668 



Fishery Bulletin 97(3), 1999 



42" N 



4rN 



Fall 



Ln strata coefficients 



E3 -2.0 to -1.5 



 -1.5 to -1.0 

 □ -10 to 



 Oto 0.5 



420N 



4rN 



67 W 



66 W 



670W 66°W 



Figure 3 



The relative abundances between strata as characterized by the In strata coefficients from the multiplicative model for the 

 NMFS fall and spring surveys. Strata which have a circle ( ) around them, e.g. 17u , were not used in the multiplicative model 

 and missing observations were given a value of "0." Stratum 20 was also not used in the model because the 5Z stratum means 

 were used for missing observations. Strata depth zones are as follows: 19 and 20, 27-55 m; 16 and 21. 56-110 m; 17 and 22, 

 111-183 m; 18, >183 m. 



in several years but, in those years, the total abun- 

 dance on the U.S. side was usually small. For the Ca- 

 nadian side, in most instances, very little of the abun- 

 dance was estimated from the multiplicative model. 



An examination of the fall ratios of relative abun- 

 dance, (Fig. 4), revealed two distribution patterns in 

 haddock age 1 and older. Between 1963 and 1972, 

 haddock were highly variable in their relative abun- 

 dance and were found throughout the bank. After 

 1971, the majority of haddock ages 1 and older were 

 found on the Canadian side as indicated by the pre- 

 dominance of ratios >0.75. The ratios of relative abun- 

 dance indicated that age-0 haddock were very vari- 

 able throughout 1963-93 and were often more abun- 

 dant on the U.S. side. 



Friedman's nonparametric test revealed that the 

 ratios for ages l-9^- during the period 1972-93 were 

 similar but the ratios for age differed from these. 

 During the period 1963-71, the ratios for all ages, 

 0-9-t-, did not show persistent patterns (Table 4). 



We conclude then that the relative abundance of 

 age-0 haddock across the ICJ line is different from 

 that for ages 1-9-h but only during the period 1972- 

 93. Therefore, for further analysis of time trends, ages 



1-9-h were combined (Fig. 5). From 1963 to 1971 had- 

 dock of ages 1+ were distributed throughout 5Zj,m 

 but were generally found in greater numbers on the 

 U.S. side. Since 1972, however, very few haddock at 

 these ages have been found on the U.S. side. Between 

 1963 and 1971, there was one anomolously high 

 value, 1968 with a ratio of 0.99, but all other ratios 

 were lower than any of the values after 1971. In con- 

 trast to ages 1-I-, the ratios of relative abimdance of 

 age-0 haddock were very variable and indicated a 

 less patterned distribution, being distributed 

 throughout 5Zj,m. All except one of the stronger year 

 classes, 1972, 1975, 1978, 1983, 198.5, 1987, and 1992, 

 with relative abundance ratios at age of 0.34, 0.28, 

 0.47, 0.62, 0.40, 0.21, and 0.08, respectively, were 

 more abundant on the U.S. side. Interestingly, the 

 weak year classes that followed these ( 1973, 1976, 

 1979, 1984, 1986. 1988, and 1993) were found in 

 greater abundance on the Canadian side with ratios 

 of 0.85, 0.85, 0.64, 1.0, 1.0, 1.0, and 0.93, respectively 

 A likely mechanism has not been determined. 



The distribution of age-l-f- haddock determined by 

 using the catch-per-tow data from the NMFS fall sur- 

 vey, pooled over 1985-95, are shown in Figure 6. As 



