514 
Fishery Bulletin 95(3), 1 997 
Island, i.e. to those areas sampled during 1992. Two 
groups of variables explain much of the observed dis- 
tribution. These variables are substrate composition 
and a depth-temperature factor. The relative impor- 
tance of depth and temperature or of gravel, sand, 
and mud is difficult to assess because each group is 
Flathead sole 
temperature <8.9 deg.C > 8 9 deg.C 
in 
+ 
LU 
O 
a. 
o 
cn 
o 
■sf 
co 
CM 
O 
1 2 
3 4 
Node 
Figure 6 
Pruned regression tree for age-0 flathead sole with predictions of log(CPUE+l) at each terminal 
node, and number of stations in parentheses (top figure). Bottom figure contains box plots of 
log(CPUE+l) by terminal node. Each box plot summarizes distribution of CPUE at stations rep- 
resented by the terminal node directly above it. Box plots indicate median (filled circle), 
interquartile range (box height), and outliers (open circles). Whiskers indicate upper quartile 
plus 1.5 times interquartile range and lower quartile minus 1.5 times interquartile range. Width 
of boxes is proportional to square root of number of stations at that node. See text for sediment 
abbreviations. 
