Recent reports from the Intergovernmental Panel on Climate Change (IPCC, 2001a; 2001b) indicate that global average temperatures may rise by 1.4–5.8 ^OC in the present century.
     Rainfall and temperature changes may have substantial impacts on (among other things) agriculture and human health in Africa, and Africa is the continent most vulnerable to the impacts of projected changes because existing widespread poverty limits adaptation capabilities. In a preliminary attempt to assess the possible impacts of these climate changes on production systems, we carried out some analyses for Africa related to changes in the length of the growing period that might be anticipated over the next five decades.
     The LGP defines a period when crop production is possible based on temperature and moisture limitations—roughly when the precipitation exceeds half the potential evapotranspiration (FAO, 1978). Previously, McDermott et al. (2001) had estimated a current LGP surface for SSA based on long-term monthly climate normals (rainfall, daily temperature and daily temperature diurnal range) from over 7000 stations (Jones, 1987). A 10 minutes-of-arc surface was fitted, based on the NOAA data set TGOP006 (NOAA, 1984) using inverse square distance weights for spatial interpolation, and a correction for elevation effects. For each grid cell, a simple daily water balance model was run, interpolating the monthly climate normals to daily values using a fast Fourier transform (Jones, 1987). To derive a predicted LGP surface for 2050, mean values of maximum and minimum temperature and precipitation for 2041–2070 were obtained from the IPCC Data Distribution Centre on the worldwide web (http://ipcc-ddc.cru.uea.ac.uk/). These predictions are based on a greenhouse-gas-only experiment conducted at the Hadley Centre using the Unified Model (Cullen, 1993) at a resolution of 2.5 by 3.75 degrees of latitude and longitude. These data were then interpolated to a 10-minute pixel size (Jones and Thornton, 2000) and the water-balance model rerun to produce the 2050 LGP surface
     Map 16a shows the difference between the 2000 and 2050 LGP surfaces. Major reductions in LGP are predicted for areas of West Africa, southern Sudan, Uganda and some areas of Ethiopia, and increases are indicated in southeastern Kenya, northeastern Tanzania, southern Cameroon and other areas of Ethiopia. 
     To see how this might bring shifts in the cropping boundary, Map 16b shows changes in areas related to a 60-day LGP, which we assume to be the point at which transition can occur between pastoral systems (<60 days LGP) and agro-pastoral systems (>60 days LGP). Because of the general predicted decline in LGP for Africa, the total area that will experience a drop in LGP below 60 days is greater than the areas that will see a gain in LGP above 60 days. Areas that would be more suitable for pastoral systems in 2050 are predicted to stretch in a band across the Sahel and Sudan, the transition zones to lower elevations in Ethiopia, and a band cutting across southern Angola and central Zimbabwe. Areas that could move from pastoral to mixed systems are predicted to be located primarily in Kenya, Tanzania and Ethiopia.
     To highlight potential changes within mixed systems, we have mapped the movement of the boundary for growing maize, which we defined at 120 days LGP. Map 16c shows a similar pattern in movement to that observed for the pastoral/agro-pastoral transition zones. Most of the potential reduction in areas suited for growing maize is predicted to occur in the Sahel and Sudan. Kenya is predicted to see some modest gain in area that is suitable for growing maize. Interestingly, some of the expansion here is expected to occur in districts in which poor households rely significantly on the sale of maize for their income from crops.
    Data from the 1994 Kenya Welfare Monitoring Survey (Kenya Government, 1998; 2000) show that the districts where poor households had the highest share of monthly crop income from the sale of maize are located either in areas we classified as mixed systems or rangeland/ livestock-based systems. For the mixed systems, most of these districts include the transition zones to low suitability for growing maize. For example, the proportion of maize sales in mean monthly crop income for poor households is as follows: West Pokot 99%, Uasin-Gishu 97%, Kericho 91%, Trans-Nzoia 93%, Narok 90%, Nandi 73%, Elgeyo-Marakwet 64%, Laikipia 63%, Baringo 50% and Bomet 46%.Three districts in the rangeland only, livestock-based systems also have a high proportion of crop income from maize sales for poor households (Wajir 96%, Mandera 80% and Marsabit 72%). The dependence of poor households on maize, often in more marginal production environments, parallels the general shift in Kenya away from traditional, drought-resistant crops, expansion into areas with higher rainfall variability, and a strategy by farmers to produce a significant marketable surplus during a favourable growing season at high risk, relying on food aid as a type of crop insurance (see, for example, FEWS In-Depth Report—June 27, 1996, Kenya in Depth, at: http://www.fews/org/fb960627/fewsidl.html). 

     Expected changes in production systems to 2050 in Africa are shown in Map 16d particularly noteworthy are the predicted shifts in West Africa from range land-based systems to mixed, and in East and Southern Africa, the disappearance of mixed highland systems.