Malaria prevalence data were collated from surveys of childhood populations in Mali since 1960. Altogether 101 such surveys were identified yielding suitable estimates of malaria Background Good maps of malaria risk have long been recognized as an important tool for malaria control. The production of such maps relies on modelling to predict the risk for most of the map, with actual observations of malaria prevalence usually only known at a limited number of specific locations. Estimation is complicated by the fact that there is often local variation of risk that cannot be accounted for by the known covariates and because data points of measured malaria prevalence are not evenly or randomly spread across the area to be mapped. MethodWe describe, by way of an example, a simple two-stage procedure for producing maps of predicted risk: we use logistic regression modelling to determine approximate risk on a larger scale and we employ geo-statistical ('kriging') approaches to improve prediction at a local level. Malaria prevalence in children under 10 was modelled using climatic, population and topographic variables as potential predictors. After the regression analysis, spatial dependence of the model residuals was investigated. Kriging on the residuals was used to model local variation in malaria risk over and above that which is predicted by the regression model. ResultsThe method is illustrated by a map showing the improvement of risk prediction brought about by the second stage. The advantages and shortcomings of this approach are discussed in the context of the need for further development of methodology and software.

Sizes of young field-collected males of Glossina morsitans morsitans Westw. were compared with those of teneral males emerging in the laboratory from field-collected puparia. Based on the mean size, indications were that smaller flies in field populations were selected against for about seven months in the year. A method is described for estimating the proportion of flies lost in the field population, and for the calculation of standard deviations for the estimates. Estimates of the extent of elimination showed that up to 35·2% of the total fly population in the field was eliminated in the cool months and up to 75·5% in the hot dry months.

An evaluation of sampling techniques was conducted on Highland Sourveld in the Natal Drakensberg. The quadrat, Levy bridge, step point, wheel point, metric belt transect and 't Mannetje & Haydock methods were used. In order to compare these methods the scores were standardized against the wheel point. The results of analyses of variance showed that the quadrat and point techniques were the most consistent. The metric belt transect and 't Mannetje & Haydock methods were shown to be not suitable for detailed botanical analysis. Operator differences and time of day had little effect on the results. It is concluded that the wheel point method is most suitable for determining grassland species composition in the Natal Drakensberg.

No abstract

Some subtle difficulties in optimal design are highlighted by the example of unreplicated field trials laid out on plots with spatial errors defined by uniformity trials. There is a dual problem of the arrangement of control plots and maximizing the number of test-line entries. A simulation study is conducted by randomizing the allocation of genotypes to the plots of four uniformity trials in accordance with the rules defining a number of competing designs. Results are summarized in terms of the 'SE ratio', which reflects the improvement in precision of a given design relative to a completely random design on the same plots. The definition of the SE ratio overcomes problems induced by differential shrinkage and consequent precision of test and control lines. A general result applying to all designs shows a curvilinear improvement in SE ratio with increasing error degrees of freedom of the design. The actual arrangement of check plots is of less importance than their increasing number, which contributes to increasing error degrees of freedom. Overall measures, including expected genetic gain, are used to illustrate the choice of a balance between the total number of test-line entries and the error degrees of freedom.