Estimation of growth parameters L∞ and K using length-based methods generally gave higher values of L∞ with corresponding low values of Ks but gave similar values of Æ estimated using tag-recapture analysis (Table 7). This was the case despite the fact that ELEFAN uses the seasonalised growth curve of the VBGF and some cases the oscillations were high (high “C” and “WP” in Table 4). The tag-recapture method using the Gulland and Holt plot does not incorporate seasonality in the analysis. Results emphasize the importance of using the Æ values for comparison of growth performance of fish stocks
Differences in Æ values between the two methods only occurred between male fish of Mbvoniha (Table 7). It could be suggested that the estimation of K for Mbvoniha using tagrecapture analysis of 1.57 cm was exaggerated. This was the highest K of all the values estimated for the different sexes and small water bodies.
The estimation of biomass from the length-based analysis (VPA) gave more biomass than any other method. The estimated biomass was questionable. In the relative terms the results showed that biomass from female was higher than biomass of the male fish in both Mikolongwe and Mbvoniha. The overestimation of biomass could be as a results of the following reasons:
catches from gill nets were used for length frequency analysis and because of adjusting the catches for selectivity to minimise bias, the numbers were over estimated (especially fish of small length classes, compare Tables 1–4 to Tables 5–8 in appendix I). While this adjustment was suitable for estimation of growth parameters (because it reflects relative abundance of each length class), it was not good for estimation of biomass using the VPA.
One of the problems of using the VPA is its assumption for steady-state biomass i.e. it assumes constant recruitment. Growth parameters estimated using ELEFAN had low Rn values which suggested that there were recruitment pulses. Recruitment was therefore not constant.
Biomass estimated from multiple catch tag-recapture study gave lower biomass than the length-based methods. Apart from the reason that the length-based methods overestimated biomass, the biomass from tag-recapture study probably underestimated the biomass:
While the length frequency analysis included fish from low sizes (7 cm and above) which were caught using gill nets, biomass from the tag-recapture estimates were based on size of fish which were tagged, and this included fish of only 10 cm and above. This excluded biomass of all fish lower than 10 cm. This may have lowered the estimated biomass.
Survival rates were ratios of tagged fish at two different sampling dates. If say, fish were tagged on 10/2/93 and the next tagging was on 10/5/93, survival rate was the ratio of the number of tagged fish (tagged on 10/2/93) just before the tagging on 10/5/93 to the number of tagged fish soon after tagging on 10/2/93. It is therefore impossible to have survival rates of greater than one as was the case for Mbvoniha (in September) and Mikolongwe (in June) (Table 10). To estimate survival rates a few parameters are involved: survival rate (Si)
Si = Bi + 1/Bi-mi +Mi (formula given in 3.2.2). and
| since: | Bi = (Miki/Ri) + mi (formula, and explanation of the notations in 3.2.2) |
It is easy to note that incomplete reporting of tags affects ki, Ri and mi and therefore affects the Bi and then eventual Bi + 1(Jones 1977). It is possible that incomplete reporting of tags occurred in this study. For Mikolongwe, 15 tags were reported from fishing activities by local community in March 1994. For Mbvoniha, 8 and 2 tags were reported from hook and line fishing in June and July respectively; 5 and 2 tags were reported from traps in June and July 1993 respectively. From the same traps, 2 tags were reported in February and 1 was reported in March 1994. All these were from catches by the local community. These data emphasize the fact the fishing by local people occurred during the study period and the possibility of having many tags unreported can not be overruled. Such high unrealistic survival rates could be explained by such a possibility. Further, the experiment took a long period of time (about 10 months) that there were possibilities of tagged fish escaping out during flushing of the small water bodies in times of rainy seasons. Tags were recorded down the stream of Mbvoniha (personal observation). The possibility of emi and immigration of tagged fish and therefore affecting tag reporting was there.
From Mbvoniha, Mattson (1994) using a single recapture study in a period of 3 to 4 weeks and analysing data using the Peterson method gave higher estimates of biomass than what was found in this study. Proper conditions were set:
the experimental period was short to discount any possibility of migration of tagged fish;
fishing by other people was stopped over the experimental period;
fish of less than 10 cm were tagged (although from experience, fish of lower sizes suffer from mortality due to tagging (Mattson, personal communication).
Production/yield estimated from empirical methods were lower than those estimated using length-based and tag-recapture analysis. Inconsistent yield estimates were obtained using the models 1 and 2. One would have expected the models to give similar differences between Mbvoniha and Mikolongwe since they were both based on Total Phosphorus. Similar differences were found using model 4 and 5. Model 3 using transparency gave highest production estimation.
