On growth parameters, both length-frequency and tag recapture seem to estimate them with some degree of precision. The decision to use either of them will depend on consideration of inputs required to conduct the study (Table 14). Using non selective gears will ease up the work in that selectivity studies to adjust length-frequency catches will not be undertaken.
Regarding mortality rates, the length frequency programs can be used to estimate fishing mortality provided an estimate of natural mortality rate is made. For tag-recapture method, the survival rate estimates can be made provided conditions are met. It is noted however that it is not easy to meet the conditions.
On estimation of standing biomass, using length-based methods gives complete analysis of the fish population, but better results can be obtained when using non selective fishing gears like the seine net. While adjusting the catches for selectivity leads to better estimates of growth parameters, it does not seem to give better results for population estimates. With multiple tag-recapture studies, as was the case of this study, the study takes a long time, and there are many conditions which have to be met to have the correct analysis. For example reporting of all lost tags which seem to have influenced results of this study can be crucial. Further, there is an additional work load to take measurements of weight and length of fish i.e. tagging of fish. Tagging is a cumbersome exercise especially when a good proportion of the population has to be tagged to have better results. It can be noted from this study that the number of fish declined with time for Mikolongwe (Table 3). Perhaps people who were involved in the tagging exercise got tired in the end or number of fish suitable for tagging diminished. Further, as it can be noted in Table 14 that labour inputs for multiple catch tag-recapture studies are much higher than for the length-frequency and other studies.
Using a single catch tag-recapture methods, estimates from Mattson (1994) seem to be better than any of the methods used in this study for estimation of fish biomass. The single tag-recapture studies require a period of about 4 weeks while the multiple tag-recapture and length-based methods (using VPA) require about a year to estimate biomass. Empirical models seem to underestimate potential yields of the stock. Models 4 and 5 using MEI gave consistent differences and could therefore be improved on to give realistic yields. Model 3 which used transparency gave highest estimates and more work needs to done on it to see if it can give reliable estimates.
It is therefore recommended that:
For estimation of growth parameters, mortality rates and management of the reservoirs, length frequency data be used using non-selective gears (considering its low manhour requirement). This data will be used to estimate standing stock as well.
For estimation of standing stocks at a particular time, carefully planned single catch tag-recapture studies fulfilling all the conditions required should be used while item (c) is in progress.
Studies should be carried out to determine important morphoedaphics, physical, chemical or even biological parameters which are strong predictor variables for estimation of fish yields in this southern part of African. This will replace (b)as quick method to estimate standing biomass and production.
