model. In the following discussion of the model, 

 the recruitment index was determined as the frac- 

 tion that age 3 (21.4-28.0 cm) fish made up of 

 each annual Niantic River adult CPUE. These 

 lengths represented the midpoints of the 10th and 

 90th percentiles of length between ages 2 and 3 

 and ages 3 and 4, respectively. The index of age 

 3 fish was multiplied by a scaling factor (1.938) 

 reflecting their lifetime reproductive contribution 

 (maximum life span was assumed to be 12 years). 

 Age 3 fish were used as the best estimate of re- 

 cruitment since a majority of them were mature 

 and in the Nicintic River for spawning each year. 

 This also extended the period of compensatory 

 mortality throughout the entire period of imma- 

 turity without restricting it to a particular larval 

 or juvenile life stage. The index of parental stock 

 producing the age 3 fish was the fraction of each 

 annual CPUE made up by all adults age 3 and 

 older (^21.4 cm) during the spawning season 3 

 yeais prior to the age 3 recruitment estimate. 



Jones (1982) noted that the largest stock abun- 

 dances for a number of species were usually about 

 three to six times that of the smallest ones and 

 that the difference between them usually approx- 

 imated the mean. Ursin (1982) reported that 

 recruitment variation over 13 years for the North 

 Sea plaice {Pleuronectes platessa), which is an 

 European flatfish closely related to the winter 

 flounder (Burton and Idler 1984), varied by a 

 factor of five from the smallest to the largest year- 

 class. Many other species, such as cod {Gadus 

 morhud) and herring {Clupea harengus), also had 

 ratios of this order. However, a few fishes (e.g., 

 haddock, Melanogrammus aeglefmus) varied by a 

 factor of 100 or more, suggesting the lack of a 

 stabilizing mechanism for recruitment. For the 

 Niantic River parental stock index, the range seen 

 from 1976 through 1987 was 10.2 to 32.3 (differ- 

 ence of 22.1) with a mean of 19.0. This suggested 

 that our stock and recruitment data set probably 

 included representative small and large stock sizes 

 typical for this population. The two Rhode Is- 

 land commercial fishing indices had ratios of abun- 

 dance that were 4 and 1 1 and the URI series had 

 a difference of 15. However, the former may 



have been influenced by economic and social fac- 

 tors and the latter time-series included many ju- 

 veniles, which, as illustrated by our data, can be 

 quite variable in number and may not absolutely 

 reflect true abundance. 



A plot of recruit versus parental indices sug- 

 gested a dome-shaped curve with reduced recruit- 

 ment at high levels of adult abundance. This is 

 typical for fishes with high fecundity (Gushing 

 1971; Gushing and Harris 1973), although the 

 reliability of this relationship will be ascertained 

 as more data are collected. Because a dome was 

 presumed for the stock-recruitment data, Ricker's 

 (1954, 1975) model was fit to the data. His two- 

 parameter (a and P) model did not explain much 

 (R = 0.44) of the variability seen in recruitment 

 (Fig. 13). For example, the 1978 and 1984 year- 

 classes were remarkably different, although pro- 

 duced by similar adult abundance. 



Numerous examples have been given where 

 recruitment success of a species has been strongly 

 linked with environmental factors (Gushing 1973, 

 1977; Sissenwine 1974, 1977, 1984; Roff 1981; 

 Shepherd et al. 1984; Lorda and Grecco 1987). 

 In particular, water temperature has been found 

 to be inversely related to strong year-classes of 

 winter flounder in Rhode Island (Jeflries and 

 Johnson 1974; Jeffries and Terceiro 1985; Gibson 

 1987). Roff (1981) also noted that abnormally 

 cold temperatures have been related to strong 

 year-classes in several other winter-spawning 

 flatfishes. To examine possible temperature ef- 

 fects, yearly mean water temperatures during the 

 winter flounder spawning and larval seasons 

 (January- May) were calculated for single months 

 and for various combinations of two or more 

 months. For each month or group of months, 

 deviations from the long-term mean were com- 

 puted and compared to annual recruitment indices. 

 The strongest negative correlation (r=-0.78) was 

 found between February temperature deviations 

 and recruitment indices (Fig. 14). Gorrelations 

 between recruitment and March and February- 

 March temperatures were also significant. 



Winter Flounder Studies 



173 



