1985, with colored electrical tape on uncolored cable ties). Linear growth was determined by measurements 

 made from the top of the most recently formed vesicle to the apex of the developing axis, or apices if 

 branching had occurred. Vesicles had not developed sufficiently to be tagged in April or May, so five tips 

 were measured on each of 50 randomly chosen Ascophyllum plants. Monthly measurements of tagged 

 plants began in .lune. Lost tags were not replaced, and the pattern of loss was used as a measure of both 

 mortality and environmental stress. I^ss of the entire plant was assumed when the base tag and tip tags 

 were missing. Tip survival was determined in terms of remaining tip tags. 



Temperature 



Water temperatures were derived from the EDAN (Environmental Data Acquisition Network) system, 

 which continuously records a variety of environmental parameters and reports at 15-minute intervals. 

 Ambient water temperatures were recorded by sensors in Unit 1 and 2 intake bays, and effluent water 

 temperatures by sensors in the quarry cuts (Eig. 3). Temperatures at EE and the experimental Ascophyllum 

 stations (EL and EN, I'ig. 3) were interpolated, based on measured AT, and verified over several tidal 

 cycles with a portable thermistor and strip chart recorder. 



Data Analysis 



Relative abundance of intcrtidal organisms was estimated on the basis of percentage of substratum 

 covered by each taxon. Unoccupied substrata were classed as free space. Similarity between communities 

 was determined by a percent standardi/.ed form of the Bray-Curtis coefficient (Sanders 1960), calculated as: 



where P(ii) is the percent of species (i) at station (j), P(ik) is the percent for station (k), and (n) is the 

 number of species in common. A flexible-sorting, clustering algorithm was applied to the resulting 

 similarity matrix. The calculations were performed on untransformcd percentages. 



Ascnphvllum growth data are reported as mean growth ± 2 standard errors; means are compared using 

 2-sample t-tcsts (F'ROC TTEST, SAS Institute Inc. 1982). A probability level of n = 0.05 was used to 



