Previous Page Table of Contents Next Page




R.L. Welcome
Fishery Resources Officer
Aquatic Resources Survey and Evaluation Service
Department of Fisheries
Rome, Italy


Simple but crude models of fish catch are useful for the formulation of broad policies for the management of river basins for fisheries. Rivers may be treated as sets from which the expected catch of a typical waterway may be extrapolated. Variations from the typical state are brought about by differences in morphology, nutrient richness, intensity of exploitation and also by year-to-year variations in the intensity of the flood. A series of relationships are given which indicate preliminary findings using this approach but much more precision is possible in the future following further research.


Des modèles simples mais non ajustés des captures de poisson sont utiles pour formuler les politiques générales de l'aménagement halieutique des bassins versants. Les cours d'eau peuvent être considérés comme des ensembles à partir desquels il est possible d'extrapoler les captures probables d'un cours d'eau typique. Les variations par rapport à une situation type sont imputables à des différences concernant la morphologie, la teneur en éléments nutritifs, l'intensité d'exploitation et les fluctuations annuelles de l'intensité des crues. L'auteur décrit une série de rapports fondés sur les résultats préliminaires de cette méthode; une plus grande précision pourra être ettente lorsque de nouvelles recherches auront été réalisée.


The effective management of inland waters for fisheries requires some knowledge of the size of the resource in any river or lake, its potential yield, and its state at its present state of exploitation. It also needs an understanding of other natural and maninduced processes taking place within the river basin, which may affect the quality and quantity of the water as well as the living aquatic organisms. This information is necessary initially to determine the level of investment in fishing technology, research and adminisstrative infrastructure appropriate to the size of the resource. Subsequently, it is needed to monitor the effects of any decisions taken for the management of both the fishery and the environment.

Estimates of the magnitude and yield of riverine fish stocks based on adequete exploratory surveys have been rare, although more recently systematic investigations have been carried out in some systems (e.g., the Kafue, University of Michigan, 1971). On the whole the information that is available has been drawn from examination of catch statistics combined with more or less qualitative assessments of the state of the stocks. Unfortunately catch statistics from rivers are often of low quality because of the difficulties inherent to collecting data from fisheries which operate from many small landings dispersed along a system that can traverse several countries. Because of the lack of concise information on many of the individual systems, attempts have been made to extrapolate general principles from those small groups of water bodies for which something is known. This paper reviews some of the models that have been suggested to describe the basic factors which determine the potential yield from tropical rivers.


Despite the inadequacies of the data, an analysis of the fish yield patterns from African rivers has given a fairly coherent picture of the factors involved in determining the catch that can be expected from any particular system (Welcome, 1976). In the rivers used in the analysis, all of which are moderately to heavily fished, a good relationship was found between the drainage basin area of the river system (A), and the catch obtained from it (C). Rivers without exceptionally large floodplains, i.e., those with “normal” development of their fringing floodplains, conform to the relationship C = 0.12A0.85. Basin area and the total length of the longest channel are themselves simply related and a relationship also exists for yield in t as a function of main channel length in km; Ct = 0.0033L1.95, or approximately the catch in kg equals 3.33 times the square of the length of the river from its source (Ckg = 3.33L2).

The catch in tons of any reach of river of length xkm at distance y from its source can be calculated from xCy = Cy+x - Cy, where values of Cy can be obtained from the preceding equation. At its simplest form where x = 1 km, this yields a theoretical equation that might be expected for any kilometer of river: Ckmy = 0.00641y0.95.


Estimates made using these relationships are of course averages over a number of systems from which actual values from individual rivers deviate widely. Such deviations can be attributed to two main natural causes, edaphic factors and morphological factors, as well as to differences in intensity of exploitation.

3.1 Edaphic factors

The richness of the waters in a river are to a certain extent summarized by their conductivity. A listing of the conductivities of representative African rivers (Welcomme, 1975) shows that while the smaller streams differ widely with respect to this parameter, the conductivities of longer rivers tend to be grouped towards the centre of the distribution, i.e., between 50 and 150 μ mhos. There are nevertheless considerable differences in conductivity, covering a range of between 8 μ mhos (Rio Negro, Brazil) to 900 μ mhos (Semliki river, Uganda) and black water rivers, such as the Zaïre are notably lacking in nutrients. Differences in conductivity accounted for about 60 percent of the variation from the predicted values of catch in the African rivers studied. This variation occurs both between systems and between rivers of the same system.

3.2 Morphological factors

As is suggested below, floodplains have a very similar potential yield per unit area. Thus, reaches of river or river systems with a greater area of floodplain not unexpectedly produce more fish. By taking into account the relative areas of floodplains it was possible to divide African rivers into two distinct sets each with its own regression line. Actually these sets are somewhat arbitrary, but as there is insufficient information on floodplain areas in most rivers relatively broad classifications are needed for the present. There was firstly a group of rivers with extensive floodplain systems - such as the Niger, the Senegal or the Ouémé, whose flooded areas cover between 2.5 and 3.8 percent of their total basin area. Secondly, there were those rivers with “normal” floodplain development, which are harder to define as the fringing plain is rarely measured, but where less than 1.5 percent of the basin area seems to be liable to flooding. When the catch of the major floodplain areas is subtracted from the total catch of rivers having extensive floodplains, the remaining catch conforms to the normal relationship for a river of the same basin area. When systems from elsewhere in the world are compared with these relationships they conform reasonably well. The lower Mekong, the Danube and the Magdalena river, with floodplain areas of between 3 and 8 percent of their respective basins, are distributed around the extensive floodplain regression line, whereas the Mogi Guassu and Indus rivers, with no extraordinary development of their floodplains, fit the normal relationship.

3.3 Influence of fishermen on catch

The floodplains included in Welcomme's (1976) analysis were selected because of their active fisheries, but many other floodplains are less intensively exploited. The effects on the catch of the number of fishermen operating in a body of water has already been investigated for some African lake systems, and sufficient information is available for a similar analysis for rivers, bearing in mind that it will inevitably be revised as more data becomes available. Table 1, shows the number of fishermen, catch and total flooded area for a number of tropical rivers. Two indices derived from these data, number of fishermen km-2 and catch fisherman-1, are plotted in Fig.1. This shows that the catch per fisherman (C) declines with increasing number of fishermen (N) according to the relationship:

C = 3.922 (0.91 number of fishermen km-2) [r = -0.89]

which tends to a mean catch per fishermen of 3.955 t yr-2 at very low fisherman densities. Obviously, deviations from this will occur in different systems and catch rates as high as 10 t yr-2 have been recorded from the Kafue. This relationship can be used to generate a curve for total catch per unit area of floodplain as a function of the density of fishermen (Fig.2). According to this catches on a floodplain should increase until there are about 10 fishermen km-2, after which the total yield declines. From this rather preliminary analysis, it would seem that floodplains may reach their theoretical maximum level of yield with considerably more fishermen per unit area than do tropical lakes, which appear to read their optimum at between 1 and 2 fishermen km-2. Two reasons may contribute to this. Firstly, there are high proportions of part-time and occasional fishermen on the plains whose numbers are likely to increase as the plain becomes more densely occupied. Number of fishermen cannot, therefore, be taken as directly equivalent to effort. Secondly, in heavily exploited fisheries there is a tendency for forms of fish husbandry to be developed which probably allow the productivity of the unaltered environment to be exceeded to a certain degree. It is, however, by no means certain that floodplains can sustain their yield under conditions of such intense exploitation, and a few have been fished at these levels for any length of time, the ultimate faite of their fisheries is difficult to predict.

The relationships derived in Fig. 1 and 2 depend much on the apparent similarity in performance of artisanal fisheries from a number of river systems. Should more efficient methods of capturing fish be successfully introduced, than the catching power of any one fishermen will be increased with a net result that the number of fishermen that can be supported by any fish stock is reduced.

Fig. 1

Fig. 1 Catch per fisherman as a function of number of fishermen km-2 with calculated regression line C = 3.955 (.910N)

Fig. 2

Fig. 2 Theoretical relationship between number of fishermen and catch from tropical rivers derived from the regression equation plotted in Fig. 1. Also plotted are actual values from Table 1

Table 1

Number of fishermen, catch and maximum flooded area of some tropical rivers

RiverNumber of fishermen
Catch in tons
Area in

kg hm-2
Shire (1972)  2 445  9 039     6653.683.70135.9  Ratcliffe, 1972
        (1976)  3 93113 300     6655.913.38200.0  Willough by and Walker, 1976
Kafue (1963)  1 112  8 554  4 3400.267.6919.7Zambia, 1965
         (1970)     670  6 747  4 3400.1510.06  15.5Zambia, 1971
Senegal10 40036 00012 9700.803.4627.8Reizer, 1974
Central Delta, Niger54 11290 00020 0002.711.6645.0Konare, 1977
Pendjari       65     140       401.652.1535.0FAO, 1971
Ouémé, 195725 000104 000    1 00025.00  0.42104.0  CTFT, 1957
            1968/6929 852  6 500  1 00029.85  0.2265.0FAO, 1971
Niger, Niger  1 314  4 700     9071.453.5851.8FAO, 1971a
Niger, Nigeria  4 60014 350  4 8000.963.1229.9FAO, 1969a; 1970
Benue, Nigeria  5 140  9 570  3 1001.661.8630.9FAO, 1969a; 1970
Barotse     912  3 500  5 1200.183.84  6.8Zambia, 1974
Magdalena30 00065 00020 0001.502.1732.5Bazigos, et al., 1976

When data from Table 1 is combined with data on catch from other systems, from which there is unfortunately no information on fisherman number (Table 2), a plot of catch relative to total flooded area is obtained. The points fconform to a linear relationship, C = 3.83A (r = 0.93), which is equivalent to a constant yield of 38.3 kg hm-2. Deviations from this line can be explained almost entirely by differences in the degree of exploitation, and 92 percent of the variation is resolved when the number of fishermen is plotted against deviations from the expected values. The mean yield of 38.3 kg hm-2 is attained at a theoretical level of 1.33 fishermen km-2, and it may be assumed that floodplains supporting above this density of fishermen are approaching full exploitation. There are indications that above a certain density fishing pressure becomes excessive, but, because of the flexibility of the behaviour of part-time fishermen, this point is difficult to fix. From the data available, it may be concluded that normally exploited floodplains can be expected to yield between 40 and 60 kg hm-2y-1. Exceptions to this will obviously arise, for instance in especially poor areas where low fertility will limit productivity, such as the black water rivers, or systems where the amount of water remaining in the system in the dry season is unusually high, permitting a longer growing and fishing season. In the Shire river, for instance, the permanently flooded area is some 48 percent of the total, accounting for the very high mean values for catch from this system. Much of the catch comes from larger permanent lagoons such as the Bangula lagoon studied by Shepherd et al., (1976). Here yields fluctuated around a mean of 333 kg hm-2 and reached values as high as 539 kg hm-2. Year-to-year variations were strongly correlated with differences in the depth and extent of flooding in previous years.

Fig. 3

Fig. 3 Catch per maximum flooded area of various floodplain rivers. Calculated regression line C = 3.83A

Table 2

Catch and maximum flooded area of some tropical rivers

Catch unit area
kg hm-2
Mahaweli (Sri Lanka)    121       41334.1Indrasena, 1970
Ganges-Brahmaputra (Bangladesh)93 000727 00078.2FAO Yb.Fish.Stats., 1976
Lubuk Lampan       12         2924.2Arifin and Arifin, 1976
Lower Mekong54 000220 00040.7Mekong Study (pers.comm.)
Yaérés  7 000  17 50025.0Stauch (pers.comm.)
Massilli     150       47531.7Barry (pers.comm.)
Niger, Dahomey     274    1 20043.7FAO/UN, 1971
Okavango16 000       8000.5Cross (pers.comm.)
Kamelondo  6 639    7 35511.1Poll and Renson, 1948
Danube26 450  49 40018.7Liepolt, 1967


The yield analysis presented above is applicable to estimates of catch averaged over several years, which may be termed the potential average yield. Actual catches would be expected to vary around the mean value, and when the catches from typical floodplains are followed over a number of years, it is apparent that these fluctuate in a manner that is in some way dependent on changes in the flood cycle. Data from the Kafue, analyzed by several authors including the University of Michigan et al., (1971), University of Idaho (1971) and Muncy (1973), have shown relationships to exist between some measure of flood intensity and catch. Similar relationships emerged when sets of data from the Shire and Niger rivers where examined (Welcomme, 1975). Interestingly, the best correlations were reached when catch was compared with the floods of one or two years previously. This effect was also found by Krykhtin (1975) in the Amur river. The lag time between the year of flooding and the time when its effects are reflected in the catch is probably dependent on the time taken for fish to enter the size range captured by the fishery. In Africa, this is extremely short, less than a year in some cases, because of the rapid growth and small size of some fish, and also because of the heavy fishing for fish of the year as they leave the plain. In the Amur, where fish are slower growing, and are taken at larger sizes, the best correlations of catch were with the flood regime 3–4 years previously.


The simple models described above are intended as an initial guide to some of the factors influencing the yield from natural floodplain systems. As has been emphasized there is a need for much more precise data on catch, fishermen numbers, flood regimes and floodplain morphology, upon which improved estimates can be made. For the time being, however, the formulae derived could be used with caution to provide first-cut estimates. In systems where there are large floodplains, estimates based on a yield per area are probably sufficient to derive the expected catch of a particular section of river. In doing this it should be remembered that yields estimated from natural systems are apt to be misleading when applied to rivers which have been modified extensively for uses other than fisheries. Experience has shown that the alteration of the flood regime for power generation, agriculture and flood control, or the changing of morphology of the river by canalization and the construction of irrigation and drainage ditches, generally tends to lower both species diversity and catch.


Arifin, O. and Z. Arifin, 1976 Fisheries in the floodplain area of south Sumatra, a case study of the Lubuk Lampam in 1973. Paper presented to the IPFC Symposium on the Development and Utilization of Inland Fishery Resources (ABSTRACT), 27 October-5 November 1976. IPFC/76/SYM/10

Bazigos, G.P. et al., 1977 The present state of the fishery of the Magdalena river basin, Colombia. Rome, FAO, FAO Working Paper No.2. FI:DP/COL/72/552:30 p.

CTFT, 1957 Notes et documents sur la pêche et la pisciculture. (Dahomey, Vallée inférieure de l'Ouémé). Serie D.G. No.2. Nogent-sur-Marne, Centre Technique Forestier Tropical, Ser.D.G.(2)

FAO, 1976 Yearbook of fishery statistics. Catches and landings, 1975. FAO Yearb.Fish.Stat., (40):pag.var.

FAO/UN, 1969 Report to the Government of Zambia on fishery development in the Central Barotse floodplain. Second phase. Based on the work of D. Duerre. Rep.FAO/UNDP(TA), (2638):80 p.

FAO/UN, 1970 Report to the Government of Nigeria on fishery investigations on the Niger and Benue rivers in the Northern region and development of a programme of riverine fishery management and training. Based on the work of M.P. Motwani. Rep.FAO/UNDP(TA), (2771):196 p.

FAO/UN, 1970a Report to the Government of Zambia on fishery development in the Central Barotse floodplain. Based on the work of G.F. Weiss. Rep.FAO/UNDP(TA), (2816):19 p.

FAO/UN, 1971 Rapport au Gouvernement du Dahomey sur l'évolution de la pêche interieure, son état actuel et ses possibilités, établi sur la base des travaux de R.L. Welcomme, Spécialiste de la pêche. Rapp.FAO/PNUD(AT), (2938):97 p.

FAO/UN, 1971a Rapport au Gouvernement du Niger sur le développement et la rationalisation de la pêche sur le fleuve Niger, établi sur la base des travaux de N. Bacalbasa-Dobrovici, technologiste des pêches. Rapp.FAO/PNUD(AT), (2913):33 p.

Indrasena, H.H., 1970 Limnological and freshwater fisheries development work in Ceylon. In Proceedings of the IBP Section PF (Freshwater productivity) meeting of Inland Water Biologists in Southeast Asia, at Kuala Lumpur and Malacca (Malaysia), 5–11 May 1969. Sponsored by Unesco. Djakarta, Unesco Field Science Office for Southeast Asia, pp. 45–7

Konare, A., 1977 Collecte, traitement et commercialisation du poisson en plaines inondables. In CIFA Working Party on River and Floodplain Fisheries. Consultations by members of the Working Party, pp. 32–45 (mimeo)

Krykhtin, K.L., 1975 Causes of periodic fluctuations in the abundance of the non-anadromous fishes of the Amur river. J.Icththyol., 15(5):826–9

Liepolt, R. (ed.), 1967 Limnologie der Donau. Stuttgart, E. Schweizerbart'sch Verlagsbuchhondlung, pag.var.

Muncy, R.J., 1973 A survey of the major fisheries of the Republic of Zambia. Rome, FAO, FI:DP 9/10 ZAM 511/3:69 p.

Poll, M. and H. Renson, 1948 Les poissons, leur milieu et leur pêche au bief supérieur du Lualaba. Bull.Agric.Congo Belg., 39(2):427–46

Ratcliffe, C., 1972 The fishery of the lower Shire River area, Malawi, 1972. Zomba Fish.Bull., (3):79 p.

Reizer, C., 1974 Définition d'une politique d'aménagement des ressources halieutiques d'un ecosystème aquatique complex par l'étude de son environnement abiotique, biotique et anthropique. Le fleuve Sénégal Moyen et Inferieur. Docteur en Sciences de l'Environnement. Dissertation Arlon. Fondation Universitaire Luxembourgeoise, 4 vols:525 p.

Shepherd, C.J. (ed.), 1976 Investigation into fish productivity in a shallow freshwater lagoon in Malawi 1975/76. London, Ministry of Overseas Development, 90 p.

University of Idaho, et al., 1971 Ecology of fishes in the Kafue river. Report prepared for FAO/UN acting as executing agency for UNDP. Moscow, Idaho, University of Idaho, FI:SF/ZAM 11: Tech.Rep.2:66 p.

University of Michigan, et al., 1971 The fisheries of the Kafue river flats, Zambia, in relation to the Kafue Gorge Dam. Report prepared for FAO/UN acting as executing agency for UNDP. Ann Arbor, Michigan, University of Michigan, FI:SF/ZAM 11:Tech.Rep.1:161 p.

Welcomme, R.L., 1975 The fisheries ecology of African floodplains. CIFA Tech.Pap., (3):51 p.

Welcomme, R.L., (ed.), 1975a Symposium on the methodology for the survey, monitoring and appraisal of fishery resources in lakes and large rivers. Aviemore, Scotland, 2–4 May 1974. Panel reviews and relevant papers. EIFAC Tech.Pap., (23) Suppl.1, 2 vols:747 p.

Welcomme, R.L., 1976 Some general and theoretical considerations on the fish yield of African rivers. J.Fish.Biol., 8:351–64

Willough by, N.G. and R.S. Walker, 1977 The traditional fishery of the lower Shire Valley, Malawi, Southern Africa. In CIFA Working Party on River and Floodplain Fisheries. Contributions by members of the Working Party, pp. 20–31 (mimeo)

Zambia, 1965 Fisheries Research Bulletin 1963–64. Lusaka, Game and Fisheries Department. Fish.Res.Bull.Zambia, (1963–64):211 p.

Zambia, 1971 Central Statistical Office, Fisheries statistics (natural waters) 1970. Lusaka, Central Statistical Office. 140 p.

Zambia, 1974 Department of Fisheries. Annual report 1974. 39 p.


The discussions following the presentation of the papers highlighted certain points.

Simple Production Models and Need for Statistics

Simple models of production from rivers have been derived by averaging data over a number of systems. Such models are useful mainly for first estimations of the value of a new fishery, or to roughly estimate the impact of management measures undertaken in fields other than fisheries. The models are at present very rough and need improved statistical data for increasing precision. The need for such data, which is also necessary for the management of river fisheries and for the assessment of environmental impacts, has already been noted in the case of lakes in Session I, and the participants felt that any centres set-up for the study of fish stocks, should not be simply directed at lakes but should also consider the problems of rivers and their floodplains.

Environmental Effects of Dams

Much of the discussion during this session centred around the environmental effects of dams. It was firstly noted that rich fisheries have emerged below dams, such as Kainji. Such “tailrace” fisheries have been noted below most dams throughout the world, and are based on a mixture of migratory species which have been frustrated in their movements up-stream and of predatory fish. The improved production of such fisheries, however, rarely penetrates far downstream, and the more general effect is for a diminution in catch to be experienced often far downriver. Intervention both by participants from Nigeria and Ghana, served to illustrate just how distant and complex such effects could be.

In Nigeria effects possibly attributable to the Kainji dam have been detected as far downstream as the Anambra basin. Downstream of the Akosombo dam in Ghana, fish yields have generally declined, although Tilapia has profitted by the increase in submerged vegetation. Freshwater mussels Aetheria eliptica, on the other hand, have increased in abundance. New fisheries in the reservoir upstream of dams can more than compensate for the loss of the catch previously originating from the river downstream, and in planning the fisheries of impounded rivers a balance must be considered between upstream gains and downstream losses. It was also felt that much remains to be done in maintaining fisheries in the downstream areas, either by the seasonal release of a controlled flood which would simulate the previous natural regime, or by the development of aquaculture perhaps associated with irrigation systems. The further possibility was mentioned for the installation of fish ladders to enable fish to pass the dam. Experiences have not been encouraging with these in the past in Africa, although in some tropical waters some success has been achieved in maintaining migrations by this means. It was pointed out that such ladders should only be considered where it can be shown that the life cycle of a major commercial species is endangered by the interruption of its migrations.

Due to the concern expressed on the environmental impact of dams it was recommended by the Symposium that the competent authorities dealing with fisheries should seek most actively to introduce fisheries considerations into the planning of developments liable to have impacts on the fish stock. The aid of FAO and other organizations is requested where necessary to carry out assessments.

Other topics touched upon in discussions were:

Fish diseases

The impact of fish diseases was noted by several participants with regard to wild stocks within their own areas. Large numbers of parasitic organisms have been detected, and heavy infestations were noted in the case of Tilapia in lakes Naivasha and Barringo, and Labeo niloticus in the Nile system. It was, nevertheless, felt that there is little to be done in the natural environment where the parasites form part of the ecosystem. Parasitic infestations do, however, become critical where fish are to be removed from the natural system for stocking or culture elsewhere.

Ageing of Fish

The value of hard structures for ageing fish was discussed. It was concluded that, although the use of annuli for ageing has been questioned in lakes, various structures, including scales, otoliths, vertebral or spines can be so used in rivers. Several participants had already successfully done such work, on a variety of genera including Clarias, Bagrus, Chrysichthys and Tilapia, but there remains the need to excercise caution in the interpretation of such structures especially in the young fish where marks are not always clear.

Previous Page Top of Page Next Page