within ± 50% of actual values. As population 

 size decreases, as it has in recent years, accuracy 

 decreases and abundance estimates may become 

 negatively biased. 



A second, and perhaps less biased, measure of 

 abundance for winter flounder was the CPUE 

 during a 4-week period from mid-March through 

 early April, the only period which included com- 

 parable data for all annual surveys. The median 

 CPUE was used as the most appropriate catch 

 statistic because the trawl catch data were not 

 normally distributed and were positively skewed. 

 A 5-mean CPUE was used as an index of abun- 

 dance for other trawl data sets where the data 

 series had zero observations. However, the 

 Niantic River abundance surveys had only a few 

 tows without fish throughout the past 12 years. 

 A comparison of the two indices showed that 

 they were highly correlated for winter flounder 

 both larger and smaller than 15 cm (Spearman 

 rank-correlation, r = 0.96 and 0.98, respectively). 

 Therefore, the median CPUE was retained as the 

 best measure of abundance for the Niantic River 

 spawning stock. 



The 1987 median CPUE of 13.7 was similar 

 to the 1986 value of 12.0 (Table 5; Fig. 4). Mean 



tow duration and catch of winter flounder from 

 tows of one-half (n = 2; 6.6 min; 6.0 fish) and 

 two-thirds (n = 79; 7.8 min; 1 1.3 fish) of the stan- 

 dard tow were reasonably proportional to those 

 for the regular distance of 0.55 km (n= 137; 12.1 

 min; 18.2 fish) and shorter tow length did not 

 affect the calculation of median trawl CPUE in 

 1987. Annual trends in median CPUE generally 

 corresponded with the JoUy composite index of 

 abundance until 1982. The CPUE in 1982 (42.6) 

 was nearly the same as in 1981 (43.4), but the 

 Jolly abundance index increased 72% (28.7 to 

 49.4). However, the latter 1982 estimate had a 

 large confidence interval (± 16.7). The decline in 

 CPUE for fish larger than 20 cm from 1983 (22.1) 

 to 1984 (12.8) and 1985 (12.6) was greater than 

 for the Jolly abundance index (29.9 to 29.3 and 

 21.6). The Jolly index for 1986 (8.3) decreased 

 more than 60% relative to 1985, but the CPUE 

 decreased by only about 20%. Both the JoUy 

 index and CPUE showed similar increases (22%i 

 and 14%, respectively) in 1987. The CPUE for 

 1984-87 indicated population levels about one-half 

 of those during 1976-80, which also contradicted 

 the Jolly abundance indices. These differences 

 between abundance indices and biases of both 

 were discussed at length in NUSCO (1987). 



TABI/E 5. Median CPUE of Niantic River winter flounder larger than 15 cm from 1976 through 1987 

 during the period of mid-March through mid-April. 



Zero for symmetrically distributed data. 



Winter Flounder Studies 



161 



