By Arill Engas and Egil Ona Institute of Marine Research P.O.Box 1870, N-5024 Bergen, Norway The constraint technique, i.e. using a rope between the warps, has the intention of minimizing the variability in trawl geometry with depth. Experiments with the constraint technique on two different bottom trawls used in the Norwegian surveys are reported and discussed. In addition, modifications of trawling technique, which enable the trawl doors to maintain constant ground contact under different conditions, are presented. INTRODUCTION Observations of bottom trawl performance using acoustic gear monitoring equipment have shown that the geometry of the trawl is quite variable. This is considered to be a major problem, since the data from surveys are interpreted, assuming a constant catch efficiency for each age group over areas and years (Canothers, 1981; Hylen et al., 1986). In the Norwegian bottom trawl survey for cod and haddock in the Svalbard Barents Sea area the depth of trawling varies from 50 m to about 600 m. Similar, the International Bottom Trawl Survey (IBTS) in the North Sea is canied out in areas from 30 m to 250 m depth, and difficulties with keeping a constant swept area over the entire depth range have long been recognized (Hagstr0m, 1987; God0 & Engas, 1989). The equipment and procedures used give reduced door and wing spread in shallow water and overextended door spread in deeper waters. Trials to con·ect the survey estimates for a depth dependent swept area have been made, based on measurements of door and wing spread (God0 & Engas, 1989; Koeller, 1990; Yatsu & J0rgensen, 1988). 2 The problem, however, is that when changing the door spread, the effectiveness of the entire trawl gear change. In patticular, a change in door spread will change the position of the herding sand cloud relative to the wing ends of the trawl, a critical area for escapement of fish (Main & Sangster, 1981a; Kortokov, 1984). A change in door spread may also influence the bottom contact of the trawl itself, as the bottom contact of the ground gear is reduced when overspreading the trawl.. As shown by Main & Sangster (1981b); Korotkov (1984); Engas & God~ (1989) and Walsh (1992), bottom contact of the trawl is critical for catching downward escaping species. Equally important is the ground contact of the door, i.e. the part of the weight of the door actually resting on the bottom, which affect the sand cloud fotmation, the acoustic noise produced by the door and the bottom contact of the traw 1 itself. We therefore question the validity of corrections only based on geometrical adjustment of the swept area, and recommend methods aiming to stabilize trawl petformance. In order to maintain constant trawl petformance at all fishing depths, several methods have been applied. Increased sweep length with depth have been recommended (Wilemann 1984; Palsson et al. 1989) and also a specified warp-to-depth ratio for different depths are commonly used (Anon. 1992). However, instrumented studies of tr·awl performance have shown that use of different sweep lengths is an unsatisfactory solution; increasing sweep with increasing depth results in higher door spread (Hagstr0m 1987). Furthermore, different sweep lengths have been shown to have a profound effect on the catch efficiency for different length groups of fish (Engas and God~ 1989). Realizing that the catch efficiency of the trawl is detetmined not only by the geometr·y of the tr·awl, but rather its entire performance, a method to reduce systematic vat·iability by using a constraint rope between the wat-ps was presented by the authors (Engas & Ona 1991). In this paper, experiences of using the method in order to obtain constant door spread and bottom contact with two different bottom trawls used in Notwegian surveys at·e presented and discussed. MATERIAL AND METHODS The trawls used during the experiments were the 36/47 m GOV-tr·awl and the Campelen 1800. The GOY-trawl is used during the IBTS in the North Sea, while Campelen 1800 is used during the Norwegian bottom tr·awl survey for cod and haddock in the Bat·ents Sea and Svalbard area. The experiments with the GOV trawl were conducted throughout the foutth quruter of the International Bottom Trawl Survey (IBTS) 1992 on RN "G.O. Sars". The GOV tr·awl was rigged with the recommended kite, floats, groundgeat·, 60 m sweeps (including backstr·ops of the doors) and 4.5 m Polyvalent doors (1100 kg) as described in the IBTS Manual (Anon, 1992). The recommended wat-p-to-depth ratio for the trawl was used on all hauls. At six of the fixed stations of the survey, the tr·awl haul was repeated over the same at·ea using the constraint technique. A rope (9 .0 m, 12 mm polyamide) was mounted between the wat-ps 150 m (the shottest wat'_P length permitted according to the IBTS Manual (Anon 1992)) in front of the doors (exact desctiption of mounting is found in Engas and Ona, 1991). A calibrated depth sensor was mounted in the middle of the constr·aining rope, or on the wat'_P 150 m in front of the door. 3 Investigations using the Campelen 1800 bottom trawl were carried out during surveys on RN "Johan Hjort" and RN "G.O.Sars in September 1992, and in February 1993 at depths between 50 and 370 m. The trawl was dgged with 40 m sweeps and 6.0 m Waco doors (1500 kg in air) according to the IMR trawl manual (Engas, 1988). Dudng trawling, the speed was maintained constant, 3.0 knots, as measured by the GPS. Constraining of the trawl warps was done as for the GOV trawl, and the normal warp-depth ratio used during the survey was not modified. A series of 25 comparative hauls coveting a depth range from 50 to 370 m , south west of Bear Island were made in September at the end of the 0-group survey from "Johan Hjort", alternating between free and constrained door spread over the same depth. Dudng the standard bottom trawl survey for cod and haddock in the Barents Sea, in Febtuary 1993, the constrain technique was used on 57 bottom trawl stations on both of the research vessels. In situ warp load were measured with a tensiometer (Scantrol, Bergen, Norway, Prototype with accuracy± 120 kg) on some hauls with both trawls. When measudng warp tension, the tensiometer was shackled into the wire between the warp and the door bracket. The speed of the gear through the water was measured during this experiments with a Scanmar speed sensor mounted on the central headline. The trawl geometry (vertical opening, door spread, and depth of constraining rope) was either logged continuously or each 30 sec. throughout the hauls, using Scanmar sensors. The depth sensor was calibrated with an accurate pressure sensor on the STD system. RESULTS AND DISCUSSION