FISHERY BULLETIN: VOL. 75, NO. 3 



integration interval that would not result in un- 

 reasonable numerical errors being propagated 

 through many years of simulated time. This was 

 done by increasing the time step until the simu- 

 lation results converged. With an integration 

 interval of 0.005 yr, the results converged suffi- 

 ciently so that a numerical error of less than 57c 

 is expected after 23 yr of simulation (the length of 

 the data record used for verification). This level 

 of error was considered acceptable in light of the 

 precision of all the data upon which this work 

 was based. It was noted that each decrease in 

 the time step was accompanied by an increase in 

 the simulated catch; therefore, the predictions 

 yielded by the computer simulations are probably 

 slightly lower than would have resulted from an 

 exact solution of the model. 



The average length of age-groups 2-5 according 

 to the model for 1943-66, 1957-62, and 1962-71, 

 and the average length of these age-groups as 

 reported by Lux and Nichy ( 1969), and of the catch 

 for 1962-71 are compared in Table 7. Age-groups 

 2-5 were considered because they were most 

 abundant in available samples; and, therefore, 

 their means have smaller standard errors than 

 less abundant age-groups. 



Most of the fish measured by Lux and Nichy 

 were collected during 1957-62. Model results for 

 this period compare favorably as expected since 

 the model was designed to simulate the situation 

 reported by Lux and Nichy. The average simu- 

 lated lengths for 1943-66 are generally lower 

 than for the 1957-62 design period since the 

 design period had a lower temperature (favorable 

 to growth) than the longer time interval. 



The model tends to overestimate growth for 

 1962-71. The mean length of fish of a particular 

 age-group collected from the catch for 1962-71 is 

 lower than is predicted by Lux and Nichy's growth 

 function. This situation cannot be explained as an 

 effect of temperature. As has been the practice 

 throughout this work, the model was designed 



TABLE 7. — Average length (millimeters) of yellowtail flounder 

 age-groups 2-5 according to the model for 1943-66, 1957-62, and 

 1962-71 according to Lux and Nichy (1969), and for samples 

 from the commercial catch collected January-March 1962-71 

 (data provided by Northeast Fisheries Center). 



in accordance with the published literature; 

 therefore, some apparent overestimation of 

 growth in later years of the simulations is in- 

 evitable. This situation makes application of the 

 model less satisfactory for recent years, but part 

 of the effect of overestimating growth would be 

 compensated for by a shift in age-group structure 

 of the catch. If the model slightly overestimates 

 growth, there is a tendency to catch younger fish; 

 and, therefore, the effect of overestimating growth 

 is partially offset. 



The simulated size-category structure of cap- 

 tured (landed and discarded) fish for 1943-65 is 

 compared with unpublished data for 1963 as re- 

 ported by Brown and Hennemuth (see footnote 2) 

 in Figures 2 and 3. Clearly, it would have been 



i 



u 



OBSERVED O 

 SIMULATED • 



300 



350 375 



LENGTH , mm 



400 



FIGURE 2.— Simulated (1943-651 and observed (unpublished 

 data for 1963 as reported by Brown and Hennemuth (see footnote 

 2)) size-category structure of catch ( including discards) of yellow- 

 tail flounder. 



OBSERVED O 

 SIMULATED • 



300 325 



LENGTH, mm 



FIGURE 3.— Simulated (1943-65) and observed (unpublished 

 data for 1963 as reported by Brown and Hennemuth (see footnote 

 2)) size-category structure of discards as percentage of catch 

 (including discards) for yellowtail flounder. 



474 



