Once the basic decisions are made about sample design (vessel, gear, season, and area of survey, stratification and station allocation scheme, 12 h or 24 h operation, speed and duration of standard haul) there is still a great deal of planning and preparation to be done including scheduling various segments of the survey, staffing the scientific parties and crew, assembling supplies and sampling equipment, etc. In this section only the basic elements are briefly outlined, and other survey manuals should be consulted for additional details (e.g., Grosslein, 1969a and 1969b; Doubleday, 1981).
Stations should be selected in advance of the survey and plotted on the navigation charts to be used by the ship's officers. In the case of stratified random sampling it is simply a matter of drawing simple random samples of stations within each stratum, with or without replacement. Drawing samples without replacement spreads sampling slightly more evenly but in the strict sense inflates stratum variances slightly. The gains of sampling without replacement probably outweigh the losses but the differences are very small.
Known areas of untrawlable bottom should be excluded in advance from the survey area in the stratification scheme. Careful records should be kept on locations of new rough bottom areas encountered during each survey with appropriate revisions in the sampling strata. Repair or replacement of gear plus lost time can be very costly when attempting to trawl on rough bottom. In some areas it may be necessary to fish essentially in the same location on each survey because of the risk of damage, e.g., edges of canyons or coral reefs. Different gear types (e.g., traps or gillnets) might be suitable for untrawlable grounds.
Selecting a cruise track can be done mathematically in a way which minimizes total steaming distance. However, there are nearly always practical constraints which mitigate the value of using a strict mathematical minimization technique. A simple track drawn by eye using common sense is usually adequate, and in most cases the track has to be changed or interrupted anyway due to weather or other contingencies.
Upon arrival at a station, standard routine procedures should be followed. Frequently oceanographic measurements (temperature, salinity, etc.) are made first and then the trawl is set. Ideally the direction of haul should be randomized at each station unless it is possible to tow in a constant direction relative to the current (the latter requires accurate real time information on the current, which is rarely available). In practice, it is often not possible to tow strictly at random (because of depth contours or nearby rough bottom), and in any case cost is of sufficient importance so that most survey designs call for towing in the direction of the next station to reduce total steaming time.
The same exact procedure should be used in setting and hauling the trawl at each station, allowing of course for standardized changes in scope relative to depth. It should be emphasized again that speed, and rigging of the trawl and otter doors must be kept exactly standard at all times. Trawl performance in terms of fishing power probably depends on both speed through the water and over the bottom and they are not necessarily the same, depending on currents. Furthermore, vulnerability of fish, especially those on or near the bottom, may depend on direction of tow relative to direction and speed of current against which the fish may orient themselves. Also, wingspread and headrope height of the trawl may vary whether towing against or with the current even when standardizing speed through the water by means of a speed log at the surface. It is not possible to standardize both speed through the water and over the bottom at the same time; the choice will depend on instrumentation available. If no speed logs are available then standardization of rpm and duration of haul are the next best; adjustment of actual distance towed after the fact may then be possible depending on accuracy of navigational equipment.
In the event of damage to the gear, some protocol must be established to decide whether or not the haul will be counted. Also, if rough bottom is encountered a decision rule should be devised for choosing another trawling location, e.g., the next sampling unit in the direction of the next originally scheduled station (for examples see Grosslein, 1969a, Doubleday, 1981).
It is not possible to design an optimum biological sampling protocol for a survey without a detailed listing of the kind of biological information sought ranked in order of priority, and a detailed knowledge of the operational characteristics of the vessel relative to sampling (deck space, laboratory space, etc.), size of the scientific party, type of sampling equipment, etc. It will suffice here to indicate that the basic minimum data on multispecies trawl surveys should include the total weight and a length frequency sample of each individual species in the catch. The size of sample and the proper sampling procedures vary with circumstances but nearly always involve sorting major species from the catch prior to taking stratified (by length) samples for length frequency.
The total catch of small uniform-size fishes is often recorded simply as the number of baskets of unsorted fish, and a small random sample of baskets are then weighed, sorted by species, and sampled for length frequency. The principal requirement is to use a sampling procedure which provides an unbiased (representative) sample of the catch (assuming only a sample is processed), and careful records of the exact size and nature of the sample so that proper expansion factors can be calculated. Examples of detailed procedures are given in Grosslein (1969a).
Careful record-keeping at sea is as important as using care in standard trawling methods and proper catch sampling techniques. Again the exact data to be recorded and the optimum format of logs vary with the type of information sought, training, and size of scientific party, and also the kind of data processing to be done. In this case it is assumed that data will be recorded on some kind of log for processing ashore. A trawl log format should be designed by the scientists who will be responsible for collecting and analysing the data. The log should allow for recording all the essential information but be simple and compact. However, clarity and accuracy should be the first priority. There are advantages to having most data recorded on a single log, i.e., station data and biological catch and sampling records, so that recording time at sea is reduced, and having to identify different logs from the same station is avoided. It is extremely important that the scientist in charge of the watch check each log for accuracy and completeness immediately after finishing the sampling at each station. Examples of trawl log formats and record keeping methods are given in several documents issued during the training course (Grosslein, 1969a; Halliday and Kohler, 1971; Doubleday, 1981). It should be noted that the single log used in US east coast surveys serves the dual function of a record form for use at sea and a form for coding data for entry into an ADP system.
