SISSENWINE: COMPARTMENTALIZED SIMULATION MODEL 



the mean size of younger fish was higher than 

 observed while the converse applied to older fish. 

 The differences were generally small. The sizes 

 of the most abundant fish in the catch (age-groups 

 3 and 4) were well simulated. While the model 

 adequately simulates growth, more precise re- 

 sults might have been obtained by assuming an 

 average maximum size in excess of 500 mm. The 

 result, with fishing, would be an average max- 

 imum size near the value assumed by Lux and 

 Nichy ( 1969). Thus the assumed value of k 2 would 

 have been more appropriate. 



The parameters of c 12 and c 14 specify the tem- 

 perature dependence of the model. Estimates of 

 these parameters were based on Sissenwine's 

 (1974, 1975) calculations of recruitment and 

 average growth per fish for 1944-65. No attempt 

 was made to improve these estimates by tuning 

 them to the model. While Figures 4 and 5 indicate 

 the adequacy of the model and its parameters, 

 these figures also reveal that catch was generally 

 overestimated during warm years and under- 

 estimated during cold years. This implies that 

 the fishery was probably more sensitive to tem- 

 perature than indicated by estimates of c 12 and c 14 . 

 Rather minor adjustment of these parameters 

 would probably account for much of the remaining 

 unexplained variation in catch. On the other 

 hand, since tuning in effect reduces the residual 

 degree of freedom and, more subjectively, reduces 

 confidence in the model, no attempt was made to 

 improve the initial estimates of c 12 and c u . 



Adult female yellowtail flounder are generally 

 longer than males of the same age. The model does 

 not distinguish between sexes. To do so would 

 require doubling the central processing time 

 required to run the model. Fishing pressure would 

 tend to shift the sex ratio in favor of males because 

 of this size difference. Since the sex ratio (c u = 0.5) 

 was estimated for the exploited population, the 

 influence of fishing was incorporated into the 

 model through the estimation of this parameter. 

 Variations in c u resulting from changes in level 

 of fishing were not simulated. 



Since females are larger than males, the total 

 fecundity of the population is underestimated 

 when based on the mean size of the age-size com- 

 partments. This bias is probably compensated for 

 by overestimating mean recruitment per egg (c 13 ). 

 Thus, expansion of the model to segregate fish 

 according to sex should not affect the results re- 

 ported here, although some revision of c J3 would 

 be required. 



In recent years, several changes have occurred 

 in the Southern New England yellowtail flounder 

 fishery that were not reflected in the model. 

 During the late 1960's, more active industrial and 

 distant water fisheries (using small mesh nets) 

 for the yellowtail flounder developed. The fish- 

 eries statistics used in this report do not include 

 the catch of the industrial fishery which in a few 

 years equaled 20 r /r of the total yield. Estimates 

 of the catch of the distant water fishery are in- 

 cluded and the fishing effort of the distant water 

 fleet is estimated by assuming that the catch per 

 unit effort was the same as for the domestic fish- 

 ery. Since 1971, the fishery has been regulated by 

 quotas set by ICNAF. During the 1970's, landings 

 of yellowtail flounder within ICNAF Subarea 6 

 (south of the Southern New England ground 

 which is within ICNAF Subarea 5) have in- 

 creased. The relationship between the fisheries in 

 these two areas is unknown (Brown see footnote 3; 

 Parrack see footnote 4). These recent changes 

 necessitate several modifications of the model 

 before it can be used to simulate the present 

 fishery. 



The work reported here indicates the potential 

 for predicting future trends of certain well-studied 

 fisheries in which the role of a specific environ- 

 mental variation can be described. Two important 

 limitations of this approach are demonstrated. 

 Firstly, model parameters may change with time; 

 thus it is important to keep the model up-to-date. 

 This does not imply that the model should be regu- 

 larly tuned to assure that it successfully predicts 

 each additional year of data but rather that 

 parameters be updated as evidence of change in 

 the fishery becomes available. Secondly, numer- 

 ous fundamentally different models may be 

 nearly as successful in simulating a specific sys- 

 tem. Therefore, it is dangerous to limit considera- 

 tion to a single model or regulatory mechanism. 



ACKNOWLEDGMENTS 



I thank Saul B. Saila for his support throughout 

 this work. Numerous valuable constructive com- 

 ments on the manuscript were provided by Brad- 

 ford Brown, Judith Brennan, and Richard Henne- 

 muth. Ilene Sissenwine edited and proofread the 

 typescript. Part of this work was completed in 

 partial fulfillment of the requirements for the 

 degree of Doctor of Philosophy at the University 

 of Rhode Island and was sponsored by the Office 

 of Sea Grant, NOAA, U.S. Department of Com- 



481 



