compared to other indices of winter flounder 

 abundance to determine if similar trends were ap- 

 parent. These included National Marine Fisheries 

 Service (NMFS) statistics for Rhode Island and 

 a University of Rhode Island (URI) long-term 

 (1966-86) trawl survey in Narragansett Bay (M. 

 Gibson, Rhode Island Department of Environ- 

 mental Management, pers. comm.), and 

 Connecticut-licensed trawler catches from 1979 

 through 1987 (E. Smith, CT DEP, pers. comm.). 

 The NMFS statistics were CFUE expressed as 

 the number of pounds of winter flounder per 

 50-ton unit for Rhode Island vessels (1947-87) 

 and as the number of pounds per directed trip in 

 Statistical Area 539 (Narragansett Bay and Block 

 and Rhode Island Sounds), available from 1964 

 through 1987. The Connecticut commercial 

 CPUE was calculated as the number of pounds 

 per hour of trawling. The URI data are annual 

 5-means determined from monthly sums of winter 

 flounder taken during weekly sampling by trawl. 

 Due to an apparent preponderance of juveniles 

 in the latter time-series (Gibson 1987), the annual 

 CPUE values were lagged 1 year before making 

 comparisons. Additional information concerning 

 the URI time-series may be found in Jeffries and 

 Terceiro (1985) and Jeffries et al. (1986). Abun- 

 dance indices were correlated using the 

 nonparametric Spearman rank-order correlation. 

 Historical records of winter flounder catches given 

 by Perhnutter (1947) and NMFS (1987) were also 

 examined. 



The annual abundance data for the Niantic 

 River spawning stock from 1976 through 1987 

 were used to develop a stock and recruitment 

 model. All fish 21 .4 cm and larger were considered 

 to be adults (age 3 and older). Annual median 

 CPUE values of winter flounder 15 cm and larger 

 were adjusted to account for fish larger than this 

 minimum length to estimate relative parental 

 stock size for each year. Similarly, the CPUE 

 were partitioned to determine a value for age 3 

 (21.4-28.0 cm) fish each year. The latter value 

 was multiplied by 1.938 in order to account for 

 life-time reproductive contribution and to scale 

 the values into similar units for comparison with 

 parental stock size estimates. This scaling factor 



was determined from estimates of maturity and 

 mortality (NUSCO 1987) and an assumed max- 

 imum life span of 12 years for Niantic River win- 

 ter flounder. The adjusted CPUE values of the 

 age 3 recruits were plotted against the parental 

 stock size CPUE for 3 years prior. As a dome- 

 shaped recruitment curve with reduced recruit- 

 ment at high parental stock sizes was suggested 

 by this procedure, a Ricker (1954, 1975) stock 

 and recruitment model was fit to these data using 

 nonlinear regression methods (SAS Institute Inc. 

 1985) where: 



R = aPexpC-pP) 



(2) 



where R = CPUE index of the number of age 3 

 recruits 



P = CPUE index of parental stock size 



a = dimensionless parameter associated 

 with density-independent mortality 



P = parameter describing the instantane- 

 ous rate of density-dependent mor- 

 tality with dimension of 1/P 



Water temperature, believed to be an important 

 environmental variable in the early life history of 

 winter flounder, was considered as an additional 

 parameter in order to improve the fit of the ob- 

 served data to the model. Continuous water tem- 

 perature records were available from the intakes 

 of MNPS Units 1 and 2 and mean water temper- 

 atures during 1976 through 1987 were found for 

 individual and for various combinations of months 

 encompassing the January through May spawning 

 and larval recruitment period. Means over an 

 1 1-year (1976-86) period were also determined for 

 the same months and groups. For each time 

 period, deviations from the long-term mean were 

 computed and compared (Spearman rank- 

 correlation) with annual estimated age 3 recruit- 

 ment. Water temperature, expressed as a positive 

 or negative deviation from the long-term mean, 

 having the highest correlation with recruitment 

 was used as a third parameter in the Ricker model: 



R = aPexp(-pp)exp(-(pT) 



(3) 



Winter Flounder Studies 



153 



