LAURS and WETHERALL: GROWTH RATES OF NORTH PACIFIC ALBACORE 



Mendocino, Calif., and the central North Pacific 

 east of 180°(Laurs'*). 



The hypothesis of equal growth rates was tested 

 using the weighted zero-intercept analysis of 

 covariance, and was rejected at the 0.57c 

 significance level (Table 4, Figure 2). In pairwise 

 comparisons between individual groups, the only 

 nonsignificant difference in growth rate was be- 

 tween Groups B and C. 



Are the observed differences in growth rates of 

 tagged albacore consistent with other informa- 

 tion? To check this, we examined the length 

 composition of albacore catches along the U.S. 

 west coast (Figure 3). The length-frequency plot 

 for catches north of lat. 38° N during the period 

 when most recaptures were made, 1972-78, 

 showed modes at about 64 and 76 cm and a hint of 

 one at 54 cm. Catches south of lat. 38° N showed 

 the 54 cm mode, but had primary modes at about 

 66 and 79 cm. The discrepancy between modes of 

 the older albacore is further evidence of a slower 

 growth rate for North fish, assuming these modes 

 represent fish of the same age. To see if length- 

 frequency data and tag data agreed, we computed 

 the expected fork lengths under each K at annual 

 time steps and compared these with observed 

 modes in the length-frequency distributions. 

 Starting with some initial fork length, Li, we 

 used the equation L, =a +bL,-i, i =2, 3,..., 

 where a = Lj- (1 - expi-K)) and b = expi-K). 

 Setting Li = 54 and Ly- = 125 cm, we found the 

 sequence of lengths 54.0, 66.0, and 76.0 cm for the 

 North group albacore; and 54.0, 68.6, and 80.3 cm 

 for the South fish. These are reasonably con- 

 sistent with the observed sequences of length 

 modes. 



Extended Model 



Plots of residuals from the standard model 

 against days out (Figure 4) showed a tendency 

 toward negative deviations during the first sev- 

 eral months after tagging, suggesting that some of 

 the residual variation could be attributed to "lack 

 of fit" (Draper and Smith 1966). For example, of 

 the 221 recaptures analyzed in Group A, 90 were 

 taken within 6 mo of tagging, and 70%of the 

 Model 1 residuals corresponding to these early 

 recaptures were negative. We therefore fit Model 



^Laurs, R. M. 1979. Results from North Pacific albacore 

 tagging studies. Southwest Fish. Cent. La Jolla Lab., Natl. 

 Mar. Fish. Serv., NOAA. Admin. Rep. LJ-79-17, 10 p. 



Table 4. — Analysis of covariance comparing growth rate of 

 Group A North Pacific albacore with growth rate of Group (B + 

 C) albacore, assuming stock boundary at lat. 40" N. 'Probability 

 of obtaming F statistic this large under null hypothesis is 

 • 0.005. 



Source of variation 



df 



Residual SS 



MS 



30.72 



I 



01 



UJ 

 3 

 O 

 UJ 



a: 

 u. 



>- 

 o 



z 



LiJ 



O 



UJ 



35 



30 



25 



20 



15 



10 



^ 



I FISH CAUGHT 

 SOUTH OF 38° NORTH 

 I (n = 49.920) 



I 



T>v^ 



45 50 55 60 65 70 75 80 85 90 95 



FORK LENGTH (cm) 



Figure .3. — Composite length-frequency distributions for 

 North Pacific albacore caught north of lat. 38° N and south of lat. 

 38° N off the U.S. west coast during the 1972-78 fishing seasons. 



2 to each set of data, using the sequential estima- 

 tion procedure (Equation (2)) with Ly-_ = 125 cm. 

 Resulting estimates of K were 3-6% larger than 

 the corresponding estimates from the standard 

 linear model; thus, if Model 2 is correct, system- 

 atic bias in the latter estimates does not appear 

 to be serious. 



However, estimates of a and /3 were relatively 

 large in all cases, suggesting that the growth rate 

 may drop abruptly to near zero immediately after 



299 



