Climate variability and maize production in Sub Saharan Africa

Climate variability and maize production in Cameroon: Simulating the effects of extreme dry and wet years Munang Tingem1, Mike Rivington2 and Jeremy Colls1 1Agriculture & Environmental Science Division, School of Bioscience, University of Nottingham, Nottingham, UK 2Macaulay Institute, Craigiebuckler, Aberdeen, Scotland Correspondence: Munang Tingem (email: m.tingem@googlemail.com) The effects of interyear variability of extreme rainfall events on maize yields at locations in Cameroon, in central-west sub-Saharan Africa were investigated through a simulation assessment combining a weather generator with a crop growth model. This study analyzes the potential of using dry/wet year predictions to reduce risk in subsistence agricultural production associated with climate variability at the site level. Weather data sets from eight provincial study localities were classified into three precipitation scenarios – dry (lower threshold), normal and wet (upper threshold) years. According to the modelling results, there is a less than 12 per cent variance in mean maize yields across six out of the eight localities when planting occurs in March, May and August. The variance is equivalent to approximately 100–300 kg per ha, which represents a significant amount of food in the household security of the majority impoverished sectors of rural and urban society, and which could greatly impact the socioeconomic activities of the entire populace. The results lead to the conclusion that all extreme dry and wet years are not equal in terms of their regional manifestation. This calls for precise monthly and sub-seasonal local level forecasts and the effective dissemination of this information to farming communities in Cameroon, thereby facilitating the adaptive management of indigenous cropping practices and reducing their vulnerability to climate related disasters. Keywords: agriculture, climate variability, CropSyst, maize yield, Cameroon Introduction Agricultural production in sub-Saharan Africa including Cameroon is predominately a rainfed subsistence system (Tadross & Hewitson, 2005; Stige et al., 2006). Year-to-year variation and extreme climatic conditions attributed to climate variability invariably result in acute food deficit and affect the fortunes of the majority farming population (Ati et al., 2002; Tadross & Hewitson, 2005). This poses risks and uncertainties to agricultural communities in this region where food production per capita has decreased over the last twenty years (Sultan et al., 2005). Rainfall which produces available soil moisture, and thus controls the utilisation of water by crops is by far the most important climatic factor that influences the pattern and productivity of rainfed agriculture in sub-Saharan Africa (Amissah-Arthur, 2003). In a general context climate variability is virtually synonymous to rainfall variability. Access to relevant knowledge and climate forecast information is key for farmers to cope with or adapt to climate variability (Challinor et al., 2007), but year-to-year variability in yield associated with interannual climate variability is not well studied (Mearns et al., 1997; Phillips et al., 1998; Tsvetsinskaya et al., 2003). In order to provide farmers with information specific to their local conditions there is a need to understand and document how rainfall and yield in different regions within a country varies during different extreme events (Amissah-Arthur et al., 2002). This calls for research and pilot doi:10.1111/j.1467-9493.2008.00344.x Singapore Journal of Tropical Geography 29 (2008) 357–370 © 2008 The Authors Journal compilation © 2008 Department of Geography, National University of Singapore and Blackwell Publishing Asia Pty Ltdstudies to characterize and quantify the effects of climate variability at different spatial scales so as to increase the resilience of farmers to the risks posed by climate variability in sub-Saharan African countries, Cameroon amongst them. Maize (Zea mays L.) is the staple crop in Cameroon. It is consumed mainly as a dry fermented dough (fufu), and in multiple other forms including corn beer (kwacha), which is produced and sold in huge quantities in villages and towns, forming an important contribution to the local economy and thus reduces hunger among the rural poor. Because of its multiple uses, maize provides the bases for financial profitability after the devaluation of the CFA franc in January 1994 (Adesina & Coulibaly, 1998; Ngoko et al., 2002). Smallholders form around 40 per cent of the rural population and any crop yield depreciation affects survival needs of the majority. Many farmers – both men and women alike – lack the capacity to adapt farming practices due to their difficult socioeconomic conditions, the insufficient provision of institutional supports and technical assistance and, inadequate infrastructure (e.g. micro-loan facilities and long-range forecasting systems). Because of inadequate research and training in the agricultural sector, the unavailability of early warning systems and lack of access to climate forecasting, domestic food production and the livelihood of subsistence farmers remain vulnerable to climate variability (Molua, 2006). Resource-poor farmers in Cameroon, as elsewhere in sub-Saharan Africa, have traditionally relied on indigenous knowledge to cope with climate hazards; forecasts of the onset and cessation of the rainy season and quantity of rain are based upon observations and interpretation of natural phenomena (e.g. the height of ants’ nests in trees, colour of frogs and so on), but rainfall patterns that depart from normal variations such as a run of wet years followed by one of dry years, are often unanticipated. Farmers’ overriding concerns are achieving household food security. However, population growth coupled with climate variability pose serious challenges on food security in Cameroon, which point to the need to realign and adopt new strategies that contribute to greater resilience of the agricultural sector. Modelling assessments help to estimate crop production under variable climatic conditions and in so doing support practical decision making on crop management issues taking into account climate variability. These assessments can be done combining weather generators and crop growth simulation models (Hoogenboom, 2000; Utset et al., 2006). Application of these simulation models in agricultural systems ideally requires observed long-term daily weather data. Observed short-term data represents just one realization of the climate, but many realizations are needed to provide a wider range of feasible situations and allow a good estimation of the probability of extreme events. Deterministic mathematical models (known as stochastic weather generators) that simulate time series climatic variables have addressed this (Richardson & Wright, 1984). The weather generators use observed weather data as inputs to generate data, which are statistically similar to the observed data (Hoogenboom & Soltani, 2003). Using a crop simulation model (CropSyst) and a weather generator (ClimGen), this paper aims to first, determine the proportion of interannual maize yield variability that could be ascribed to climate variability; second, characterize the rainfall events that result in the most substantial negative impact on maize yield; third, define options for farm improvements and appropriate strategies in order to optimize yields and reduce risk of crop failure; and finally, provide a benchmark of yield variability against which future climate change induced variability may be compared. 358 Munang Tingem, Mike Rivington and Jeremy CollsAn understanding of the patterns of dry and wet years/cycles in the region will also provide some important insights into issues of management of food resources during bumper crop years to minimize the effects of famine and food shortages during poor harvest years. Materials and methods Weather data Daily maximum temperature, minimum temperature, and rainfall data from 1979–2003 were obtained from the University Cooperation for Atmospheric Research (UCAR) (http://dss.ucar.edu/datasets/) for the eight provincial localities chosen for study that would provide as complete a representation of the agricultural environments in Cameroon . Since we are evaluating long-term effects of climate variability on maize yield, it was necessary to expand the temporal range of the weather data for use in the crop model to allow a good estimation of the probability of extreme events. We used the ClimGen software (version 4.1.05; http://www.bsyse.wsu.edu/climgen/) to produce generated climatic data to supplement the 1979–2003 UCAR data. The model requires inputs of daily series of weather variables (minimum and maximum temperature, precipitation) to calculate parameters used in the generation process for any length of period at a specific location. ClimGen is well documented elsewhere (e.g. Castellvi & Stöckle, 2001; Stöckle & Nelson, 2003). According to Richardson (2000), long data records are needed to obtain stable representative estimates – at least 10 years of weather data, for estimation of temperature parameters and 20 or more years for the estimation of the precipitation parameters. Hence, we used the UCAR observed 25-year historic daily records of temperature and precipitation to generate a further 25 years’ modelled daily weather data to extend the coverage to 50 years for each study locality. The generated weather data series was compared with the observed weather data for its distribution of monthly mean and variance of precipitation and temperature using the t-test and F-test respectively. Table 1 summarises the results of the series of statistical comparisons for all the test localities. ClimGen performed well in simulating the range of monthly mean precipitation and temperature values at the test localities. Wet and dry years classification Consideration of dry and wet years is important in agricultural studies since both can significantly affect crop growth and yields. In this study, wet years were distinguished from dry years using the 10th and 90th percentiles of yearly accumulated precipitation distributions as indicators to identify upper and lower threshold values (Table 2), since rainfall under or above these percentiles can be considered as extreme (IPCC, 2001; Unganai, 2002; Sanchez et al., 2004; Utset et al., 2006). The 50-year baseline (25-year observed plus 25-year generated) continuous time series of daily air temperature and precipitation for each locality were sorted into dry and wet years. A year was classified as wet (WetBase) if the total rainfall at the site was above the 90th percentile (upper extreme) and as a dry year (DryBase) if the total rainfall was below 10th percentile (lower extreme). The years within these two thresholds were classified as normal years (NormalBase). Study area Cameroon is a tropical country located in the sub-Saharan region of central-west Africa (Figure 1). The country displays highly contrasting physical and biogeographical Climate variability and maize production, Cameroon 359Table 1. Comparison of the observed (Obs) precipitation (mm) and maximum temperature (Tmax) means and variances with those of generated 25-year data (Gen) at eight localities, and the probability levels (p-value) calculated by the t-test and F-test for the monthly mean and variance (0.05 or lower indicates a departure from the observed data significant at 5 per cent level). Bamenda Batouri Garoua Kribi Maroua Ngaoundere Tiko Yaounde Precipitation Obs mean 195.8 123.2 83.1 219.5 65.9 124.7 266.5 135.7 Obs variance 25 185.0 5375.9 8425.6 22 020.0 6842.8 12 638.0 38 782.2 8083.0 Gen mean 176.1 134.7 88.3 226.6 72.3 126.1 260.3 153.6 Gen variance 11 202.4 8506.1 7600.7 19 342.7 7646.4 11 874.9 41 799.1 7593.5 p-value t-test 0.724 0.740 0.888 0.838 0.854 0.976 0.940 0.624 p-value F-test 0.097 0.229 0.434 0.417 0.429 0.460 0.452 0.460 Tmax Obs mean 23.8 29.6 34.9 30.0 34.5 29.0 30.0 28.4 Obs variance 2.93 2.50 9.27 2.05 8.81 4.00 3.25 2.47 Gen mean 24.7 29.1 34.3 30.0 33.5 28.1 30.1 27.8 Gen variance 2.96 3.35 8.89 2.00 8.08 3.83 3.30 2.33 p-value t-test 0.188 0.457 0.617 0.933 0.447 0.287 0.937 0.329 p-value F-test 0.493 0.318 0.473 0.484 0.445 0.472 0.490 0.463 Table 2. The calculated 10th and 90th percentile of accumulated precipitation for 50 years in the eight study localities. Location Latitude (°N) Longitude (°E) Elevation (m) Annual rainfall (mm) 10th percentile 90th percentile Bamenda 6.05 10.1 1239 2378 1658 2213 Batouri 4.47 14.37 656 1499 1447 2062 Garoua 9.33 13.38 244 1090 806 1289 Kribi 2.95 9.89 16 2634 2375 3094 Maroua 10.44 14.25 422 834 629 1072 Ngaoundere 7.34 13.57 1104 1514 1233 1810 Tiko 4.08 9.37 52 3198 2512 3389 Yaounde 3.83 11.51 760 1655 1553 2122 360 Munang Tingem, Mike Rivington and Jeremy Collsfeatures. The climate, reflecting the topography and latitudinal range is diverse but comprises two principal zones: the equatorial zone (2–5° N), encompassing the southern and the mountainous western part of the country, that manifests a classic Guinean type climate, and the tropical zone encompassing the rest. Three regional subtypes are discerned. First, the seaboard (e.g. provincial capitals Kribi and Tiko) with abundant annual rainfall (2634 and 3198 mm respectively). Second, the inland areas (e.g. the national capital Yaounde), with total annual rainfall of less than 1660 mm, characteristic of the southern part of the south Cameroon plateau, extending into the east of the country around Batouri. And third, the Sudan-Sahel subtype (tropical climate) north of 5° N, where annual rainfall decreases from 1513 mm to 834 mm northward near Lake Chad. The mean temperature of Cameroon ranges from 22° to 29 °C, increasing from south to north and from the coast to the hinterland. [Detailed regional climatic differences reflecting this spatial variation at the eight study localities are provided at Appendix 1.] The maize staple crop growing season in Cameroon is related to the rainy season and planting is fine-tuned to very specific times of the year. In the equatorial zone consisting N N I G E R IA C H A D C E N T RAL AF R I CAN R E P U B L I C GAB O N Bight of Biafra 0 2000km N CAMEROON A F R I C A Kribi YAOUNDE Batouri NgaoundereGaroua Maroua Tiko Bamenda 0 50 100 150km Annual rainfall range (mm) 500–1000 1000–1500 1500–2000 2000–4000 South East Centre Adamawa North Far North Southwest Northwest Figure 1. Map of Cameroon showing the annual rainfall of the eight provincial study localities. Climate variability and maize production, Cameroon 361of the localities Bamenda, Batouri, Kribi, Tiko and Yaounde, there are two plantings coinciding with the two rainy seasons: the March planting heralding the ‘long’ March to July season, and the August planting for the ‘short’ August to November season. In the tropical zone localities – Garoua, Maroua and Ngaoundere – where there is only one rainy season, from May to October, planting starts in May (Ndemah, 1999). Table 2 provides the coordinates for each locality. Soil and crop production Representative soil properties (thickness and texture) for the simulation points were extracted from the International Soil Reference and Information Center database (http://www.isric.org) (Batjes & Bridges, 1994; Batjes, 1995). Agrodata (e.g. yield, phenological parameters and so on) for the maize staple was readily available from the Central Bureau of Statistics’ published district reports for the chosen provincial study localities (AGRISTAT, 2001). Crop growth model CropSyst (http://www.bsyse.wsu.edu/cropsyst), the multiyear,multicrop, daily time step cropping system simulation model (Stöckle et al., 2003) used in this study has been applied and used extensively to simulate crop growth and yield for a range of crops such as wheat, maize, soybean, sorghum, groundnut and forage crops in diverse environments includingWest and South Africa (Badini et al., 1997; 2007; Abraha&Savage, 2006). It has also been used in detailed studies of maize cropping (i.e. Jara & Stöckle, 1999; Bellocchi et al., 2002) and has been shown to be robust and accurate for a diverse range of local environments, including those found within Cameroon. The different submodels included in CropSyst are described with a similar level of detail, so it is a well-balanced crop simulator, simulating different crops from a common set of parameters. Cropsyst simulates the soil water budget, soil–plant nitrogen budget, crop canopy and root growth, crop phenology, dry matter production, yield, residue production and decomposition, and erosion. The main inputs are daily weather data and the model allows the user to specify management parameters such as sowing date, cultivar genetic coefficients (photoperiodic sensitivity, duration of grain filling, maximum leaf area index, and so on), soil profile properties (soil texture, thickness, water and initial nitrogen content), fertilizer and irrigation management, and tillage. Crop growth is simulated for the whole canopy by calculating unstressed (potential) biomass based on crop potential transpiration and on crop intercepted photosynthetically active radiation. This potential growth is then corrected by any water and nitrogen limitations to determine actual daily biomass gain. The simulated yield is then obtained as the ratio between actual total biomass accumulated at physiological maturity and crop-specific harvest index (harvestable yield/aboveground biomass) (Monteith, 1981; Tanner & Sinclair, 1983). The simulation of crop development is based on the thermal time required to reach specific development stages. Thermal time is calculated as growing degree days (GDD, °C day) accumulated throughout the growth cycle (starting from planting until harvest). Average air temperature above a base and below a cut-off temperature is considered for GDD calculations. The accumulation of thermal time may be accelerated by water stress. Water balance processes in CropSyst include rainfall, runoff, interception by the crop canopy and residues, infiltration, redistribution in the soil profile, crop transpiration and soil evaporation. Potential evaporation is estimated by the Priestley-Taylor method (1972) and water dynamics in the soil is handled by the Richard’s equation, which is solved numerically using the finite difference technique (Stöckle et al., 2003). 362 Munang Tingem, Mike Rivington and Jeremy CollsCropSyst has data requirements that can be reasonably met and provides support utilities to substitute for missing parameters based on well established procedures (e.g. using pedotransfer functions to derive soil hydraulic parameters). Hence, it provides a conceptually unified modelling system for many crops, minimizing the dangers of structural uncertainty in making both cross crop and interspatial comparisons (Rivington et al., 2006). As such it ably represents the variation in yield determined by weatherdriven environmental conditions and respond to specific management regimen. In this study, CropSyst was applied for estimating final potential maize yields using the 50-year weather datasets. Total yields were matched with values from CropSyst corresponding to WetBase, NormalBase and DryBase. Calibration of crop parameters were based on typical values obtained from the literature and default values from the CropSyst user manual. A number of parameters accounting for cultivar-specific differences were calibrated based on outputs of development and growth characteristics. Further parameterization was achieved by minimizing differences between actual and simulated yields for a limited number of simulation trials using available field reported data. Remaining parameters were adjusted within a reasonable range as provided by the manual (Stöckle & Nelson, 2003). The calibration of the phenological parameters (e.g. growing degree days) was made using data provided by the Institute of Agricultural Research (IRA), Cameroon. Simulations were run for a two-year maize-maize rotation with sowing dates set to 15 March (the 74th day of the calendar year) and August 15 (the 227th day) in Bamenda, Batouri, Kribi Tiko and Yaounde, and to 15 May (135th day of the year) in Garoua, Maroua and Ngaoundere, which agrees with the traditional crop management practices in these two climate zones (Ndemah, 1999; Molua, 2003). Corresponding to the observed maximum maize root length, a 1-m soil depth was considered to simulate the soil water balance (Farré, 1998; Farré & Faci, 2005). The finite difference soil water balance function, by which water moves up and down depending on the soil water potential of vertically adjacent layers, was used for the redistribution of water in the soil under nonlimiting soil fertility. An implementation of the Priestley-Taylor (1972) equation was used to compute the reference evapotranspiration. Forty per cent of crop residue was assumed to remain in the field after harvest for recycling purposes (Abraha & Savage, 2006). No irrigation was used as this is not a common practice. Results The coefficients of variation (CVs) values of yield for the March, May and August plantings were computed over the entire time-series available at each of the respective localities (discussed further below). The per cent CV represents a measure of the farmer’s risk, low CVs indicate stable year-to-year production, while high CVs denote high inter-annual variability (Rosenzweig & Tubiello, 2007). The parameterization of CropSyst was deemed to be successful at each locality, as estimates showed that simulations of maize yield agreed within 0.1–1.9 per cent with the AGRISTAT-reported yields across Cameroon (Figure 2). Hence the model can be seen as robust under the diverse range of environmental conditions found within Cameroon. Statistical test using t-test and F-test (5 per cent level of significance) indicated there were no significant differences between the generated values and observed weather data; thus representative long-term weather data of precipitation and temperatures could be generated from historical weather data using ClimGen. This finding has particular relevance for agricultural modelling applications in Cameroon where the Climate variability and maize production, Cameroon 363limited observed record would make it otherwise difficult to evaluate long-term effects of weather on crop yield. March planting Table 3 shows that year-to-year variations in maize productivity is lowest at Tiko for the March planting (average CV 6.5 per cent) and generally also at Kribi and Yaounde. Under March plantings the dry scenarios introduced higher CVs in Bamenda and Batouri, where there was a large variation even under the NormalBase scenario (25.2 and 37.5 per cent respectively). At these two localities the WetBase scenario resulted in a decrease in CV from that of the NormalBase. The highest CVs across all scenarios occurred in both Batouri (38.2 per cent) and Bamenda (31.5 per cent) with the DryBase data. Generally the changes in CV were very low between all three scenarios at Kribi, Tiko and Yaounde. Although higher simulated yields would be expected because of not taking weeds, pests and pathogens into account in the modelling, the results reported here can be taken as probable indicators. May planting May plantings at Garoua, Maroua and Ngaoundere (Table 4) showed low interannual yield variability across all three scenarios where yield amounts remained consistent. Maroua showed the highest interannual yield variability (12.3 per cent under the WetBase scenario), but with only a 7.9–10.7 per cent range in average CV between the three localities. Only at Garoua did the mean yield decrease under the WetBase scenario, possibly due to excess rainfall resulting in nitrogen losses from the soil profile during the growing season. Maroua and Ngaoundere showed no real change in CV under the DryBase scenario, whilst Garoua evidenced a slight increase which was associated with only a very small change in mean yield from the NormalBase value. The magnitude of variance was similar to that found at Kribi, Tiko and Yaounde for the March planting, but slightly higher than for the August planting at these localities. August planting For the August planting Batouri produced a CV of 31.5 per cent under the dry scenario, indicating the severity of risk in this region. The CV reduces across all five localities for Bamenda Batouri Garoua Kribi Maroua Ngaoundere Tiko Yaounde 1000 1200 1400 1600 1800 2000 2200 2400 Maize grainyield (Kg/ha) Simulations Observed Figure 2. Calibration simulations versus observed crop yields (AGRISTAT, 2001) under current climate grouped by provincial locality. 364 Munang Tingem, Mike Rivington and Jeremy Collsthe WetBase scenario, most noticeably at Batouri (down to 4.2 per cent) which is also associated with a 400 kg/ha increase in mean yield (Table 5). Maize planted during August showed little variability in yield estimate (except for Batouri), but with a very slight increase in mean yield under the WetBase scenario. As with the March planting, there is little change in CV for maize planted in August between the three scenarios at Kribi, Tiko and Yaounde. Generally, the average CV for Bamenda, Batouri, Kribi, Tiko and Yaounde varied between 6.5 and 23.3 per cent for the March planting and 4.0 and 18.7 per cent for the August planting (Table 3). For the May planting at Garoua, Maroua and Ngaoundere the average CV varied between 7.9 and 10.7 per cent (Table 4). Discussion This study highlights the agricultural impacts of climate variability and provides a tool for identifying/investigating alternative management strategies that can reduce agricul-Table 3. Mean and coefficient of variation (CV) of simulated maize yields in Cameroon during March planting for the baseline scenario. Location March planting Mean yield (kg/ha) CV (%) Average CV (%) DryBase NormalBase WetBase DryBase NormalBase WetBase Bamenda 1126 1253 1314 31.5 25.2 13.3 23.3 Batouri 1085 1241 1419 38.2 37.5 20.0 31.9 Kribi 1860 1791 1854 8.8 8.5 6.4 7.9 Tiko 2392 2423 2413 7.8 5.8 5.7 6.5 Yaounde 2170 2138 2196 5.9 8.2 7.0 7.0 Table 4. Mean and coefficient of variation (CV) of simulated corn yields in Cameroon during May planting. Location May planting Mean yield (kg/ha) CV (%) Average CV (%) DryBase NormalBase WetBase DryBase NormalBase WetBase Garoua 1900 1940 1862 10.2 7.2 9.2 8.9 Maroua 2137 2156 2180 9.9 10.0 12.3 10.7 Ngaoundere 2302 2281 2329 7.6 7.3 8.8 7.9 Table 5. Mean and coefficient of variation (CV) of simulated maize yields in Cameroon during August planting. Location August planting Mean yield (kg/ha) CV (%) Average CV (%) DryBase NormalBase WetBase DryBase NormalBase WetBase Bamenda 2101 2045 2102 3.2 7.0 5.0 5.1 Batouri 1630 1800 2032 31.5 20.4 4.2 18.7 Kribi 2125 2094 2047 7.5 8.6 7.1 7.7 Tiko 2585 2590 2604 6.8 9.1 7.9 7.9 Yaounde 2687 2696 2709 4.9 4.0 3.0 4.0 Climate variability and maize production, Cameroon 365tural risk in Cameroon and elsewhere. It also provides a benchmark of yield variability against which future climate change induced variability may be compared. Year-to-year variability in rainfall reflected in the yield estimate and variance of maize at each locality shows that some, most noticeably Kribi, Tiko and Yaounde, have relatively low (<10 per cent) interyear variation between wet and dry scenarios for both March and August plantings (Tables 3 & 5). Similarly, for the three localities with planting in May there was very little change in both yield amount and variability. The level of variance between March and August plantings for Kribi, Tiko and Yaounde was similar for the May plantings (Table 4) at Garoua, Maroura and Ngaoundere. This indicates a reasonable stable environment for maize growth at these localities under current climate variability – but without consideration of biological (such as pests and pathogens) or practical (soil workability, management and the like) restrictions. Conversely, Batouri and Bamenda showed a high yield variance under a dry scenario for the March planting. The approach taken in this research indicates that some localities (illustrated by Bamenda) would benefit from adopting an August planting regime, with possible substantial reductions in variance risk. However, localities such as Batouri may still have substantial risks associated with yield variance regardless of when the crop is planted under a dry scenario, whilst also having high variance under normal and wet conditions. Generally, there is spatial variation in the response of the maize crop to different precipitation scenarios. Dry scenarios generally showed a slight decrease in maize yields (in the order of 100–300 kg/ha) and an increase in yield variability across locations, with some localities showing slight increases in yield of the same order. Yield changes under dry scenarios are due to varying degrees of water stress during critical phenological stages, which decrease grain numbers (Rosenzweig & Hillel, 1993). Conversely, wet scenarios produced on average a positive impact, by showing increasing yield and reduced yield variability, especially for the August planting at Bamenda, Batouri, Kribi, Tiko and Yaounde. In some areas yield reduced with increased variance under wet scenarios. This could be attributed to the uneven distribution of rainfall at these locations, where heavy rains might be received over a few days followed by a dry spell, with leaching of nitrogen from the soil profile. These findings complement previous simulation studies in Zimbabwe (Phillips et al., 1998) and in China (De Costa et al., 2006); maize yields showed a decreasing trend during growing seasons with decrease in precipitation. Olesen & Bindi (2002), Amissah-Arthur (2003), and Adejuwon (2005) also identified low precipitation as a reason for cereal production decrease in the arid and tropical areas. Using the CERES-maize (Crop Environment Resource Synthesis) model, Southworth et al., (2000) found variations of a similar type in maize yield estimates across years, explained by the precipitation changes during the growing seasons in the midwestern USA. For resource-poor subsistence farmers in Cameroon and sub-Saharan Africa, the implications of outstandingly good or bad years are profound. Forecasts of the quality (extreme dry or wet, or normal) and date of onset and end of the rainy season could aid in production planning and advisory services, and thus increase and enable the capacity of farmers to adapt to climate variability (Unganai, 2002). These results point to the need to downscale seasonal climate forecasts for local levels: knowing what the coming season is likely to be will provide farmers with the lead time to secure resources for planting as soon as the rains starts. The CVs under each scenario at each locality could be used as determinants/indicators by government and food security agencies to proactively increase assistance including food aid to tide the community over periods of hardship brought on by dry years. 366 Munang Tingem, Mike Rivington and Jeremy CollsThe simulated approach demonstrated here is useful and will assist in planning optimal strategies to help farmers reduce risks and increase productivity conditioned on climate information such as building food reserves/storage, providing disaster relief and crop insurance. Conclusion This work investigates the implications for subsistence rainfed agricultural systems of regional extreme dry/wet year events at the local level in Cameroon. This effort is directed at the application of seasonal climate forecast to agricultural management at the farm level to reduce risks in production of the maize staple associated with climate variability. The modelling approach taken indicates a less than 12 per cent variance in mean maize yields between wet, dry and normal climatic scenarios across six out of the eight study localities (whether planting occurs in March, May, or August); at two localities yield variation was substantially higher under dry conditions. However, the actual variability at all localities will be higher, given the modelling limitations for losses due to pests and pathogens. The results suggest that wetter conditions generally favoured yield amounts and reduced variability at most of the localities. Drier conditions considerably reduced yield and increased variability at Bamenda and Batouri for March plantings and in Batouri for August plantings as well. For locations where planting occurs in May, mean yield amounts were consistent, with little variation (7.6 to 12.3 per cent) between dry and wet scenarios. The anomalous conditions of extreme dry/wet events, and their demonstrated impact on crop yield and household food security among the majority subsistence farming population in sub-Saharan countries such as Cameroon, highlights the importance of seasonal forecasts at local farm levels to facilitate timely relief measures and avert the persistence of climate-related disasters such as famine. Seasonal climate predictions should be effectively disseminated in an appropriate form to end users who bear the brunt of climate variability on rainfed subsistence production systems. Using the conceptual framework outlined here, there is potential for identifying and enabling management strategies to reduce and buffer against climate-related agricultural risk in Cameroon and regions similarly affected by extreme dry/wet growing seasons. The results gained here provide a useful benchmark against which climate change induced variability can be compared. Investigation of yield response using a crop model confirms the potential for further studies of the impacts of longer-term climate change through use of future scenario projections. Acknowledgement We acknowledge the assistance of Claudio O. Stöckle and Roger L. Nelson (Biological Systems Engineering Department, Washington State University) in the application of CropSyst and ClimGen, as well as the contributions of Arnold R. Salvacion and Gianni Bellocchi. References Abraha MG, SavageMJ (2006) Potential impacts of climate change on the grain yield of maize for the midlands of KwaZulu-Natal, South Africa. Agriculture Ecosystems and Environment 115, 150–60. Adejuwon JO (2005) Food production in Nigeria. I. Present effects of climate variability. Climate Research 30, 53–60. Adesina AA, Coulibaly NO (1998) Policy and competitiveness of agroforestry-based technologies for maize production in Cameroon: an application of policy analysis matrix. Agricultural Economics 19, 1–13. Climate variability and maize production, Cameroon 367AGRISTAT (2001) Semi-annual bulletin of the statistics of the agricultural sector 2000/2001. Ministry of Agriculture and Rural Development, Yaounde. Amissah-Arthur A (2003) Targeting climate forecast for agricultural applications in sub-Saharan Africa: situating farmers in user-space. Climatic Change 58, 1–20. Amissah-Arthur A, Jagtap S, Rosenzweig C (2002) Spatio-temporal effects of El Nino events on rainfall and maize yield in Kenya. International Journal of Climatology 22, 1849–860. Ati FO, Stigter CJ, Oladipo EO (2002) A comparison of methods to determine the onset of the growing season in northern Nigeria. International Journal of Climatology 22, 731–42. Badini O, Stöckle CO, Franz EH (1997) Application of crop simulation model and GIS to agroclimatic assessment in Burkina Faso. Agriculture Ecosystems and Environment 64, 233–44. Badini O, Stöckle CO, Jones JW, Nelson R, Kodio A., Keita M. (2007) A simulation-based analysis of productivity and soil carbon in response to time-controlled rotational grazing in the West African Sahel region. Agricultural Systems 94, 87–96. Batjes N (1995) A homogenised soil data file for global environmental research: a subset of FAO, ISRIC and NRCS profiles (version 1.0). Working paper 95/10, International Soil Reference Information Center (ISRIC). ISRIC, Wageningen. Batjes N, Bridges E (1994) Potential emissions of radiatively active trace gases from soil to atmosphere with special reference to methane: development of a global data base (WISE). Journal of Geophysical Research 99, 16479–6489. Bellocchi G, Maestrini C, Fila G, Fontana F (2002) Assessment of the effects of climate change and elevated CO2: a case study in Northern Italy. In Proceedings of the VII European Society for Agronomy Congress, Cordoba, Spain. 15–18 July 2002, 763–64. Available at http://www.sipeaa.it/tools/ClimGen/ESA2002_ClimateChange.pdf/(accessed 29 August 2007). Castellvi F, Stöckle CO (2001) Comparing the performance of WGEN and ClimGen in the generation of temperature and solar radiation. Transactions of American Society of Agricultural and Biological Engineers (ASABE) 44, 1683–687. Challinor AJ, Wheeler TR, Garforth C, Craufurd PQ, Kassam A (2007) Assessing the vulnerability of food crop systems in Africa to climate change. Climatic Change 83, 381–99. De Costa WAJM, Weerakoon WMW, Herath HMLK, Amaratunga KSP, Abeywardena RMI (2006) Physiology of yield determination of rice under elevated carbon dioxide at high temperatures in a subhumid tropical climate. Field Crops Research 96, 336–47. Farré I (1998) Maize (Zea mays L.) and Sorghum (Sorghum Bicolor L. Moench) Response to Deficit Irrigation. Agronomy and Modelling. PhD dissertation, University of Lleida, Spain. Farré I, Faci JM (2005) Comparative response of maize (Zea mays L.) and sorghum (Sorghum bicolor L. Moench) to deficit irrigation in aMediterranean environment. Agricultural Water Management 83, 135–43. Hoogenboom G (2000) Contribution of agrometeorology to the simulation of crop production and its applications. Agricultural and Forest Meteorology 103, 137–57. Hoogenboom G, Soltani A (2003) Minimum data requirements for parameter estimation of stochastic weather generators. Climate Research 25, 109–19. Intergovernmental Panel on Climate Change (IPCC) (2001) Climate Change 2001: Impacts, Adaptation, and Vulnerability, A Contribution of Working Group II to the Third assessment Report of the IPCC. Cambridge University Press, Cambridge. Jara J, Stöckle CO (1999) Simulation of water uptake in maize, using different levels of process detail. Agronomy Journal 91, 256–65. Mearns LO, Rosenzweig C, Goldberg R (1997) Mean and variance change in climate scenarios: methods, agricultural applications and measures of uncertainty. Climatic Change 35, 367–96. Molua EL (2003) Global Climate Change and Cameroon’s Agriculture: Evaluating the Economic Impacts. PhD dissertation. Georg-August University, Germany. Molua EL (2006) Climate trends in Cameroon: implications for agricultural management. Climate Research 30, 255–62. Monteith JL (1981) Climate variations and the growth of crops. The Quarterly Journal of the Royal Meteorological Society 107, 749–74. 368 Munang Tingem, Mike Rivington and Jeremy CollsNdemah RN (1999) Towards an Integrated Crop Management Strategy for the African Stalk Borer Busseola Fusca (Fuller) (Lepidoptera: Noctuidae) in Maize Systems in Cameroon. PhD thesis, University of Hannover, Germany. Ngoko Z,Cardwell KF, Marasas WFO, Wingfield MJ, Ndemah, Schulthess F (2002) Biological and physical constraints on maize production in the Humid Forest and west highlands of Cameroon. European Journal of Plant Pathology 108, 893–902. Olesen JE, Bindi B (2002) Consequences of climate change for European agricultural productivity, land use and policy. European Journal of Agronomy 16, 239–62. Phillips JG, Cane MA, Rosenzweig C (1998) ENSO seasonal rainfall patterns and simulated maize yield variability in Zimbabwe. Agricultural and Forest Meteorology 90, 39–50. Priestley C, Taylor RJ (1972) On the assessment of surface heat flux and evaporation using large scale parameters. Monthly Weather Review 100, 81–92. Richardson CW (2000) Data requirement for estimation of weather generation parameters. Transactions of American Society of Agricultural Engineers 43, 877–82. Richardson CW, Wright DA (1984) WGEN: A model for generating weather variables. US Department of Agriculture, Agricultural Research Service, ARS-8, Washington, DC. Rivington M, Matthews KB, Bellocchi G, Buchan K, Stöckle, C.O. Donatelli M (2006) An integrated assessment approach to conduct analyses of climate change impacts on whole-farm systems. Environmental Modelling Software 22, 202–10. Rosenzweig C, Hillel D (1993) Agriculture in a greenhouse world. National Geographic Research and Exploration 9, 201–21. Rosenzweig C, Tubiello FN (2007) Adaptation and mitigation strategies in agriculture: an analysis of potential synergies. Mitigation and Adaptation Strategies for Global Climate Change 12, 855–73. Sanchez E, Gallardo C, Gaertnar MA, Arribas A, Castro M (2004) Future climate extremes in the Mediterranean simulated by regional climate model: a first approach. Global planetary Change 44, 163–80. Southworth J, Randolph JC, Habeck M et al. (2000) Consequences of future climate change and changing climate variability on maize yields in the Midwestern United States. Agriculture Ecosystems & Environment 82, 139–58. Stige LC, Stave J, Chan KS et al. (2006) The effects of climate variation on agro-pastoral production in Africa. Proceedings of the National Academy of Sciences 103, 3049–53. Stöckle CO, Donatelli M, Nelson R (2003) CropSyst, a cropping systems simulation model. European Journal of Agronomy 18, 289–307. Stöckle CO, Nelson R (2003) Cropping System Simulation Model User’s Manual. Washington State University, Pullman. Sultan B, Baron C, Dingkuhn M, Sarr B, Janicot S (2005) Agricultural impacts of large scale variability of the West African monsoon. Agricultural and Forest Meteorology 128, 92– 110. Tadross MA, Hewitson BC (2005) The interannual variability of the onset of the maize growing season over South Africa and Zimbabwe. Journal of Climate 18, 3356–72. Tanner, CB, Sinclair, TR (1983) Efficient water use in crop production: research or re-search? In Taylor HM, Jordan WR, Sinclair TR (eds) Limitations to Efficiency Water Use in Crop Production, 1–27. American Society of Agronomy, Madison. Tsvetsinskaya EA, Mearns LO, Mavromatis T, GaoW,McDaniel L, DowntonMW(2003) The effects of spatial scale of climatic change scenarios on simulated maize, winter wheat, and rice production in the south-eastern United States. Climatic Change 60, 37–71. Unganai L (2002) Application of long-range rainfall forecast in agricultural management: A review of Africa’s experiences. Proceedings of the International Forum on Climate Prediction, Agriculture and Development, 26–28 April, 13–25. International Research Institute for Climate and Society, Columbia University, Palisades. Utset A, Martinez A, Farre I, Cavero J (2006) Simulating the effects of extreme dry and wet years on the water use of flooding-irrigated maize in a Mediterranean landplane. Agricultural Water Management 85, 77–84. Climate variability and maize production, Cameroon 369Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature ( C) O Mean ppt Mean temp Bamenda Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature (C) O Mean ppt Mean temp Batouri Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature ( C)O Mean ppt Mean temp Garoua Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature (C) O Mean ppt Mean temp Kribi Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature (C) O Mean ppt Mean temp Maroua Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature (C) O Mean ppt Mean temp Ngaoundere Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature ( C)O Mean ppt Mean temp Tiko Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 100 200 300 400 500 600 700 0510 15 20 25 30 35 Rainfall (mm) Temperature (C) O Mean ppt Mean temp Yaounde Appendix 1: Mean monthly rainfall and temperature distribution at the eight study localities. 370 Munang Tingem, Mike Rivington and Jeremy Colls