Predictive models of larval habitats from topography have in common that the resolution of the DEMs used is too coarse. Other studies indicate that the Anopheles mosquitoes (malaria vector) are keen to breed in irrigated valleys utilized as rice fields. Therefore, irrigated rice fields could be potential breeding sites for Anopheles while puddles smaller than can be analysed from the commonly used DEM’s are found suitable as well. The question that emerges is what topographic information would be more suitable for the geo data based identification of breeding sites; A high resolution DEM, as it could derive smaller topographic features, the locations of rice fields, being a special type of topographic feature or local inhabitants’ perception on topographic features of stagnant water.
The Malaria Elimination Program in Ruhuha (MEPR) is a project to eliminate the malaria infection rate in the Ruhuha sector, Bugusera, Rwanda. Questionnaires have been conducted to analyse local inhabitant’s view towards malaria. Furthermore entomological surveys have been executed to count the number of mosquitoes at 10% of all the households in the sector.
The first aim of this study was to find whether there are any spatial patterns (by clustering) within the MEPR entomological surveys’ data. Additionally, the study focussed on DEM derivatives to see whether there is a correlation between clusters of (Anopheles) mosquito counts and these DEM derivatives. For both a low-resolution ASTER DEM (30m) and a high-resolution DEM (10m) from the RNRA, correlations have been calculated. These results have been compared to see if it gave a more accurate result. Besides deriving topographic features from DEMs well delineated areas, rice fields, have been associated with the entomological data. Finally the perception of topographic features of stagnant waters by local inhabitants was explored to find spatial patterns in relation to the entomological clusters.
Two global and two local clustering methods assessed that none of the data from the entomological surveys was globally significantly clustered. However, local clusters were present in the data. All of the clustering occurred between 250m and 1500m. High mosquito-count clusters over 1500m that became evident from all analysis tools were located in the west- and northern part of the Ruhuha sector. In contradiction to Kernel Density Estimations, where the clusters were defined by the number of mosquitoes present, the Local Convex Hulls technique created hulls around positive sites and thus created sharper clusters. Nevertheless the locations of both these clusters concurred between the two methods. Overall the DEM derivatives were not directly found to be explanatory in bivariate analyses of any clusters.
Average (Anopheles) mosquito densities in 250m bins from the rice fields showed a parabolic relation, with increasing mosquito densities until a distance of 1000m outside the rice fields boundaries. The results from comparing human perception- and clusters of mosquitoes were not univocal though there was a tendency towards a positive relationship.
Local human perception could possibly be used to identify Anopheles breeding sites as local inhabitants are able to identify topographic features that cause water stagnation. However, the inventory technique caused for a subjective bias in the data. The DEM derivatives, at any resolution, were not sufficient to identify the topographic features related to mosquito’s habitats for the Ruhuha sector’s case. Overall the high resolution DEM gave slightly higher correlations. Irrigated valleys, utilized as rice fields, had the strongest relation to mosquito densities. Therefore this special type of topographic feature seems the most favourable mosquito habitats in the Ruhuha sector.
One of the most important factors that could have influenced the results is the entomological data used. An interesting alternative dataset would be to sample for both adult mosquitoes and larval breeding site in other areas as well, especially near rice fields.
Keywords: Malaria vector (Anopheles), breeding habitat, topographic features, Digital Elevation Models, rice fields, human perceptions, clustering, Kernel Density Estimation, Local Convex Hull.