Synthesis of simple predictive models for river fish yields in major tropical rivers


River fisheries are difficult to assess, with tributaries often spreading over vast areas. Although many people in developing countries benefit from their proximity, difficulties in assessing the scale of exploitation arise because subsistence fishermen fish purely for their own consumption and catch does not enter a market or trade centre.

Previous analysis of limited African river data has produced simple predictive relationships between fish catch and physical features such as river length and basin area. These relationships, which have been subsequently refined using additional data, have provided indications of yield from river systems in Africa. However, the collation and analysis of data for other river systems was required in order to derive predictions for tropical rivers in other continents.


The project developed simple predictive relationships comparable to those for African river systems, through compiling river data from tropical regions other than Africa into a database. Data was sourced from the scientific literature, particularly regional journals sourced from the Institute of Freshwater and the FAO Fisheries Library. A database was constructed to facilitate data storage and retrieval for later analysis, providing a tool for future assessments of river or floodplain systems. To accompany the database, an application program was produced to enable user entry of new data and references, editing of data already entered and output of data suitable for analyses.

Simple predictive relationships were derived using simple and multiple linear regressions. The analysis included a limited examination of morphometric relationships of river basins but was mainly confined to deriving the relationships of total annual fish catch to physical, edaphic, hydrological and social indices.


The project successfully refined earlier relationships for predicting fish yields in tropical rivers. A primary database and accompanying manual have been produced, containing annotated statistics on physical, morphological, hydrological, edaphic, fisheries and demographic data for 45 rivers in South America and Asia.

Most significantly, fish catch has been shown to be positively correlated with river discharge rate, providing a further series of empirical relationships based on information which is fairly widely available. It also, however, has considerable biological implications for interpreting the basis of production in rivers and opens up the way to providing a dynamic model for management of tropical river fisheries centred upon the hydrological cycle.

Given information on drainage basin area, river length, flood plain area or mean discharge rate, it is now possible to provide a first order planning estimate for fish production in South American, African or in a more generalised fashion, Asian rivers. Multiple regressions amongst the three physical variables with respect to fish catch have also been produced in order to increase the precision of the estimate.

The database highlights gaps in the knowledge of tropical rivers, the biggest being for Asian rivers, and so can be used to assign future research priorities.

FMSP Ref: 


Project Documents: 


Project Dates: 

Thursday, 1 April, 1993 to Tuesday, 31 May, 1994