2. MATERIALS AND METHODS

Fish populations were sampled using multi-mesh monofilament nylon gillnets consisting of 14, randomly distributed panels of various mesh sizes. The mesh sizes used were 6.25, 8, 10, 12.5, 16.5, 22, 25, 30, 33, 38, 43, 50, 60, and 75 mm. bar height, and twine diameter between 0.12 and 0.25 mm. Due to problems with net saturation the panels with the mesh size 6.25 and 8 mm. were removed from the nets used in Botswana (Nermark and Mmopelwa, 1994). Each panel was 3 m. long and 1.5 m. high. (For a detailed description of the nets and the use see Fjalling and Furst (1991).) The standard method used a predetermined number of nets according to the size of the lake and its maximum depth. The nets were set on the bottom and stratified per depth zone. The standard method also used pelagic nets with the same 3 metre panels but with a depth of 6 m.

This standardized method was first adopted by ALCOM, but later slightly modified. The pelagic nets were not always used, because most reservoirs were too shallow, and because the pelagic nets were not always available. The stratification per depth zone was not always followed. The oxygen contents of the deeper zones were often too low for most fish species. In most reservoirs oxygen concentration drops considerably after 3 metres of depth. To enable comparison of data between reservoirs, and to simplify the comparison, only the catches in the 0–3 m zones with the bottom set nets are discussed here.

The method used the catch per unit of effort (CPUE) as an index of abundance. The CPUE was considered to be directly proportional to the actual abundance of a species and a constant called “catchability” (Hamley, 1975). In formula:

CPUE = Catchability × Biomass

The CPUE was defined as the amount of fish, in number or in weight, caught in one net set for a period of one night. In the analysis the weight of the catch was used instead of the number of fish, which Degerman et al. (1988) used. The reasons were that the individual maximum size of the different fish species varied much more in the reservoirs in southern Africa than in the lakes in Sweden, and small reservoirs often had a population of small stunted fish, therefore the numbers of fish would not be a good indicator of biomass.

The fish caught in each net were sorted by species and their total weight was recorded. The CPUE per net and per species was calculated after In(catch+1) transformation. The average transformed CPUE was back calculated to a linear scale.

Transformation into logarithms is commonly used when the mean is positively correlated with the variance. The logarithmic transformation is likely to make the variance independent of the mean. Frequency distributions skewed to the right are often made more symmetrical by transformation to logarithmic scale. A consequence of logarithmic transformation is that the standard error is of no value to the retransformed data. Instead confidence limits in the logarithmic scale are retransformed to a linear scale and indicate the reliability of the data. When retransformed the confidence limits become asymmetrical around the mean.

The number of nets required to attain a certain precision was determined according to Degerman et al (1988):

A C.V.M. of 0.10 meant that differences between two sampling occasions of 50% in CPUE would be statistically significant at the 95% level, whereas a C.V.M. of 0.05 would allow changes of the CPUE of 20% to be detected.

In one reservoir a tagging and recapture experiment was carried out to determine the absolute selectivity of the used nets for the dominant fish species.

In total results from 34 reservoirs in 5 countries were analyzed. A number of these reservoirs had been fished on several occasions to determine the seasonal fluctuations. The total number of sampling occasions was 87.

During the sampling of the reservoirs data that could have influenced the sampling like the weather, water quality, fishing pressure and physical characteristics of the reservoirs were not collected in a systematic way and could therefore not be used in the analysis. Fishing gears, to complement the catches of the sampling nets, were not used.

Table 1: Fish species caught in the different reservoirs in countries of southern Africa.