During this experiment 3 different hydroacoustic systems were used: a split-beam system from SIMRAD, a dual-beam system from BioSonics and a conventional single-beam echosounder, the Simrad EY-M.
All these systems use different techniques to assess the pelagic fish population. In the Lake Tegel study both direct and indirect in situ target strength measurements were made: indirect measurements using the single-beam system, direct in situ target strength measurements using the split-beam system and the dual-beam system.
The single-beam system is described by Bayona (1984) and Dahm et al. (1985), the split-beam system in Foote et al. (1984), and the dual-beam system in Ehrenberg (1978).
Hydroacoustic estimates using echo integration require a scaling factor based on the average target strength of the population being assessed. In this study the scaling factor was determined by direct and indirect methods: the indirect method used a single-beam echo sounder and the Craig and Forbes technique, while the direct method used the dual-beam echo sounder. An additional target-strength distribution was provided by the split-beam system. All three systems used in this exercise have methods to sort out echo signals from single fish. The shape and duration of the echo signal usually indicates whether the echo comes from more than one fish. In order to calculate the scaling factor for estimating the total density of fish in the scattering layer, it is assumed that the size distribution of fish registered as singles reflects the size distribution of fish responsible for the multiple echoes.
The limitations of the method are discussed by Foote, (1980), Dahm et al. (1985) and others.
The transformation of target strength (TS) into fish length is discussed by Dahm et al. (1985). The following formula have been used in the Lake Tegel experiment:
| Single-beam system | : TS (dB) = 20 | log L (cm) - 67 |
| Dual-beam system | : TS (dB) = 19.1 | log L (cm) - 64.4 |
The formula used by the single-beam system was found by experiments with the same echo sounder and trawl catches (Lindem and Sandlund, 1984). The formula used by the dual beam system is from Love (1971).
By using an echo sounder “with a 40 log R” time-varied-gain (TVG), received echo signals were independent of range. The echo amplitude then depended only on the angular coordinates of the target. The directivity pattern of the system was obtained by combining the transmit and receive directivity from the transducer. In this experiment the directivity pattern was divided into 2-dB class intervals.
By applying the simple algorithm proposed by Craig and Forbes (1969), it was possible to remove the effect of the beam pattern from the received echo distribution. This statistically transformed echo distribution to a target-strength (TS) distribution. This indirect method is based on solving a set of equations recursively. An error in one equation will have great influence in the next set of equations to be solved. To obtain good results with this method it was essential that each transect contained many single fish echoes, preferably several thousands.
Earlier work demonstrated a precision of echo survey estimate which was better than 10 % (Lindem, 1983). When used together with pelagic trawl and gill net it was possible to relate the target strength data to the actual length distribution of the fish population (Lindem and Sandlund, 1984).
By simultaneous echointegration on both single fish echoes and on multiple echoes the relationship between them is estimated. In this way the total fish density could be obtained.
The split-beam system is a direct in situ method to determine fish target strength by removing the beampattern effect for each individually resolved echo. The transducer is under reception divided into four quadrants. Half beams are formed in each of the two planes to allow measurements of phase differences in the received echosignal. This gives us the direction to the target, and the beampattern can easily be removed.
An advantage of the split-beam is the precise localisation of scatterers in the beam. Individual fish can be tracked through the beam and their reaction to the passing survey vessel can be determined.
In a theoretical study Ehrenberg (1979) found the split-beam method superior in noisy environments and in the presence of interfering targets.
In the Lake Tegel experiment the split-beam echosounder was used to measure fish target strength. An integrator to produce measurements on fish abundance was not used. The system was operated at a frequency of 70 kHz.
Hydroacoustic data were also collected with a BioSonics Model 105 Portable Sounder configured to operate with a dual-beam transducer. A dual-beam transducer transmits sound pulses on a narrow-beam element, and then receives echo signals on both narrow - and wide - beam elements. The system could operate at either 120 kHz or 420 kHz. The dual-beam transducers were: 120 kHz with 10° / 22° beamwidths and 420 kHz with 6° / 15° beamwidths. For the mobile surveys, the transducers were mounted and towed in a hydrodynamic V-fin (one at time). For fixed-location data collection, the transducer was mounted in a floating life ring.
All data were collected at 420 kHz except for the 14 November grid survey and trawl transect 3, which were at 120 kHz.
Hydroacoustic data were recorded digitally on video tapes using the BioSonics VCR Digital Hydroacoustic recording system. This digital recording system was used because its 80 dB dynamic range is at least 100 times greater than those of analog recording systems (reel-to-reel tapes and cassettes). Three channels of data were recorded to permit simultaneous echo integration (narrow beam at 20 log(R) TVG) and dual-beam target strength measurements (narrow and wide beams at 40 log(R) TVG). Data were also displayed on a BioSonics Model 115 Portable Chart Recorder (narrow beam at 40 log(R) TVG).
The transducers, mounted in the V-fin, were towed at about 0.5 m below the surface at an estimated boat speed of 2 m/sec. The transmitted pulse width was 0.4 m/sec, and the pulse rate was 2.5 pulses/sec. Because of the transducer/sounder characteristics and towing depth, data collection began at about 1m below the surface. Data analysis began at 1.5 m below the surface.
After the surveys, the recorded data were played back through two signal processors: one to measure fish-target strength and the other to measure fish densities. The outputs were processed with computer programmes.
Fish populations often utilize different areas of a water body over the year. It is important to make hydroacoustic surveys at a time when most of the fish is found in the deeper part of the lake. Earlier recordings of the fish population in Lake Tegel had shown such a concentration in late autumn. November was found to be the best month for an echo survey.
Schooling behaviour is a problem when it is looked for single fish echoes. The best survey results are obtained during night when the schools spread out to form a layer of fish. Thus densities will be low at night and single fish be detected easily. In this experiment most of the data were recorded at night. However, some species may migrate to the surface layer at night, thereby making them inaccessible to the hydroacoustic system towed from a boat.
Transect lines should be made to give good statistical coverage of the whole lake area (see discussion).
In Lake Tegel hydroacoustic data were collected along six “grid” transects and three “trawl” transects on 12,13 and 14 November 1985 (Figure 1). The trawl transects were so designed that distributions of fish target strengths could be compared with the distributions of fish lengths obtained from the trawl net catches. Generally, the trawl transects were sampled hydroacoustically immediately before catches were made with the trawl.
Because of the selectivity of the codend-mesh size in a single gear three different types of trawl were used during the exercise: the YFT (Young Fish Trawl) and the CVT (Commercial Vendace Trawl) of the Institut für Fangtechnik and the PMT (Pikeperch Midwater Trawl) of the Fischereiamt Berlin. Both the YFT and the CVT have been described elsewhere (Bagenal et al., 1982). The PMT was very similar to the CVT the only difference being that the codend-mesh opening (internal stretched diameter) was 40 mm.
All nets were towed with a 100 m towing-warp. With the slightly negative buoyancy of the total gear the desired depth could be reached by suspending the trawls by lines connected to floats running at the surface. These lines were fastened to the points where the towing warp and the sweepline joined. An additional weight of 15 kg per side helped to reduce the lifting effect of water resistance on the floats. At night the distance between the two towing vessels was controlled by a marked line of 60 m length which the skippers kept tight by steering at an appropriate angle to the general course. Trawling speed was 1.8 kn as an average.
All three trawls were used in the two-boat mode (Steinberg and Dahm, 1974). The deck layout of the two towing vessels “Hecht” (Grosch 1973) and “Oberhavel” (overall length 15 m, beam 3 m, draft 0.6 m, 55-hp Diesel engine) contributed greatly to the success of the experiment which was intensive and demanding of time and space.
There were two main restrictions on trawling. On one hand all three nets had to be maintained at the same level for reasons of comparability. This meant that all tows in a given depth and on a given course had to be repeated three times. On the other hand trawl fishing in the two-boat mode, where only very slowly course changes can be made, was possible only on very few transects due to the existence of aerators in the lake (Figure 1) which had to be avoided. The problem was aggravated by the fact that all trawling had to be carried out at night.
The group agreed in advance that trawling be confined to three selected transects (Figure 1) and that trawling samples be taken with all three trawls at the surface, at 5 m depth and at 10 m depth.
It should be clearly pointed out that the experiment was not intended to estimate the total fish number in the lake from trawl catches. Previous experience has shown that dimensional changes of the net during the tow, the unknown and probably different catchability of the three nets, the patchy distribution of fish and a size- and species-dependent behaviour of the fishes would bring unacceptably high elements of uncertainty into such a calculation. What the trawls could provide were representative samples of the relative species composition and length distributions of the available fishes. It is clear that because of mesh selectivity those samples were not representative below a fish length of approximately 5 cm. An unknown degree of underrepresentation due to net-avoidance reactions can also be assumed with bigger fishes. Nevertheless it is the firm conviction of the group that the trawl catches represented an adequately accurate sample in the length range from 5 to 30 cm (see figure 2).
Trawl haul number (T1 – T27), courses, fishing depths, trawl types and catch results split into numbers per species are recorded in table 3.
One trawl unit of the chapters 3.1.3 and 3.1.4 consisted of nine 10 minute-hauls with three types of nets, by three depths or three locations.
The boat-operated seine had a length of 200 m and a depth of 11 m. The mesh opening is 40 mm in the wings and 30 mm at the knot-less cod end.
To reduce the effects of selectivity on samples taken with gill nets, many workers have used fleets of different mesh sizes or multimesh-gill nets. Another approach is the multimesh-gill net of the “Latin Square” type, designed by Büttiker and colleagues (1985).
According to the theory of Regier and Robson (1966) mesh sizes followed a geometrical progression (10,13,17, 21, 27,34,44,56 mm). The eigth different mesh sizes were arranged in four horizontal squads, four net pieces formed a vertical column (Table 4). Each rectangle was 10 m long and 1.5 m high, so the overall dimensions of the net were 80 × 6 m. The material was white-blueish PA monofilament (0.12 mm).
The net was anchored floating in about 10 – 12 m water depth (Figure 1) 1 m below the surface, thus fishing from about 1 to 7 m. It was set three times (G1 - G3) in the late afternoon and emptied the next morning.
Catches from the main experiment (Nov. 12 – 14) were usually processed the day following the night of sampling although the rich seine catches needed four days for complete processing. The fish caught was stored in the open at temperatures near freezing point until it could be treated.
Fishes were grouped into four main categories: bream (A. brama), white bream (B. björkna), ruffe (Acerina cernua) and others, consisting of 573 fish: 362 roach (Rutilus rutilus), 24 perch (Perca fluviatilis), 140 pikeperch (Stizostedion lucioperca) and some others.
Lengths were expressed as standard length, with a range, for example, of 15.5 – 16.4 cm being recorded as 16.0 cm.
The A. brama from hauls S1, T8, T19, T22, T27, and G3 were subsampled to n = 173, 77, 10, 11, 17, 16; the B. björkna to n = 348, 222, 79, 150, 35, 76; the A. cernua of hauls S1 and G3 were subsampled to n = 127 and 42; and the R. rutilus of haul no S1 were subsampled to n = 152.
Species composition of the catches was compared with a 2 × 2 contingency Chi2 - test with n = 200. Samples were combined on the basis of percentage (instead of absolute number) of caught fish. Consequently, fish length distributions were compared on a percentage basis with t- test, Weirs solution of the Behrens - Fischer problem (s2 differs), and H- test (deviation from normal distribution). Absolute numbers of trawled fish were compared with the Wilcoxon matched pairs signed rank test. In statistical pretesting normality of size distributions was tested to the (recommended) 10 % level, F- tests were conducted to the 5 % level. Final tests were conducted to the 1 % level, if not otherwise stated in the following text.
The mark-recapture experiment was carried out with the three fish species, bream, white bream and roach that dominated the biomass. An original population estimate of one to two million bream, white bream and roach was based on the average commercial catch of about 100 kg/ha/yr in the sampling area which contained 84 % of these three species. Following the suggestions of Robson and Regier (1964) a total of 20 000 fishes had to be marked and 40 000 had to be examined for marks to achieve an error level of 10 % for this population size.
Lake Tegel is connected to the River Havel. The experiment was fixed for late autumn to minimize migration. Migration was investigated by looking for marks in commercial catches: as among 758 bream and 88 white bream caught in the River Havel only one marked white bream was recorded it was concluded that migration was negligible.
Fish for marking were caught with a seine net (of. 2.2.2) on October 16 (1 ton), 23 (3.5 tons) and 31 (2.5 tons). The catches were kept in the lake in two 15 m3 net cages for a maximum of 3 days.
A total of 28 207 fish were marked comprising 3 126 bream, 14 498 white bream and 10 583 roach. The length distribution of the marked fish is given in Figure 2.
A subsample of each species was weighted to establish the relationship of length to weight. Fish were anestesized, measured, and 4 % Alcian-blue solution was injected into the fins with an automatic jet inoculator (Hypospray Jet Injector K 3, RP Scherer Corp.). The lower part of the caudal fin was marked in the first catch, the anal fin in the second, and the dorsal fin in the third. The fish recovered in aerated water. Fish showing abnormal behaviour, loss of scales or lesions of the mucus layer were rejected from marking. After recovering from narcosis the fish were uniformly distributed throughout the lake.
To monitor mortality, 302 fishes were kept in three fibre-glass tanks (2 5000 l) from October 16 to November 25: 150 fish of which had been marked, the remaining untreated. The fish were equally distributed in the tanks into each of which 0.4 l/s water were continuously pumped from Lake Tegel. The tanks were aerated and covered with a net and black foil. One tank was treated with 3 000 ppm methylene blue (Methylthionine chloride) for one week to prevent mycosis.
A total of 48 fish died, mostly with signs of fungal infection. As there was no significant difference in the mortality of marked and unmarked fish it was concluded that marking did not increase mortality.
1468 fish were examined for visibility of marks and 19 (1 %) were found to bear no mark.
Seining for recapture took place on November 12 (1.2 and 2.5 tons), 18 (0.5 tons) and 25 (2.2 tons). The catches from trawling and gillnetting were also used in the estimations. Every fish was examined for marks and the species were separated.
All roach were measured due to the low number caught. Bream and white bream were mixed together into boxes of 50 l. They were further processed in sets of 5 boxes. The fish in every fifth box were measured. In the first seine catch only the fish in the other 4 boxes were counted. All recaptured fish were measured.
A total of 40 453 fish comprising 9 545 bream, 30 209 white bream and 699 roach were checked for marks. Of these, 1 157 were marked. 35 848 of the fish in the recapture period could be used for the population estimates (Table 5), the other fish belonging to length classes which were not recaptured.
At recapture, 6 marked fish were found among 2586 fish which were controlled for undetected marks, so 0.2 % of marks were not detected. This percentage is added to the total number of recaptured fish.
The results of the Chi-square test for equality of r/m - ratio showed that it was possible to regard the whole size range as one sample for bream (13 – 35 cm) and roach (16 – 23 cm). For the white bream the length groups from 11 – 14 cm, 15 – 18 cm and 19 – 21 cm had to be calculated separately, the population estimates being calculated from the cumulation of the individual length groups.
