Previous Page Table of Contents Next Page



Robin L. Welcomme
Fishery Resources and Environment Division,
FAO, Via delle Terme di Caracalla, 00100 Rome

This paper examines historical data on fish catch and hydrological records from the Central Delta region of the Niger with a view to establishing the relationship between catch levels and fluctuations in climate and flood regimes. The data are drawn from hydrological records at Koulikoro near Bamako and catch records at Mopti. The relationship was found to be stronger at low levels of flood intensity than at high levels, with a very definite cause and effect relationship between falling floods and falling fish catch. The evidence also suggests overfishing is present and it is recommended that, in view of the recurring problems of drought in the Sahel region, serious consideration should be given by the authorities to the better management of the fisheries. Existing controls may have exacerbated the situation and improved management might result from re-establishment of traditional practices in consultation with the local fishing communities.


The fishery of the Central Delta of the Niger River in Mali is one of the most important inland fisheries of Africa. The Central Delta is an extensive floodplain which covers up to 30 000 km2 during the peak flood and which retains some 4 000 km2 of water in a series of lakes during the dry season. It is inhabited by a complex community of fishes comprising over one hundred species of which over seventy regularly enter the catch. The fishery is pursued with a wide range of fishing gears including traps, lines, gill nets and cast nets and has been estimated to have a potential of some 110 000 tons under the flood regimes that were normal in the 1950s and 1960s.

In recent years radical changes in climate have altered the rainfall patterns over the whole of Africa and have produced drought conditions in the Sahel through which the Niger River flows. The changes in precipitation have in turn affected the flood regimes of the rivers in the region, principally the Niger, the Senegal and the Chari-Logone systems. The effects of year-to-year differences in flood intensity of fish populations in rivers have been noted by several authors. Early workers such as Antipa (1910) and Wimpenny (1934) concluded that fisheries production in unmodified rivers was directly proportional to the strength and duration of the floods. Later Ivanov (quoted by Chitravadivelu, 1974) and Holcik and Bastl (1977) found correlations in the Danube River between fish catch in any year and flood strength in the preceding year. Krykhtin (1975) also found similar relationships between the catch in any one year and the flood regimes three to four years previously in the Amur river. In Africa the catch and flood history of the Shire, Kafue and Niger floodplains were analysed by Welcome (1979), who found good correlation between catch and flooding one and two years previously.

Two components of the flood regime have been found to influence catch. The magnitude of the high water phase affects the size of the stock through the improved reproduction, survival and growth of fish during years of more intense flooding. The degree of drawdown affects natural survival and ease of capture during the low water phase, and several workers (e.g. Vidy, 1983; Annibal, 1983) have found improved catches at times of exceptionally low water. Welcomme and Hagborg (1977) investigated the mechanisms regulating fish abundance in flood systems by computer simulation and concluded that differences in ichthyomass arise principally from differences in the magnitude andduration of the flood but that the greater the amount of water remaining in the dry season the more such differences are transmitted to following years.

The present paper investigates the effects of the changes in flood regime associated with this climatic change on the fish catch of the Central Delta as exemplified by the landings at the main port of Mopti.


Mopti has been a centre for the study and development of the fisheries of the Central delta since the early 1960s. As a result a continuous set of data is available, which has been summarized in a number of publications, principally the various reports of the ‘Operation Peche, Mopti’ and of the Laboratoire d'Hydrobiologie de Mopti. Data on river discharges were taken from the hydrological records of ORSTOM of which the records for Koulikoro just downstream of Bamako are the most complete. The basic data used in this analysis are given in Table 1.


Catch, mean, flood and dry season discharge rates for the Central Delta of the Niger.

YearCatchDischarge rate (m3/sec)
Koulikoro gauge
 (t × 1000)MeanDry seasonFlood
1963 157212892928
1964 161311233039
1965 150514322772
19661101443  8962736
1967  98195110013735
1971  941309  6092516
1972  88113013222040
1973  73  939  6121777
1974  631427  4102785
1975  871540  7202960
1976  891461  8892774
1977  87  89912311592
1978  771304  8822462
1979  83153715782810
1980  88  885  7031652
1981  751118  7392112
1982  73  90210341632
1983  61  825  8171514
1984  54   

The regime falls into two parts, the flood (July – December) and the dry season (January – June). These were prepared for comparison with catch by establishing a Hydrological Index (HI) consisting of the mean of the monthly discharges for each six-month period. The hydrological index for the flood was highly correlated with mean discharge (R2 = 0.99) but the correlation between flood and dry season indices was much lower (R2 = 0.43). Separate analyses therefore had to be made for the two components of the flood regime. In comparing catch with HI the form of the flood curve, with peak discharges in the second half of the year, means that the catch in any year ‘y’ derived from the flood in the preceding year ‘y-1’ consisted mainly of 0+ fish between 6 and 10 months old. Similarly fish derived from y-2 were 18 to 24 months old.

Figure 1

Figure 1: Typical flood regimes for two consecutive years

Figure 2

Figure 2: Plots of catch in year ‘y’ and flood HI for year ‘y-1’ for the Central Delta of the Niger at Mopti.

The catch for years 1966 to 1984 is plotted in Fig. 2 together with the flood HI for the preceding year.

The similarity between the two lines is apparent on inspection and persists through two cycles of recovery and drought. Simple regression analysis shows that catch is only poorly correlated with dry season HI but that significant correlations exist with flood HI. When catch in year ‘y’ is plotted against flood HI in year ‘y-1’ R2 = 0.67, however it is apparent that previous years may also influence catch so a general formula:

εy = A + B(p1HIy-1 + p2HIy-2)

was applied. In the case of the full 1966 to 1984 data set the best fit obtained was:

εy = 198.17 + 0.27 (0.7 HIy-1 + 0.3 HIy-2)

when R2 = 0.84. Inspection of the plot by eye against the fitted line indicated that the linear regression was perhaps not the most appropriate and further improvement in correlation to R2 = 0.87 was obtained with the regression (Figure 3):

εy = 151.73 log (0.7 HIy-1 + 0.3 HIy-2) - 428.26

Figure 3

Figure 3: Landings of fish at Mopti correlated with discharge in years y-1 and 7-2.


The log-normal nature of the regression indicates a slight lessening of the response of the fish community as flood intensities increase over a certain level and it is doubtful that even years of extremely high flooding would produce increases in catch much above 110 000 tons. Conversely as flood intensities reduce, the effect on the fish stock becomes greater. Thus any further failures in the flood in coming years are liable to have an increasingly severe effect.

The high correlation obtained shows hydrological parameters to account for 87 percent of the variation in catch. This implies that changes in fishing pressure, for example from 78 000 fishermen in 1972 to 99 000 in 1982, contribute little to this variation. The reasons for this probably lie in the responses of complex fish communities to fishing pressure whereby increases in effort are compensated for by changes in the structure of the fish community rather than by any fall in total catch. Examination of the catch composition over a 12 year period (1969–80) (Laboratoire d'hydrobiologie de Mopti, 1981), however, did not reveal any great changes in composition although certain trends were evident (Table 2). Catches of Lates niloticus, for instance, fluctuated according to the flood regime 2 to 3 years previously (y-2 and y-3) and thus consisted mainly of fish of the 1+ and 2+ year classes, as would be expected of such a large species. The tilapiine cichlids on the other hand tended to increase in abundance with falling discharge and rose from 19 percent of the catch at the beginning of the period to 44 percent at the end. Most other species fluctuated according to discharge in year y-1 and y-2. Some species were more severely affected either by the adverse flood regimes or by increased fishing. These included obligate floodplain spawners such as Heterotis niloticus, Polypterus spp. and Gymnarchus niloticus as well as Citharinus citharus.

There was no data on the mean size of the fish caught although some reduction in the mean lengths of the fish entering the fishery would be expected of a heavily exploited fish community. The only indication that this has occurred lies in a shift in the regressions of discharge against catch. Welcomme (1979) presented a relationship for the Central Delta fishery for the period 1966–74 based on readings of the Mopti gauge as follows:

Cy = 3 239 + 32.10 (0.5HIy-1 + 0.5HIy-2)

and the present data set based on the Koulikoro gauge confirms this with a relationship for the same period of:

Cy + 10.136 + 0.03 (0.55HIy-1 + 0.45HIy-2).

In other words the catch between 1966 and 1974 was composed of about 45 percent 1+ fish and 55 percent 0+ fish. The influence of subsequent years changed this ratio and a relationship calculated for 1974–84 indicates that catch is best explained by a composition of 76 percent 0+ fish and 24 percent 1+ fish.

Regressions were calculated from limited series of values within the data set in order to test the accuracy of prediction of future known catches. Results demonstrated that, with 14 years data or more, predictions of catch from hydrological values are at an acceptable level (less than ± 10%) for certain types of management decisions. Such equations can, therefore, be used in management provided there are no major changes in the resource base, such as those produced by overfishing or environmental degradation.

The fish landings at Mopti indicate a very definite cause and effect relationship between the falling floods and falling fish catch over the past twenty years. Whether these changes may be regarded solely as part of the normal fluctuation of river fish stocks in response to variations in year-to-year flood strengths or whether overfishing is also present is difficult to answer on the basis of the present evidence. Until 1978, the fish community would not appear to have been unduly overstressed as rises in flood intensity in 1974/75 and again in 1978/79 produced an immediate response in increase in catch in the following years. Furthermore, although certain floodplain spawning generally have diminished in abundance or even disappeared from the catch, the majority of taxa continued to appear in much the same relative abundance despite some year-to-year variations. However, the steady diminution in the percentage of 1+ fish in the catch and the increasing reliance on fishes in their first year of life is indicative of overfishing and does not augur well for the fishery as the risk of removing the young before they have had a chance to spawn is very high.

With the continuing drought in the Sahel leading to the drying out of several of the lakes associated with the Niger floodplain and evidence of a dramatic increase in fishing activity in the 1980s the fisheries authorities concerned should give urgent consideration to the better management of the fishery. This may be most appropriately achieved in the Central Delta both by limiting access to the fishery thereby reducing fishing effort and by imposing size restrictions on the fish caught through mesh size limitations. This would initially reduce the amount of fish caught and would also deprive many of employment in a situation where the fishery has been viewed as a famine relief mechanism as other pastoral pursuits have become increasingly limited. However continuance of the existing pattern of resource use will inevitably produce the same result through the collapse of the fishery. Through management there would be a long term benefit in the survival and continued productivity of the fish community albeit at a lower level than previously.

Regrettably the socio-economic context of the Central Delta makes such intervention difficult to put into effect. Moreover there is evidence that in some countries the management mechanisms based on fisheries licences which place no restriction on the holder as to location of fishing may have exacerbated the situation by permitting the crowding out of local fishermen and accelerating the breakdown of traditional controls on the fishery. A possible approach to such fisheries management problems might be to re-examine the traditional management mechanisms as they apply to individual localities in order to determine their relevance to the present crisis as it relates to fishermen. Re-establishment of traditional management practises is only likely to be successfull if it is carried out with the closest possible participation of the local fishing communities.


Percentage composition by genus of landings at Mopti 1964–1980



Annibal, S.R., 1983 Avaliacoa bio-ecologico e pesqueira das “pescadas” (Plagioscion squamosissimus Heckel, 1840 e Plagioscion montei Soares, 1978) no “sistema Lago do Rei”, Ilha do Careiro, A.M, Brasil. INPA, Manaus. M.Sc. Dissertation, pag. var.

Antipa, G., 1910 Reguinea inundabila a Dunarii. Starea ei actuala si mijloacele de a o pue in valoare. Bucharest, 318 p.

Chitravadivelu, K., 1974 Growth, age composition, population density, mortality, production and yield of Alburnus alburnus (Linnaeus, 1758) and Rutilus rutilus (Linnaeus, 1758) in the inundation region of the Danube-Zofin. Acta Univ.Carol.Biol., 1972: 1–76

Holcik, J. and I. Bastl, 1977 Predicting fish yield in the Czechoslovakian section of the Danube River based on the hydrological regime. Int.Rev.Gesamt.Hydrobiol., 62(4):523–32

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

Laboratoire d'Hydrobiologie de Mopti, 1981 Recherches ichthyologiques dans le delta central du fleuve Niger: Projet NOVIB. Rapport final. Mopti, Laboratoire d'Hydrobiologie, 113 p.

Vidy, G., 1983 Pêche traditionelle en bordure du Grand Yaérés nordcamerounias: Le Logomatia. Rev.Hydrobiol.Trop., 16(4):353–72

Welcomme, R.L., 1979 The fisheries ecology of floodplain rivers. London, Longman, 317 p.

Welcomme, R.L. and D. Hagborg, 1977 Towards a model of a floodplain fish population and its fishery. Environ.Biol.Fish., 2(1):7–22

Wimpenny, R.S., 1934 The fisheries of Egypt. Sci.Prog.Twentieth Century, Lond., 29(114):210–27

Previous Page Top of Page Next Page