This survey was intended to permit efficient, unbiased estimation of discarding and retention of cod, haddock and whiting by English fishing vessels over 12 m registered length in the North Sea in four sampling periods during 1997 and 1998 using only two sea-going observers. The 'probability proportional to size' (PPS) sampling method was used to avoid problems previously experienced with stratification of the survey by gear. Vessels were selected with replacement and with probability proportional to predicted catching power and effort, and inversely proportional to average trip length, thereby giving priority to those expected to catch the most fish whilst maximising information for each observer-day at sea. A trip on each selected vessel was then observed in the order of drawing so far as practically possible, giving a reasonably random selection from those made by each vessel in the sampling period. PPS estimators were formulated as if vessels were the sampling units. The values measured on each were the total numbers of fish of each species discarded and retained by the vessel over the sampling period as estimated by raising numbers observed on a trip to numbers for the period using a factor based on hours fishing. The effects of measurement errors on the PPS estimators are assessed. Results for trips which could not be observed, e.g. for safety reasons, were modelled to make use of known effort by the vessel. Discarded proportions of numbers caught by the fleet were estimated to be 20-48% for cod, 30-41% for haddock, and 51-65% for whiting. Coefficients of variation (CVs) for numbers discarded varied between 20 and 50% based on 14-25 fishing trips observed in each sampling period. Random sampling improved knowledge about the fleet and confidence in catch estimates. Reducing the number of long trips observed improved the information gained per day at sea but otherwise the PPS method proved not to be as statistically efficient as had been hoped mainly because catch and effort by the NE coast fleet were not accurately foretold by historical data. However, this was not a problem peculiar to PPS and ease of implementation commended PPS over competing designs given a small number of observers, variable trip durations, difficulties with stratification, and ideally. reliable information about the fleet in a forthcoming sampling period. (C) 2002 Elsevier Science B.V. All rights reserved.
|Number of pages
|Published - Mar 2002
- probability proportional to size
- sample survey
- North Sea