longer-term cycle of abundance. Even though the forecast percent error 

 for each of these models was high, the actual 1982-83 data fell within 

 the 95% confidence interval in most cases. However, the confidence 

 intervals themselves were fairly wide because of the variability of the 

 data. 



Yearly changes in abundance were noted both in the Jolly population 

 estimates and in the harmonic regression models of the trawl monitoring 

 program data. These changes were not unusual for the winter flounder 

 and most likely reflected long-term natural fluctuations. Jeffries 

 (1983) noted an 11-yr cycle of abundance for Narragansett Bay winter 

 flounder. He reported a decline of 86% in his standardized trawl catch 

 from 1968 to 1976, a rapid recovery culminating in a peak in 1979 

 similar in magnitude to 1968, and yet another rapid decrease of 58% by 

 1982. He attributed the cyclic changes in number to a subtle warming 

 and cooling trend which may have caused the observed changes in number; 

 his temperature model accounted for 92% of the variability. Jeffries 

 hypothesized that temperature variability affected predation on larvae 

 during metamorphosis in April, which he termed a critical month. 

 Similar correlations of year-class strength with environmental factors 

 have been summarized for other flatfishes by Roff (1981). Nisbet and 

 Gurney (1982) noted that fish populations can experience fluctuations in 

 abundance approximately two to four times longer than the length of time 

 it takes the species to mature, depending upon the population 

 age-structure of the species in question. Since the winter flounder 

 typically matures in 3 to 4 years, a 6- to 12-yr cycle of abundance 

 could be expected. 



SUMMARY 



The minimum size for marking winter flounder during the population 

 abundance survey in the Niantic River was increased from 15 to 20 

 cm in 1983 and the sampling was completed after most spawning 

 occurred. The number of adult winter flounder was estimated as 

 41, 980 ± 15,564. The 1983 data were examined for potential 

 sources of bias or error; the estimate was relatively unbiased and 

 precision good. 



55 



