Rain-fed agriculture will remain the dominant source of staple food production and the livelihood foundation of the majority of the rural poor in sub-Saharan Africa (SSA). Greatly enhanced investment in agriculture by a broad range of stakeholders will be required if this sector is to meet the food security requirements of tomorrow's Africa. However, production uncertainty associated with between and within season rainfall variability remains a fundamental constraint to many investors who often over estimate the negative impacts of climate induced uncertainty. Climate change is likely to make matters worse with increases in rainfall variability being predicted. The ability of agricultural communities and agricultural stakeholders in SSA to cope better with the constraints and opportunities of current climate variability must first be enhanced for them to be able to adapt to climate change and the predicted future increase in climate variability. Tools and approaches are now available that allow for a better understanding, characterization and mapping of the agricultural implications of climate variability and the development of climate risk management strategies specifically tailored to stakeholders needs. Application of these tools allows the development and dissemination of targeted investment innovations that have a high probability of biophysical and economic success in the context of climate variability.

This study examines farmers' perceptions of short-and long-term variability in climate, their ability to discern trends in climate and how the perceived trends converge with actual weather observations in five districts of Eastern Province in Kenya where the climate is semi-arid with high intra-and inter-annual variability in rainfall. Field surveys to elicit farmers' perceptions about climate variability and change were conducted in Machakos, Makueni, Kitui, Mwingi and Mutomo districts. Long-term rainfall records from five meteorological stations within a 10 km radius from the survey locations were obtained from the Kenya Meteorological Department and were analysed to compare with farmers' observations. Farmers' responses indicate that they are well aware of the general climate in their location, its variability, the probabilistic nature of the variability and the impacts of this variability on crop production. However, their ability to synthesize the knowledge they have gained from their observations and discern longterm trends in the probabilistic distribution of seasonal conditions is more subjective, mainly due to the compounding interactions between climate and other factors such as soil fertility, soil water and land use change that determine the climate's overall influence on crop productivity. There is a general tendency among the farmers to give greater weight to negative impacts leading to higher risk perception. In relation to long-term changes in the climate, farmer observations in our study that rainfall patterns are changing corroborated well with reported perceptions from other places across the African continent but were not supported by the observed trends in rainfall data from the five study locations. The main implication of our findings is the need to be aware of and account for the risk during the development and promotion of technologies involving significant investments by smallholder farmers and exercise caution in interpreting farmers' perceptions about long-term climate variability and change.

a b s t r a c tWe estimate the potential value of general circulation model (GCM)-based seasonal precipitation forecasts for maize planting and fertilizer management decisions at two semi-arid locations (Katumani and Makindu) in Southern Kenya. Analyses combine downscaled rainfall forecasts, crop yield simulation, stochastic enterprise budgeting and identification of profit-maximizing fertilizer N rates and stand densities. October-February rainfall predictions were downscaled from a GCM, run with both observed and forecast sea surface temperature boundary conditions -representing upper and lower bounds of predictability -and stochastically disaggregated into daily crop model inputs. Simulated interactive effects of rainfall, N supply and stand density on yield and profit are consistent with literature. Perfect foreknowledge of daily weather for the growing season would be worth an estimated 15-30% of the average gross value of production and 24-69% of average gross margin, depending on location and on whether household labor is included in cost calculations. GCM predictions based on observed sea surface temperatures increased average gross margins 24% at Katumani and 9% at Makindu when labor cost was included. At the lead time used, forecasts using forecast sea surface temperatures are not skillful and showed nearzero value. Forecast value was much more sensitive to grain price than to input costs. Stochastic dominance analysis shows that farmers at any level of risk aversion would prefer the forecast-based management strategy over management optimized for climatology under the study's assumptions, despite high probability (25% at Katumani, 34% at Makindu) of lower returns in individual years. Results contribute to knowledge of seasonal forecast value in a relatively high-risk, high-predictability context; utility and value of forecasts derived from a GCM; and risk implications of smallholder farmers responding to forecasts.

More than 400 million people in the developing world depend on dryland agriculture for their livelihoods. Dryland agriculture involves a complex combination of productive components: staple crops, vegetables, livestock, trees and fish interacting principally with rangeland, cultivated areas and watercourses. Managing risk and enhancing productivity through diversification and sustainable intensification is critical to securing and improving rural livelihoods. The main biophysical constraints are natural resource limitations and degradation, particularly water scarcity and encroaching desertification. Social and economic limitations, such as poor access to markets and inputs, weak governance and lack of information about alternative production technologies also limit the options available to farmers. Past efforts to address these constraints by focusing on individual components have either not been successful or are now facing a declining rate of impact, indicating the need for new integrated approaches to research for development of dryland systems. This article outlines the characteristics of such an approach, integrating agro-ecosystem and livelihoods approaches and presents a range of empirical examples of its application in dryland contexts. The authors draw attention to new insights about the design of research required to accelerate impact by integrating across disciplines and scales.

This article summarizes the results from an analysis conducted to investigate the spatio-temporal variability and trends in the rainfall over Ethiopia over a period of 31 years from 1980 to 2010. The data is mostly observed station data supplemented by bias-corrected AgMERRA climate data. Changes in annual and Belg (March-May) and Kiremt (June to September) season rainfalls and rainy days have been analysed over the entire Ethiopia. Rainfall is characterized by high temporal variability with coefficient of variation (CV, %) varying from 9 to 30% in the annual, 9 to 69% during the Kiremt season and 15-55% during the Belg season rainfall amounts. Rainfall variability increased disproportionately as the amount of rainfall declined from 700 to 100 mm or less. No significant trend was observed in the annual rainfall amounts over the country, but increasing and decreasing trends were observed in the seasonal rainfall amounts in some areas. A declining trend is also observed in the number of rainy days especially in Oromia, Benishangul-Gumuz and Gambella regions. Trends in seasonal rainfall indicated a general decline in the Belg season and an increase in the Kiremt season rainfall amounts. The increase in rainfall during the main Kiremt season along with the decrease in the number of rainy days leads to an increase in extreme rainfall events over Ethiopia. The trends in the 95th-percentile rainfall events illustrate that the annual extreme rainfall events are increasing over the eastern and southwestern parts of Ethiopia covering Oromia and Benishangul-Gumuz regions. During the Belg season, extreme rainfall events are mostly observed over central Ethiopia extending towards the southern part of the country while during the Kiremt season, they are observed over parts of Oromia, (covering Borena, Guji, Bali, west Harerge and east Harerge), Somali, Gambella, southern Tigray and Afar regions. Changes in the intensity of extreme rainfall events are mostly observed over southeastern parts of Ethiopia extending to the southwest covering Somali and Oromia regions. Similar trends are also observed in the greatest 3-, 5-and 10day rainfall amounts. Changes in the consecutive dry and wet days showed that consecutive wet days during Belg and Kiremt seasons decreased significantly in many areas in Ethiopia while consecutive dry days increased. The consistency in the trends over large spatial areas confirms the robustness of the trends and serves as a basis for understanding the projected changes in the climate. These results were discussed in relation to their significance to agriculture.