Use only of data that are comparable Africa-wide
As a follow-up to the work of Kapetsky (1994), this study only used data that are comparable for the whole of Africa. This approach limited the number of factors that could have been used for the evaluation, but it enabled comparisons among countries. More importantly, this methodology enables development work to be planned and executed by evaluating the needs and similarities among countries or regionally. To this end, the primary objectives of this study - to stimulate improved planning for fish farming development at national levels, and at the same time to provide a tool to plan comprehensively for technical assistance by FAO and other national and international organisations - were achieved.
The present study identified potential areas for fish farming development, and it is at this stage that the assessment can allow planners to select countries or regions in which to conduct more detailed studies by bringing more factors into the analysis and taking advantage of data at resolutions that are higher than are appropriate at a continental scale. For example, addressing cage culture, and socio-economic criteria are good options for priority setting. However, such exercises are perhaps better addressed at a national-level, rather than in a continental study such as the present one, because socio-economic conditions vary greatly both between countries and within individual ones and it is unlikely that simplified weightings that would apply to all of Africa can be easily derived.
Fish farming potential
The results of this study are generally positive for the development of inland fish farming in Africa. There are apparently vast areas in Africa that have potential for both small-scale and commercial farming of Nile tilapia, African catfish and Common carp without serious constraints, either from the quality of the land or from constraints of temperature on fish growth. The most significant finding was that the final fish farming potential estimates for the three fish species together show that about 15% of the African surface contains areas with the highest suitability score for both small-scale and commercial farming.
The culture system and the yield models appear to work well separately as well as together to give reasonable estimates for fish farming potential. Therefore, refinements and not reconstruction of the models are needed to improve predictions of this study.
In correspondence to the earlier fish farming studies (Kapetsky, 1994; Kapetsky and Nath, 1997), primarily because of the continental scale used, the resolution and the assumptions employed, the estimates of fish farming potential generated in this study are essentially indicative of aquaculture potential. Not all of the area that has been identified as having potential can be allocated to fish farming as some land areas may already be used for other uses such as buildings and roads.
The estimates of area with potential (or lack of potential) for fish farming development were influenced by many factors. Some of these factors originate from the inaccuracy of the data, their spatial and temporal availability, the analytical approach and the underlying assumptions adopted. However, the increasing availability of data and the greater capabilities of computer technology continue to expand the potential role of GIS, and many of the problems affecting this study's results will be minimised or eliminated as more data becomes available, and more experience is gained with aquaculture-oriented GIS.
Thresholds
Although strong efforts were made to use objective thresholds, because the majority of the thresholds were identified through literature research (e.g., slope classes for fish ponds) and guidance from expert staff at FAO, there is some subjectivity in the results. Nonetheless, this feature is to some extent inevitable as "scoring" involved interpretation of data. Interestingly, this feature actually allows flexibility in GIS modelling, while at the same time introducing subjectivity, and the balance between the two is an important consideration in practical decision-making.
Weights
Although more field studies need to be carried out to further reduce the subjectivity of weights (or relative importance of submodels in the small-scale and commercial models), the group of experts generally agreed in choosing the appropriate weights. Moreover, the rank order of the weights and even the weights themselves of this study were quite similar to the Latin America study (Kapetsky and Nath, 1997) implying that the weight selections were based on sound decisions. For example, the urban market size and proximity submodel in this study was ranked first in importance for commercial farming and given a weight of 46% which is very similar to the value of 49% used in the Latin America study.
Verification
Overall, verification results proved to be in close agreement with the GIS predictions. There was a good correspondence between the predictions of small-scale and commercial suitability combined with the yield potential, to the locations of existing fish farms. Because none of the existing farms were located in areas scored as unsuitable, these results suggest that reasonable confidence can be placed on the predictions of the suitability of the grid cell "sites" for fish farming.
Future applications
Food security
A related use of the present results would be to update the study by Kapetsky (1995) on the potential contribution of fish farming to overall fish supply and to food security in Africa.
Evaluation and management of inland fisheries
Some of the primary data or submodels will be used directly or will be re-assessed and developed as inputs to spatial models for the evaluation and management of inland fisheries. This work is underway by the Inland Water Resources and Aquaculture Service (FIRI) at FAO led by Dr. J.M. Kapetsky. In short, the kinds of models that will be created as outlined by J.M. Kapetsky (pers.comm.) are:
1. Fishery potential under wild or natural conditions.
2. Fishery potential under various kinds of enhancements in order to estimate the likely
increases in benefits compared with unenhanced fisheries.
3. Losses of potential yield due to general environmental degradation, so that mitigation
or rehabilitation is properly scaled and financed.
Significant refinements to the data developed in the present study will be made by using "a catchment approach" to spatially delineate factors that directly or indirectly affect fishery potential. The primary data from the present study that will be used are: inland water bodies, soils, precipitation, temperature, roads, population density and land cover.
