| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Our goal in this project has been to better understand the relationship between rice irrigation and malaria in a focal irrigation project of Mali, the Office du Niger, more specifically, why it is that malaria transmission is (generally) lower there than in nearby nonirrigated villages. This might assist the management of current irrigated areas and the design of new ones.
|
Two large, intensive studies of the relationship between irrigation and disease have been conducted by the Malaria Research and Training Center (MRTC) at the University of Bamako in Mali. The first of these included three irrigated and three nonirrigated villages, described in Refs. 7, 8, and 19. They observed that malaria was quite seasonal in the nonirrigated villages, being concentrated in the wet season. Anophelines were rare there during the dry season, so there was little malaria then. In irrigated villages, however, especially those where a second crop of rice was grown, mosquitoes and malaria were observed through much more of the year. Nonetheless, the total incidence of malaria transmission in these irrigated villages was much smaller than in nonirrigated ones. For example, the average entomological infection rate (EIR) was 8.7 infective bites per month for the nonirrigated villages but only 2.4 infective bites per month for the irrigated villages. This paralleled the numbers of clinically diagnosed cases of malaria in children at the same times.8
In an effort to understand this surprising relation, Dolo and others7,19 considered the several hypotheses listed above and ultimately favored the competition hypothesis, hypothesis no. 4. They could rule out the species/form composition hypothesis, no. 3, because the composition was substantially the same in both regions. Bed nets were used almost universally in both regions, as determined from surveys, ruling out that explanation, no. 2. Generally, the same medical care was available throughout the region, at the time of this study, suggesting that differences in health care, no. 1, was unlikely to account for the differences in malaria transmission; though actual use of services could be affected by social conventions and distance to the health care facility. This left the competition hypothesis, no. 4, as the remaining possibility. It did receive some support in their study; for example, they observed that during the season of peak density, adult survivor-ship was less than at other times. Also, the sporozoite rates and vectorial capacity were lower when densities were high. These are all expected from the competition hypothesis but of course do not prove it. Our goal in this study was to examine further the claims of the competition hypothesis in the same irrigated area. Specifically, we analyzed data that had been collected in a follow-up study from a larger number of irrigated villages than in the Dolo and others7 original study (3) and over a different period of time, 1999–2001. We measured correlations between mosquito densities and survival rates, zoophilic rates, and vectorial capacity (C) over the villages and times of this expanded study. The question was, would vectorial capacity quantitatively follow mosquito densities in these additional locations and times in the same manner that Dolo and others7,19 had hypothesized it should?
|
METHODS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
A Landsat 7 ETM+ false color image of the Niono area is shown in Figure 1. This was obtained during October 1999. The color scheme is red for ETM+ band 5, green for ETM+ band 4, and blue for ETM+ band 3. In their original study of malaria transmission in Niono, Dolo1 included three nonirrigated villages (Toumakoro, Dokoboukou, and Kalanampala) and three irrigated villages (Niessoumana, Tenegué, and Tissana). The nonirrigated villages are shown in blue. The study reported here did not include any nonirrigated villages but did include the three irrigated villages studied in Ref. 1 plus 15 additional ones. There were thus 18 villages throughout the irrigated area: 8 in the Niono subzone (Nango, Tigabougou, Niessoumana, Tissana, Tenegue, Koyan Coura, Kolodougou Coura, and Mourdian), 5 in the N’Débougou subzone (Siengo, Nara, Toukoun Coura, Sounkalokan, and Sarango), and 5 in the Molodo subzone (Sokourani, Molodo, Niono Koroni, Coccodi, and Hamdalaye). The selection criteria included that the villages be at least 2 km apart to decrease the likelihood of capturing mosquitoes coming from the ricefields of a neighboring village, accessibility and the willingness of villagers to cooperate with the mosquito collection efforts. The fact that villages were located in different subzones increased the variability in the numbers of mosquitoes captured, as each subzone was irrigated independently and had been subjected to different levels of rehabilitation.
|
A useful and widely adopted estimate of a vector population’s ability to transmit malaria is the vectorial capacity, C. It may be described as follows24: Let the number of vectors per human be m and the number of bites per mosquito per night on humans be a, then a human is bit ma times per day, on average. Assuming an exponential survival rate with daily survival p, then a proportion pn of these vectors survives the incubation period (sporozoite cycle) of the parasite, so that it could then transmit the pathogen. The vectors are then expected to survive another 1/log p days and bite other persons a times per day, on average.
Combining this, a reasonable measure of transmission is the vectorial capacity, defined to be:
 |
These terms and how they were measured are summarized in Table 2.
|
Night captures were conducted at two houses in each village, at least 200 m apart, between 6 PM and 6 AM with a personnel change at midnight. At each house, a collector was posted indoors and another outdoors with a flashlight and a mouth aspirator.
Collected females were classified by abdominal status (unfed, fed, semigravid, and gravid) in the field when possible. At high density, specimens were conserved in Carnoy’s fixer (3 parts ethanol:1 part glacial acetic acid), and classification was conducted in the laboratory. The number of human occupants during the previous night was recorded for each surveyed house. To estimate the proportion of the blood-fed and semi-gravid An. gambiae s.l. and An. funestus that had fed on humans (anthropophilic rate), a blood aliquot was extracted, conserved in Carnoy’s, and analyzed with enzyme-linked immunosorbent assay (ELISA).25 Parity rates were estimated from the night catches, using the method of Ref. 26, observed the day after capture. Much of this has been compiled in the Ph.D. thesis of Mahammadou Touré21; further details may be found there.
Statistical methods. We used population-averaged panel data regression models (XTGEE models, STATA 8.0, Stata Corporation) for all analyses.27 These groups of models are extensions of generalized linear models (GLMs) that are useful to model count data, as the user can specify a negative binomial distribution for the response variable (e.g., mosquito counts). In addition, they allow the specification of a variable over which observations are not independent (either in space or over time), called the clustering variable. We used village as the clustering variable to correct for the correlation among mosquito samples taken in houses within the same village. In this way, we could relate the density measure done at the house level to the response variables measured at the village level (parity and anthropophily).
We considered the different surveys as independent, given the large temporal separation between the samples, and assumed no spatial dependency among the villages sampled, given the minimum separation of 2 km and the fact that cultivation patterns are driven mainly by irrigation patterns, which are not expected to vary consistently with distance.
We first fit a regression model with total number of mosquitoes Nt as the response variable and species, season, and year as predictors. In the second set of models, we tested the effect of human-biting rates ma and survey on parity and anthropophilic rates. A negative relationship between ma and either parity or anthropophily would decrease the rate at which vectorial capacity increases with density and could even cause C to decrease at very high densities. Finally, we used a second-degree polynomial multiple regression model to examine the relationship between ma and vectorial capacity estimated per village. We compared linear and quadratic models to see if a nonlinear density term significantly improved the fit.
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The highest An. gambiae Nt was in the middle of the main cropping season (August) in both years, followed by the off-season crop (April). In April 2000, Nt was significantly lower than in April 1999, coincident with a shortage of irrigation due to work on channel maintenance. An. funestus Nt also showed significant differences among seasons and years, with the highest abundance in October and April of both years. Given these differences between the 2 years of the study, we will hereafter refer to surveys (N = 8) instead of years/seasons. We performed all subsequent analyses for each vector species separately.
Figure 2 shows the number of indoor resting An. gambiae females, Nt, for each village and collection. Each column represents one collection, and each row shows the values per village. The indoor resting density of An. gambiae is coded in color, ranging from a low of 0 (black) to a high of 80 (red). This large variability was found even within the same survey/village (e.g., in August 1999, the numbers captured in one house in Toukoun Coura was 0, whereas in another house it was 2,487). Even with this large within-village variability, the differences in mosquito numbers were statistically significant for both An. gambiae (
2 = 155.40, df = 17, P < 0.001) and An. funestus (2 = 409.32, df = 17, P < 0.001).
|
summarizes the mean values of the malaria parameters obtained in each of the surveys. Human biting rate (ma) and anthropophilic rates (A) showed a significant negative relationship, as illustrated in Figure 3a and in Table 4, part a. There was also a statistically significant but weaker relationship between density and daily survivorship estimated through parity rates (Figure 3b and Table 4, part b). To control for the effect of the time of the year in these relationships, we included survey as a covariate in the previous models. Inclusion of survey improved the R2 to 0.33 for parity rates and to 0.49 for anthropophilic rates. Human biting rates still had a significant effect on anthropophily when the effect of survey was included, though not on parity rates.
|
|
|
). We found that adding the squared term improved the fit and that this effect was statistically significant.
|
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. At lower densities, C increases with density, it levels off at intermediate densities, and decreases at higher densities. In principle, this might go a long way toward explaining the varied response of malaria transmission to irrigation projects that are summarized in Table 1.
It should be recognized, however, that there is a great deal of variation about the curve in Figure 1. Although the relationship is statistically significant, one could question whether it is sufficient to explain the fairly large differences in malaria that were observed. A core part of the competition hypothesis is that body size decreases with larval density and that smaller adult mosquitoes survive less well. Support from this comes from Refs. 28, 29, and 30. We have begun experiments to test this, and preliminary results do suggest statistically significant, but again small, effects in support of the competition hypothesis. A full evaluation will require that all of the observed mathematical relationships be combined into a model and their respective roles compared.
It appears that the strongest relationship between density and vectorial capacity comes from a, the man-biting habit, or number of bites per human per night (Figure 3a). Although it is not hard to imagine reasons why this might occur, the most obvious is certainly that when densities are high, people protect themselves, and the mosquitoes are forced to alternate hosts. Human night catches have been made in this area, and during the rainy season the number of bites per person may exceed 550 bites/night by An. gambiae.7,19 Dolo and others7 report near-universal use of bed nets, based on surveys, but a more detailed study might be warranted.
We used adult mosquitoes from indoor PSC catches as the measure of mosquito density. This might not be correct for testing the competition hypothesis. There is ample reason to believe that competition in An. gambiae occurs in the immature stages.31,32,33 But what is important for the competition hypothesis is the numbers relative to available resources. We are aware of no good way to assess accurately how many resources are available or to measure how good they are for An. gambiae. What makes one patch of water suitable and another not? Human experts can often do a pretty good job of guessing where larvae will be found, but they are very frequently in error. Chemical composition of the water and its physical appearance both seem important. It would be helpful to have better means for doing this, both for studies of this sort and for determining species composition (Edillo FE and others, unpublished data).
Related to this issue of measuring suitability of habitats is the contentious issue of density-dependence among immature An. gambiae. Surely there is density-dependent survival in laboratory and pseudonatural circumstances,30 but more work needs to be done to demonstrate the same in nature. Larval densities in such studies (e.g., Refs. 30 and 31) are typically much greater that those commonly observed in natural water (DiukWasser MA and others, unpublished data). Service33 has reviewed much of this work, citing a variety of ways the question has been addressed and referencing several dozen studies that found evidence for or against density dependence. He could conclude only that " . . . it may be difficult to prove the existence, or otherwise, of density-dependent population regulation" (p. 757).
Finally, vectorial capacity is essentially a measure of transmissibility—the potential to transmit malaria. It is not the same as malaria transmission itself (more accurately estimated by EIR) and certainly not the same as numbers of clinical malaria human cases. Dolo and others7 found that the seasonal changes in the sporozoite index in the irrigated zone correspond well to the seasonal changes in anthropophily and parity ratios. Seydou Doumbia and co-workers at MRTC are currently compiling data about EIR in the serial studies of the villages used in this study. It is likely that they will bear directly on the observations reported here.
As to the medical relevance of malaria transmission indices, Sissoko and others8 compared the EIR to diagnosed cases of malaria in the three irrigated and three nonirrigated villages that were studied by Dolo and others.7 They observed a high correlation between the two in nonirrigated villages, though it was less so in the irrigated ones.
Received August 11, 2004. Accepted for publication December 6, 2004.
Acknowledgments: We would like to thank the collaboration of the Niono Health Center, the Office du Niger, the Institute dEconomie Rurale, Niono supervisors and village guides, and Yeya Touré, Robert Gwadz, and the members of the MRTC GIS Laboratory in Bamako for their advice and help.
Financial support: This work was supported by the National Institutes of Health (NIH), the National Aeronautic and Space Administration through an Interagency Agreement Y3-AI-5059-03 with the National Institute of Allergy and Infectious Diseases, and WARDA Africa Rice Center.
Authors’ addresses: Maria A. Diuk-Wasser, Department of Epidemiology and Public Health, Yale School of Medicine, 60 College St., New Haven, CT 06510. Nicholas Manoukis and Charles E. Taylor, Department of Ecology and Evolutionary Biology, University of California, Los Angeles, 621 Charles Young Drive, Los Angeles, CA 90095-1606. Mahamoudou B. Touré, Guimogo Dolo, Nafomon Sogoba, and Sékou F. Traoré, Malaria Research and Training Center, Faculté de Médecine, de Pharmacie and
Odonto-Stomatologie, Université du Mali, Bamako, B.P. 1805, Mali.. Magaran Bagayoko, Conseiller sousrégional pour l’Afrique Central, Organisation Mondiale de la Santé, Bureau OMS du Gabon, BP 820, Libreville, Gabon.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. J. Muturi, S. Muriu, J. Shililu, J. Mwangangi, B. G. Jacob, C. Mbogo, J. Githure, and R. J. Novak Effect of Rice Cultivation on Malaria Transmission in Central Kenya Am J Trop Med Hyg, February 1, 2008; 78(2): 270 - 275. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Sogoba, S. Doumbia, P. Vounatsou, I. Baber, M. Keita, M. Maiga, S. F. Traore, A. Toure, G. Dolo, T. Smith, et al. Monitoring of Larval Habitats and Mosquito Densities in the Sudan Savanna of Mali: Implications for Malaria Vector Control Am J Trop Med Hyg, July 1, 2007; 77(1): 82 - 88. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. SOGOBA, P. VOUNATSOU, S. DOUMBIA, M. BAGAYOKO, M. B. TOURE, I. M. SISSOKO, S. F. TRAORE, Y. T. TOURE, and T. SMITH SPATIAL ANALYSIS OF MALARIA TRANSMISSION PARAMETERS IN THE RICE CULTIVATION AREA OF OFFICE DU NIGER, MALI Am J Trop Med Hyg, June 1, 2007; 76(6): 1009 - 1015. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. DIUK-WASSER, M. B. TOURE, G. DOLO, M. BAGAYOKO, N. SOGOBA, I. SISSOKO, S. F. TRAORE, and C. E. TAYLOR EFFECT OF RICE CULTIVATION PATTERNS ON MALARIA VECTOR ABUNDANCE IN RICE-GROWING VILLAGES IN MALI Am J Trop Med Hyg, May 1, 2007; 76(5): 869 - 874. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
