Chambo samples were collected between August and December, 1990 and the corresponding period in 1991 from Lake Malawi. The fish were obtained fresh directly from the fish landings of commercial fishery.
Species identification, length and weight measurements, study of sex and gonadal maturity, and extraction of opercula were carried out in the laboratory while the samples were in fresh condition. On occasion, samples were kept frozen for a few days before examination. Total fish length and body weight of each specimen were measured to the nearest cm and gm respectively. Gonadal stages were classified on a scale given by Turner (the pers. comm.).
Opercular bones were removed from both sides of fish by cutting around the anterior-posterior edge using a scalpel and then twisting with fingers. A pair of bones from each fish were then stored in a labelled paper envelope.
A total of 533 opercular bones were taken for examination, 137, 190 and 206 for O. lidole, O. squamipinnis and O. karongae respectively. Each opercular bone was soaked in hot water (but not boiling) for a few minutes, freed from skin and tissue and left to air-dry completely. The bones were then examined under reflected light on a black background with a dissecting microscope at × 20. Glycerol solution was used as a refracting medium. The number of marks on each bone was noted. Two independent observations were made by a single reader. To reduce subjectivity in age estimations the second reading was done after an interval of 4 weeks. In case two readings of the same bone resulted in different age estimates, the bones were considered to be unreadable. Relative numbers, mean lengths and standard deviations, mean yearly growth increments were then calculated for each species in each age group.
A sub-sample of 302 fish were taken at random for back-calculation purposes. The total radius and radii to the outer edges of each ring were measured to the nearest 0.1mm using a dial callipers. The plots of total fish length versus opercular radius indicated a linear relationship. Since the regression line drawn from these data did not pass through the origin, the regression line according to Bagenal and Tesch (1978) is described by the following formula (a modification of the direct proportionality formula of Lea (1910)):
Ln-c = Sn/S(L-c)........................ 1
where Ln= length of the fish when ring n was formed, c= intercept on length axis, Sn= radius of ring ‘n’ (at length Ln), S= total opercular radius, L= length of the fish when caught. Therefore, the above equation was used to back-calculate the fish lengths corresponding to the radii.
Numbers, weighted mean length of all fish in each age group at the time of annulus formation and mean yearly growth increment were then calculated for each species.
Calculation of a theoretical growth curve is useful for modelling growth in natural fish populations. In this study the most popular theoretical growth curve, the von Bertalanffy equation Lt = Linf(1-exp(-k(t1-t0))) was used to fit the calculated average lengths at age data and back-calculated length-at-age data (Ricker, 1975; Everhart et al., 1975). The growth parameters of the von Bertalanffy's growth equation such as theoretical mean asymptotic size (Linf) and growth coefficient (K) for both sets of data were estimated graphically using Gulland and Holt, Ford-Walford and Munro plots. The theoretical time of the beginning of growth (to) was estimated using the von Bertalanffy plot.
The growth parameters (Linf, K and to) have also been estimated (Seisay, 1992) by using length frequency data available over a period of 12 months. In this case the Length-Based Fish Stock Assessment (LFSA) micro-computer package (Sparre, 1987) was used. Each length frequency sample was resolved into normally distributed cohort components using the Bhattacharya method. This method plots the difference between successive logarithms against the lower limit of the class intervals. This results in series of straight lines with descending slopes, the intercepts of which provide an estimate of the mean length of each cohort.
The Gulland and Holt plot and the von Bertalanffy non-linear least square regression method was used to estimate the growth parameters. The Gulland and Holt plot was applied after the mean lengths of each cohort component were linked to modal progression analysis in a way thought to represent the growth pattern. The resulting growth increments in length, based on a chosen time interval, were used to estimate growth paramater Linf, K and to.
When the age of each individual fish from opercular bones was read, Age-Length keys were constructed for each species. For each length class fish, the percentage or fractional age frequency distribution was obtained.
