may be imprecise unless sampling intensity is relatively high (Cormack 1979; Buckland 1980; Nichols et 

 al. 1981; Hightower and Gilbert 1984). The review concluded that sampling intensity and recapture rates 

 of Niantic River winter flounder were rather low and may have resulted in errors of the estimates. 

 Furthermore, estimates of recruitment (Jolly's B) were undoubtedly unreliable because they accumulated 

 errors from other parameters in the model (also see Amason and MUls 1981 and Hightower and Gilbert 

 1984). Adding recruitment estimates to an initial estimate of abundance (Jolly's N), as done in previous 

 years to obtain absolute abundance estimates, was shown to be highly inaccurate. This led to the 

 formulation of the composite index of abundance based on the Jolly model, which used the average of 

 the central estimates of N and a pooled variance estimate (NUSCo 1986a). 



Using results from a simulation study by Hightower and Gilbert (1984) and based on sampling 

 intensities of 3 to 5.4%, population sizes of 20 to 40 thousand, and survival of about 0.90 ("deaths" 'here 

 are mostly due to emigration from the Niantic River), it was assumed with 95% confidence that errors in 

 estimating abundance using the Jolly model ranged from 25 to 50%. These levels of accuracy are sufficient 

 for management purposes, but are not within the 10% level of error that Robson and Regier (1964) 

 recommended for research investigations. Hightower and Gilbert (1984) demonstrated that sampling in- 

 tensities had to be doubled to increase the accuracy from 25 to 50%. Substantially higher and most likely 

 infeasible levels of effort would be required to achieve sampling intensities necessary for the accuracy level 

 of 90% for current population sizes in the Niantic River. 



The CPUE was also examined for bias and accuracy (NUSCo 1983b, 1986a). The standardization of 

 fishing effort is difficult and although the catchability of a species is presumed constant, the fishing gear 

 can vary in efficiency and can be affected by subtle changes in its use, rigging, or deployment from different 

 vessels (GuUand 1983). Nevertheless, repeated surveys using relatively consistent methods can provide an 

 index of abundance free of difficulties caused by possible changes in catchability. Among the factors 

 examined that could have influenced Niantic River winter flounder CPUE were changes in boats, trawling 

 methodology, and variable annual conditions in the Niantic River. 



With some exceptions, the vessels used during sampling changed annually from 1976 through 1981 

 (Appendix XXIII). Several comparisons between boats showed significant differences in catches among 

 years (NUSCo 1983b), indicating possible differences in fishing power. Since 1981, two vessels identical 

 except for rigging have been used. Northeast /, rigged exclusively for trawling, consistently caught more 



33 



