deterministic population model (I less et al. 1975; 

 Saila 1976) were used when, only 3 to 4 years 

 into the winter flounder research program, com- 

 plete data specific to the local winter flounder 

 population were not yet available (NUSCO 1983). 

 In recent years, the work has increasingly been 

 directed towards more specific and detailed studies 

 of earl> life history and on the critical stock- 

 recruitment relationship. This has been reflected 

 by studies designed to address specific concerns 

 of NU and the Connecticut Department of En- 

 vironmental Protection (CT DEP) (NUSCO 

 1988a, 1988b). The information will be incorpo- 

 rated into the models currently under develop- 

 ment, which should be available for use within 

 about a year. This report discusses the results of 

 ongoing winter flounder research and provides 

 brief descriptions of the models that will be used 

 for impact assessment. 



Materials and Methods 



Adult and juvenile studies 



Abundance estimation of the Niantic River 

 spawning population of aduh winter flounder has 

 been based on mark and recapture methodologies 

 and details concerning annual surveys from 1976 

 through 1986 are summari7,ed in NUSCO (1987). 

 Since 1982, each survey started after ice-out in 

 the river in mid to late February and ended in 

 early April, when the proportion of reproductively 

 active females decreased to less than 10% of all 

 females examined for two consecutive weeks. The 

 Niantic River was subdivided into a number of 

 areas (stations) for each survey (I'ig. I); no sam- 

 ples were taken outside of the navigational channel 

 in the lower portion of the river since 1979 be- 

 cause of an agreement with the liast I,yme- 

 Waterford Shellfish Commission to protect bay 

 scallop [Argopecten irradiam) habitat. Collections 

 were made during 2 or 3 days each week and 

 were usually allocated to a station according to 

 the expected abundance of winter flounder with 

 more tows taken tn areas where fish were most 

 numerous. 



Winter flounder were captured with a 9.1-m 

 otter trawl (6.4-mm bar mesh codend liner), which 

 has usually been towed at a standardized distance 

 of 0.55 km since 1983; tows of variable length 

 were taken previously. The standard distance was 

 chosen because it represented the maximum tow 

 length at station 1 and because using the same 

 tow length at all stations was expected to reduce 

 variability in calculating catch-pcr-unit-effort 

 (CPUE), used as an index of abundance. In 

 1987, tows one-half or two-thirds of this length 

 were frequently taken in the upper river (mostly 

 at station 51) to avoid overloading the trawl with 

 algae and detritus. Tow time and numbers of 

 fish from tows of shorter distances in 1987 were 

 compared to those from tows of the standard 

 distance (0.55 km) to determine if the catches 

 were proportional to distance of tow for compu- 

 tation of CPUE. Because catch data from station 

 2 were also used for the trawl monitoring program, 

 hauls at that station were maintained at the stan- 

 dard distance of 0.69 km for that sampling pro- 

 gram. 



Mostly because of differences in tidal currents, 

 wind, and amounts of extraneous material col- 

 lected in the trawl, tow times for the standardized 

 distances varied and were usually greater in the 

 lower than in the upper river. For 1983-87, the 

 mean duration for standardized tows at stations 

 I and 2 was 15.7 min and at stations 4, 51, 52, 

 53, and 54 was 12.5 min. Tows from 1976-82, 

 when tow distance varied, that had shorter or 

 longer duration compared to the distribution of 

 tow times from 1983-87, when tow distance was 

 uniform, were excluded from data analyses and 

 calculation of CPUE. For comparisons among 

 years, all catches of winter flounder larger than 

 15 cm made during a 4-weck period from mid- 

 March through early April were standardized to 

 cither 15-min tows (stations 1 and 2) or l2-min 

 tows (all other stations) and annual median CPUE 

 values were determined. A 95% confidence in- 

 terval was calculated for each median using a 

 distribution-free method, where the order statistics 

 for the upper and lower confidence limits were 

 plus or minus the square root of the sample size 

 from the order statistic describing the median 



150 



