A total of 11 848 B. björkna, 3 253 A. brama, 858 A. cernua and 573 others were caught (Table 6).
Fish species composition of the catches from the five types of nets (Figure 3) differed significantly. For example the YFT contained a lower percentage of B. björnka and a higher portion of A. cernua than each of the other four nets.
Depending on the type of net, A. cernua peaked a 5 – 9 cm, B. björkna usually at 17 cm, A. brama near 21 cm, and total fish at 5 – 8 cm and near 17 cm (Table 7).
The mean size of the fish caught (Table 8) differed between the types of trawl in the case of A. cernua, but not in A. brama and B. björkna.
The variance (and standard deviation) of length indicates the specific selectivity of the different types of net. Within the three types of trawl, for A. cernua variance was highest in the YFT, for B. björkna it was highest in the CVT and for A. brama it was highest in the PMT (Table 8). According to variance of length of total catch the five types of gear rank:
| Young Fish Trawl | s = 7.5 cm |
| Commercial Vendace Trawl | s = 5.8 cm |
| Gill Net | s = 4.9 cm |
| Seine | s = 4.6 cm |
| Pikeperch Midwater Trawl | s = 3.7 cm |
Species composition from trawl transect no. 1 differed significantly from that of the two other trawl transects. For example the proportion of A. cernua was higher in the hauls from transect no. 1.
The A. brama, A. cernua and B. björkna caught at transect no. 1 were seemingly larger than those caught at transect no. 3 (Table 8), but these differences in size were statistically insignificant. Only the mean sizes of A. brama from transect no. 1 (21.2 ± 5.5 cm) and transect no. 2 (19.8 ± 4.0 cm) differed slightly (5 % level).
The numbers of B. björkna per trawl unit were significantly higher from transect no.3 than from transect no. 1 (n = 123/2.419). The rich catches from transect no. 3 may have caused sampling bias towards smaller fish.
Species composition differed between catches from different depths. For example the 10 m-hauls contained a significantly lower percentage of B. björkna than the more shallow hauls.
The mean sizes of fish trawled in different depths are given in Table 8. For example B. björkna from 5/10 m differed significantly (5 % level) in size, as did B. björkna from the surface/10 m (1 % level). The tabulated standard deviations may be biased to some extent by duration of trawling.
Trawling at 5 m depth revealed significantly more (n/trawl unit) A. brama (1 % level) and A. cernua (5 % level) than trawling at the surface. Hauls at 5 m contained more B. björkna (5% level) than trawls at 10 m.
Section 3.3.2 describes the horizontal and vertical gradients of fish density derived from echo sounding.
The population of A. brama (13 – 35 cm) was estimated at 182 000 – 194 000, B. björkna (11 – 21 cm) at 362 000 – 386 000 and R. rutilus (16 – 23 cm) at 61 000 – 65 000 fish. The error level for white bream is <10 % (Robson and Regier, 1964), and for roach and bream <20 %.
The total population estimate was 605 000 – 645 000 fish with an error level of 20 %. This is equivalent to a biomass of 317 – 338 kg/ha.
The size distribution based on 4892 single-fish echoes from 7 recordings with the split-beam system ranged from - 53dB to - 34dB with a maximum at - 44.5dB. Target strength distribution from the dual-beam system is compared in Figure 4. Based on the single-target detection criteria used, the system estimated the target-strength distribution based on 4210 single fish echoes. This distribution ranged from - 58dB to - 32dB with a maximum at - 44dB.
The single-beam system detected 4781 single fish on 11 transects. There were insufficient echoes to produce reliable target-strength distributions along the transects. The system also gave large varieties between different transects.
However, it is interesting to note the conformity between all three systems on transect no. 3. The single-beam system recorded 646 single fish echoes, and even if this is low for the indirect method it shows some similarity to the two direct methods (Figure 5). The size distribution of the trawl catches from transect 3 is given in Figure 6.
Table 9 (single-beam system) clearly shows the patchy distribution of fish. All densities represent the mean value along the transect lines. Transect I - J was recorded both on the 13th and the 14th of November. On the 14th a dense patch of fish was hit, and the mean value along the transect increased to 2 969 fish/ha. Because there were so few echosignals on each transect they could not be divided into subsamples.
On the 14th of November an upwardlooking transducer was placed on the bottom to detect fish close to the surface. During the experiment no fish were detected either by day or night in the upper two meters, and only a few fish were located between 2 and 5 meters. From this observation it was assumed that very few fish have been missed by the echosounders in the surface layer. Target-strength distribution by depth is shown in Figure 7. For gradients of fish density derived from netting see the chapters 3.1.2 and 3.1.3.
With the data of the single-beam system from the seven grid transects a mean pelagic population of 314 000 fish (95 % CL.: 649 000; 152 000) was found. The results of the single-beam system from the trawl transects together with the grid transects give an estimate of 393 000 fish (95 % CL.: 682 000; 226 000). In these calculations we have used fish with target strengths > - 56dB.
Table 10 presents population estimates from the dual-beam system. Since the TS distributions for the night-time 420 kHz surveys were not significantly different, the data from these surveys were combined to give a more precise population estimate of 310 000 fish ± 115 000.
Selected estimates of the overall pelagic fish population of Lake Tegel in November 1985 are compared in Table 11.
