1KARI/DFID NARP II, PO Box 14733, Nairobi, Kenya; 2CAB Africa Regional Centre, PO Box 76520, Nairobi, Kenya; 3Natural Resources Institute, Chatham Maritime, Kent ME4 4TB, UK

Background and objectives
Napier grass (Pennisetum purpureum) is the major source of fodder for zero-grazed livestock in Kenya. In the early 1990s a new disease of the grass was brought to the attention of the Ministry of Agriculture by farmers. The pathogen was identified as Ustilago kamerunensis H Sydow and Sydow [1]. It had been reported as a minor problem from other East African countries but has proved to be more severe and widespread in Kenya (affected plants produce a smutted inflorescence and biomass production is much reduced). This may reflect the presence of a more virulent strain, less-resistant Napier grass clones, or a greater use of the grass in Kenya. The grass has been heavily promoted for stall-fed animals since the 1960s, with a consequent increase in the area under cultivation. The overall objective of this investigation is to develop management strategies for Napier grass smut for smallholder farmers through an understanding of the pathogen's biology and ecology.

Materials and methods
A stratified two-stage random sample of 109 Napier grass plots was done in 1997 in a high-potential area (Kiambu District) of central Kenya, using qualitative and quantitative methods.

Results and conclusions
Of 3925 Napier plants examined, 171 (4.4%) had smut disease. The proportion of plots infected was 25%. Many farmers did not recognize that infected plants were diseased (they assumed that infected plants were a different species) and so the importance of the disease has probably been underestimated. Napier plants were wholly smutted (all stems on a stool infected) or partially smutted, in that only some of the stems showed signs of disease. It seems that this pathogen is partially systemic. The fungus is able to initiate disease within a stem but not to infect other stems on the same stool, i.e. infection is vertically systemic but not horizontally systemic. Experience suggests that removing diseased parts of a stool leads to disease remission within that stool, and could be recommended as a control measure. If plants can be partially smutted, it should be possible to relate the number of smutted stems to a loss value for biomass and hence develop a scoring system for smut disease severity. Spatial pattern analysis revealed non-clustering of plants in 72% of the plots, suggesting that plant-to-plant disease transmission is of minor importance, and that diseased plants do not act as major foci of infection within a plot. This suggests that the infection behaviour of U. kamerunensis is analogous to that of soilborne pathogens. A model of the proportion of smut was developed, based on a quadratic function of altitude, as the only significant parameter. This is the first published smut disease correlative model that uses environmental factors. To develop management advice for farmers, future work will investigate sources of inoculum (spore dispersal and distribution of infected planting material) and factors influencing the spread of the disease (such as soil amendments), and build on existing farmer practice in roguing diseased plants. It may be possible to recommend a fallow period if spore survival in soil is short. Longer-term control measures, as part of a basket of options for farmers, could involve resistant clones, alternative fodders such as Guatemala grass (Tripsacum fasciculatum), or fungicides. However, with such a traditionally low-input crop, farmers require methods of influencing the disease that are free or cheap.

1. Mordue JEM,1993. International Mycological Institute Biosystematics Services Report H380/93/YE1.