, 1990. Survivorship and gonotrophic cycle length of Culex tarsalis (Diptera, Culicidae) near Sheridan, Placer country, California. J Med Entomol 27 :1027–1030.

    • Search Google Scholar
    • Export Citation
  • 16

    Ecker MD, Gelfand AE, 1997. Bayesian variogram modeling for an isotropic spatial process. J Agric Biol Environ Stat 2 :347–368.

  • 17

    Spiegelhalter DJ, Best NG, Carlin BR, van der Linde A, 2002. A Bayesian measures of model complexity and fit. J R Stat Soc Ser B 64 :583–616.

    • Search Google Scholar
    • Export Citation
  • 18

    Cressie NAC, 1993. Statistics for Spatial Data. New York: John Wiley & Sons, Inc.

  • 19

    Dolo G, Briet OJ, Dao A, Traore SF, Bouare M, Sogoba N, Niare O, Bagayogo M, Sangare D, Teuscher T, Toure YT, 2004. Malaria transmission in relation to rice cultivation in the irrigated Sahel of Mali. Acta Trop 89 :147–159.

    • Search Google Scholar
    • Export Citation
  • 20

    Koudou BG, Tano Y, Doumbia M, Nsanzabana C, Cisse G, Girardin O, Dao D, N’Goran EK, Vounatsou P, Bordmann G, Keiser J, Tanner M, Utzinger J, 2005. Malaria transmission dynamics in central Côte d’Ivoire: the influence of changing patterns of irrigated rice agriculture. Med Vet Entomol 19 :27–37.

    • Search Google Scholar
    • Export Citation
  • 21

    Klinkenberg E, Takken W, Huibers F, Toure YT, 2003. The phenology of malaria mosquitoes in irrigated rice fields in Mali. Acta Trop 85 :71–82.

    • Search Google Scholar
    • Export Citation
  • 22

    Mouchet J, Brengues J, 1990. Agriculture-health interface in the field of epidemiology of vector-borne diseases and the control of vectors. Bull Soc Pathol Exot 83 :376–393.

    • Search Google Scholar
    • Export Citation
  • 23

    Doannio JM, Dossou-Yovo J, Diarrassouba S, Rakotondraibe ME, Chauvancy G, Chandre F, Riviere F, Carnevale P, 2002. Dynamics of malaria transmission in Kafine, a rice growing village in a humid savannah area of Côte d’Ivoire. Bull Soc Pathol Exot 95 :11–16.

    • Search Google Scholar
    • Export Citation
  • 24

    Temel T, 2004. Malaria from the gap: need for cross-sector cooperation in Azerbaijan. Acta Trop 89 :249–259.

Save