670 



Fishery Bulletin 97(3), 1999 



variation in catch per tow 

 (multiple i?2=o.41). The stra- 

 tum influence accounted for 

 most of the variation with an 

 F-value of 8.389 and P<0.001 

 but the year effect was not as 

 strong for the spring survey as 

 it was for fall (F-value of 1.647, 

 0.1<P<0.2). The frequency dis- 

 tribution of residuals approxi- 

 mated a normal distribution, 

 although not as well as those 

 of the fall survey. An outlier 

 from stratum 21u in 1972 in- 

 fluenced the model's estimates 

 of abimdances for that stratum; 

 estimates were 60% greater 

 than they would have been 

 had this value been excluded 

 from the analysis. A higher 

 proportion of haddock on the 



Canadian side would have resulted if this value had 

 been excluded. This stratum, however, does not con- 

 tribute substantially to the overall abundance; there- 

 fore its impact was not studied further. The relative 

 abundances between strata as characterized by the 

 strata coefficients from the multiplicative model are 

 shown in Figure 3. Two Canadian strata sections, 

 21c and 16c, which have a depth range from 56 to 

 1 10 m, had the highest coefficients. Three strata sec- 

 tions with a high number of zero values, 17u, 18c 

 and 18u, were not used in the model. 



The stratum and stratum section abundances from 

 NMFS spring surveys are given in Table 2. Stratum 

 section 16c, and 16u (secondarily), were again the 

 highest contributors to total abundance. The rela- 

 tive contribution of stratum section 21c increased 

 after 1976, whereas the contribution from stratum 

 section 17c remained stable over time. As in fall. 19u 



1965 1970 



1975 



1980 



1985 



1990 



Figure 5 



The distribution of haddock across the ICJ line in 5Zjm during the fall as indicated by 

 the ratios of relative abundance from the NMFS fall survey. 



in spring also declined in relative abundance from 

 earlier times; very few haddock were caught there 

 after 1981. The highest numbers in 19u were caught 

 during 1980 and 1981. Haddock were virtually ab- 

 sent from 18c in spring but in the fall there were 

 almost always some to be found there. Total abun- 

 dance on the Canadian side was augmented by esti- 

 mated numbers only four times, whereas most years 

 were augmented for the U.S. side. The contribution 

 made by these values to abundance on the U.S. side, 

 however, was negligible. 



Abundances by stratum for the DFO spring sur- 

 vey are given in Table 5. Haddock were always more 

 abundant on the Canadian side than on the U.S. side; 

 the Canadian side had 61-99% of the abundance in 

 5Zjm. Except for two occasions, haddock were more 

 abundant in the shallower 5Z2 stratum than in the 

 deeper 5Z1 stratum. Stratum 5Z2 and 5Z4 have the 



