This study only used data that are comparable across the world. This approach limited the number of important factors that could have been used in the evaluation (e.g. soil data), but it enabled comparisons to be made between countries based on consistent climate data. More importantly, this methodology enables crop experimentation and improvement to be planned and executed by evaluating the similarities and needs within and between countries and regions. To this end, the primary objectives of this study have been achieved. The methodology provides both an assessment of the productive potential of regions such as Africa, where the crop is widely grown but where quantitative evidence is scant, incomplete or inaccessible, and defines new regions not previously associated with bambara groundnut but where environmental factors indicate a potential for productive growth.
Much of the Mediterranean basin appears to provide the ideal agro-ecological environment for bambara groundnut with predicted biomass exceeding that in regions of sub-Saharan Africa that have been associated with the crop for centuries. Nevertheless, there remains considerable scope within its current distribution to increase bambara groundnut productivity by a clearer understanding of how factors such as the seasonal distribution of rainfall, daylength and range of temperatures influence the allocation of assimilates to pod yield. Ultimately, it is the expansion of production and consumption patterns for crops such as bambara groundnut both within and beyond their current areas of cultivation that will determine whether they become significant crops for the world or disappear.
In relation to underutilised crops, one of the major concerns of sponsors is the possibility that effort is wasted on species of unknown potential in locations of unknown suitability. At this stage, our preliminary assessment allows planners to select promising locations, countries or regions that justify more detailed studies that bring local factors into the analysis and taking advantage of data of higher resolutions.
Primarily because of the world-scale used and the resolution and the assumptions employed, the estimates of bambara groundnut potential generated in this study are essentially indicative of bambara groundnut potential. Clearly, as with any crop, not all of the areas that has been identified as having potential can be allocated to bambara groundnut cultivation as some of these are already occupied by protected areas, inland water bodies, buildings and roads.
Our analysis of the spatial productivity of bambara groundnut demonstrates how a weather data generator and a dynamic crop simulation model can be usefully linked in a GIS on a global scale. This integration allows an examination of spatially complex, non-linear, interacting environmental variables and their combined influences on crop yield. Consideration of the spatial variability of model inputs has at least two potential benefits. First, where site-specific information on particular crops is lacking - as with most underutilised species - this approach allows us to produce a reasonable assessment of likely productivity with minimal field data. Second, for all crops this approach allows us to assess the potential consequences of future climatic variability and change on agricultural production.
The major drawback with such activities is the paucity of input data. Generally, the availability of digital soil data is more limited than meteorological data. To build digital soils data is an expensive and time-consuming process (Reed and Whistler, 1990). At this stage, we have not applied soils data mainly because of their inaccessibility and processing requirements. The inclusion of soil information to explain local variations in crop yields is an important future objective. Another major limitation is that the model takes no account of the effects of pests and diseases on the likely yields of bambara groundnut at any particular locations. This is a major limitation that needs to be considered in any future developments.
Our crop model is still some way short of a being a generic and comparative model for underutilised crops. This is mainly because there are insufficient datasets available for any other underutilised species. However, the basic modelling strategy developed for bambara groundnut can be rapidly adapted to any other underutilised (or major) crop if suitable collaborators can be identified. Ideally, co-operation between growers, experimenters and modellers interested in a representative range of species would provide sufficient critical mass to test and refine the approach that we have taken for a single species. The output from such collaboration could result in a user-friendly package to benefit all interested partners with minimum cost and maximum speed. In particular, such activities that generate reliable and widespread information on underutilised species could be of greatest benefit to developing countries, where most of these species originate and grow.
The analytical system applied in this study consists of three main components; a weather data generator, a crop model, and GIS. The first two components can be continuously enhanced whenever more data are available. The updating of weather and crop data would enable us to review and refine the simulation approaches built into them. In particular, the limited datasets available in the development of the BAMnut model, especially for application across the world, are of most concern.
The strong effects of soil conditions, in particular water releases characteristics, on crop growth and development mean that better information on soil conditions as well as climate data and specific crop requirements is a priority for the future. It is also appropriate to identify those areas where soils are either highly productive or impoverished to establish likely yield limitations and potential yield ceilings within any specific agro-ecological environment. FAO's agro-ecological zoning (AEZ) methodology developed by FAO and the International Institute for Applied System Analysis (IIASA) already does this (www.fao.org/ag/agl/agll/aez.htm) and this methodology could usefully be applied in the future.
Many of the results presented in this study will help develop an interactive online information and mapping system for bambara groundnut that WAICENT has already begun to construct. The first phase of the Web site is finished and involved preparing and verifying the bambara groundnut collection by Begemann (www.dainet.de/genres/bambara/index.htm). For the second component a map for Africa was prepared showing 705 collection sites as points from which the user can click and retrieve data from Begemann's database. Then, five different thematic maps i.e. annual rainfall, climate, dominant soils, and topography for Africa were prepared that the user can select and view along with the collection sites. The next phase involves the map modelling exercise, so that the resulting maps from this study can be incorporated into the Web site. Furthermore, to make the GIS version of BAMnut constructed in this study accessible, it is intended to include this model in the Web site so that users will be able to create, online, different map outputs according to their chosen BAMnut model variables and, ultimately, for other underutilised crops.
The GIS version of BAMnut could be integrated into a Web-based Common Modelling Environment (CME) developed by Texas A & M University. The goal of the CME is to package a suite of models such as ASM, FLIPSIM, PHYGROW, EPIC, SWAT, APEX and NUTBALL into a unified environment that contains a matrix of mission-critical data files. These files would support policy-makers exploration of alternatives and emerging technology options. A detailed description of the CME is available at http://cnrit.tamu.edu/CME/. Collaborative work with the CNRIT team that is developing the CME has been established and preparations for adapting and testing the GIS version of BAMnut are currently underway.
As information and expertise on bambara groundnut expands, the user will be able to input landrace-specific information that will provide estimates of likely productivity and best management practices for the cultivation of local bambara groundnut genotypes at particular locations. The development of a wider `generic' methodology will provide comparative estimates of productivity for contrasting underutilised (and major) species at any location.
Evaluation of the BAMnut model across the world and its subsequent development into a generic framework for other underutilised crops is an activity that requires an international effort. However, for all underutilised crops, any realistic effort to expand their cultivation and establish improved genotypes for specific environments must be based on identifying those regions that are most likely to match their physiological requirements. To maximise the use of research effort the mapping exercise described in this report should be an essential pre-requisite to field-based research or breeding efforts on crops such as bambara groundnut. Subsequent field studies can serve the dual purpose of accelerating progress on each particular crop with less risk of crop failure and an improvement in the accuracy of the global mapping technique.
The widespread access of the methodology and outputs presented here to end-users and policy-makers will allow future developments to be demand-led by those most interested in the use of this information. In particular, comparison of the simulated yields of underutilised crops with existing yield maps for major crops will enable decision-makers to prioritise crop and cropping systems in terms of farmers needs and national benefits.
Finally, the approach described in this report is a first attempt to map the potential areas of cultivation of one underutilised food legume across the globe. However, the approach has wider implications in terms of food security and poverty elimination. Rather than consuming specific crops, people eat a range of agricultural products that provide nutritional compounds including proteins, carbohydrates and lipids. Thus, a development of the mapping strategy described here is that it enables the production of `nutritional maps' whereby the relative food values of different species could be assessed for individual countries or regions. In this way, policy-makers can utilise evidence of the likely yield and nutritional value of different species to design food security strategies based on the most appropriate crops to grow within each region.
