358 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



(water speed minus velocity of opposing surface 

 current) to render forward progress of trawling gear 

 almost negligible. Under the same conditions but 

 traveling in the opposite direction, indications 

 were that too high a resultant ground speed kept 

 the gear off the bottom a high proportion of the 

 time, even at reduced power settings. This was 

 an extreme situation involving operations in 

 deep water (150 fm.) and a very confused sea, but 

 it does serve to illustrate that there is always a 

 "best" combination of factors that results in maxi- 

 mum efficiency for any piece of gear. It would 

 seem quite unlikely that this combination is 

 attained in every fishing operation (see also 

 Dickson, 1961). Also, all other factors being 

 equal, larger trawlers with greater horsepower 

 ratings tend to outfish their lesser counterparts 

 (U.S. Fish and Wildlife Service, 1959). Such 

 factors interacting to varying degree and generat- 

 ing operational biases of unknown magnitude, 

 conceivably play a major role in governing the 

 accuracy of the fishable biomass index. 



At least one form of operational bias, that due 

 to differential power of trawlers, has been the 

 subject of detailed study. Gulland (1956) pro- 

 vides a method for its elimination if accurate 

 effort and catch data can be obtained for individual 

 vessels on a per unit space and time basis. As 

 already noted, however, effort statistics used in 

 the present study were projected from sample 

 interview data on the premise that all trawlers are 

 equally efficient. Such treatment automatically 

 nullifies "standardization" of available effort data. 



The second class of defects includes miscella- 

 neous error or bias arising from natural factors. 

 For example, patchy distribution of shrimp could 

 result in highly variable catches by individual 

 vessels despite uniform effort, the magnitude of 

 corresponding "sampling error" being such that 

 index precision is greatly diminished. Also, 

 "saturation" of trawls by the species sought as 

 well as bj' associated fauna, e.g., other inverte- 

 brates and fishes, reduces gear efficiency and 

 thereby compounds the inaccuracy of the popu- 

 lation density index. Finally, all of a population's 

 fishable biomass may not be available because 

 rough bottom and sundry impediments preclude 

 trawling over portions of its geographic range. 

 Assumptions 



Once the purposes of a statistical survey are 

 clearly defined, an objective should be to mini- 



mize the combined effects of sampling error and 

 bias on the estimates being sought. This implies 

 that, in situations such as described here, detailed 

 information on: weather and sea conditions; 

 trawling coiu-se with respect to wind and cm-rent; 

 water speed; vessel size, horsepower, gear reduc- 

 tion ratio, and screw specifications; fishing gear 

 specifications; incidence of miscellaneous fauna in 

 the catch; etc. should accompany basic effort and 

 catch data, all recorded on a per vessel-trip, per 

 unit space and time basis. Appropriate adjust- 

 ments would eliminate unwanted effort bias to a 

 substantial degree. Refined catch-effort data 

 would permit a more sophisticated statistical 

 treatment and hence a minimizing of sampling 

 error. Higher quality biomass indices would 

 result. 



The condition of available effort data plus the 

 lack of information that would allow adjustment 

 for bias and reduction of error, thwarted attempts 

 to rectify defects in the data used to construct 

 biomass indices. As a consequence, some assump- 

 tions regarding the magnitude and effect of error 

 and bias associated with operational and natural 

 factors had to be made. Thus, acknowledged 

 uniformity in size composition of major portions 

 of both inshore and offshore trawler fleets led to 

 the conclusion that bias due to differential effi- 

 ciency of operating unit was probably not too 

 great. Effects of varying crew ability, climato- 

 logical and sea conditions, and contagious distri- 

 bution of shrimp were, within each of the smallest 

 time increments employed, assumed random with 

 zero expectation. And, any shrimp otherwise 

 vulnerable but unavailable because of untrawlable 

 bottom presunnibly comprised a constant fraction 

 of the total shrimp biomass. Since most interest 

 attached to population trends generated over a 

 4-year period, an overriding assumption was that 

 the direction and magnitude of all error or bias 

 remained constant for each time interval within 

 that period. 

 Computation 



Calculation of monthly indices for those portions 

 of a population's fishable biomass that occupied 

 offshore areas proceeded according to Gulland 's 

 (1956) technique. Tlie seaward limit of each 

 species "commercial" range was arbitrarily set at 

 the 45-fathom contour, this decision was based upon 

 catch statistics (appendix), depth distribution 

 studies (e.g., Burkenroad, 1939), and miscella- 



