Multiscale Analysis Of Landscape Data Sets From Northern Ghana

Coffea arabica, one of the economically most important crops worldwide, occurs naturally in the undergrowth of montane rainforests of Ethiopia. The study provides the first detailed ecophysiological investigations of wild coffee populations. It demonstrates the inter- and intra-regional variability in phenotypic and hydrological characteristics of wild coffee. The results reveal very different strategies of wild coffee seedlings for coping with drought stress. The ecophysiological diversity shows the importance of Ethiopian wild coffee populations as gene pools for future breeding programs, and underlines the need for an in-situ conservation strategy. The study includes recommendations for coffee forest management and the use of wild arabica coffee in Ethiopia.

Solanum quitoense (Lulo) from the northern Andes is a fruit with an attractive taste. It has a high export potential but its production is hampered by insufficient pollination. Studies on the reproductive system of Lulo revealed that, due to their poricidal anthers, the flowers need pollinators for self- and cross-pollination. Pollination by native bumblebees significantly increases fruit set and quality. Experiments show that colonies of Bombus atratus can be reared in captivity and used in Lulo fields for providing increased pollination services. Modeling based on data on colony development reveal larva/workers ratio and the supply of pollen as main parameters for improving the rearing method.

Soils of irrigated lands in the Aral Sea Basin are often plagued by high salinity,hampering profitable agriculture on these soils. In this context, the present study has three specific objectives: to identify techniques that enable a rapid estimation of soil salinity, to characterize its spatial distribution and to estimate its spatial distribution based on readily obtainable environmental parametersusing a Neural Network Model Approach. Topsoil salinity was highly variable atshort distances and terrain attributes were the most influential factors. The use of relationships between environmental attributes and soil salinity for upscaling spatial distribution of soil salinity from farm to district level proved to be satisfactory. A possible application is the development of salinity prediction toolsfor farm-level decision making.