PETERSON ET AL.: IMPACT OF MECHANICAL CLAM HARVESTING 



I40n 



120 



e 100 



g 80 



O 



CD 



o 

 Q. 



60 



40 



20- 



— Light Kicking 



— Raking 



■- Intense Kicking I 



— Intense Kicking n 



/ 



/ 



Sp'l980t FI980i Spl98l FI98I 



FI982 



FI983 



FI984 



minor mojor 

 treatment 



Figure 4.— Percent difference between observed average biomass of seagrass in each treat- 

 ment matrix and expected biomass based on the assumption that initial differences between 

 the two control matrices and each treatment matrix would be expected to remain constant 

 across time. The expected biomass is then plotted as 100% (the no effect line). Times of the 

 two clam harvest treatments are indicated with arrows on the x-axis. 



Benthic Macroinvertebrates 



In the sand-flat habitat, the average density of 

 benthic macroinvertebrates never varied significant- 

 ly among matrices (Table 6) in any of the 3 post- 

 treatment sampling dates [one-way ANOVA's were 

 run on log (x -i- l)-transformed counts, using a 

 separate analysis for each date]. The sums over all 

 3 posttreatment dates of the average macroinverte- 

 brate densities per core are nearly identical for each 

 sand-flat matrix and a two-way ANOVA on log 

 (x + l)-transformed densities from all 3 time peri- 

 ods revealed no significant difference among ma- 

 trices. 



In the seagrass habitat, analogous one-way 

 ANOVA's done separately for each date, demon- 

 strated that the average density of benthic macro- 

 invertebrates did not differ significantly among 

 seagrass matrices in fall 1980 or spring 1981 (Table 

 6). A significant difference among matrices did ap- 

 pear in fall 1981, and in a two-way ANOVA on all 

 3 posttreatment dates together. Despite the statis- 

 tical significance of 2 of 4 ANOVA's, actual differ- 

 ences in mean densities among seagrass matrices 

 were proportionately small. Furthermore, Duncan's 

 tests revealed a pattern of differences among ma- 

 trices (Table 6) that was identical to the initial pat- 



tern of significant differences in the spring 1980 

 sampling before treatment (see Table 3). 



Although the sums of the sample means from each 

 of the 3 posttreatment sampling dates (Table 6) im- 

 ply that benthic macroinvertebrate densities in the 

 seagrass habitat were about double those in the sand 

 flat, this pattern was not consistent across seasons. 

 Nested ANOVA's, done on log {x + 1)- transformed 

 counts and performed separately for each sampling 

 date, showed that there was no significant differ- 

 ence between habitats during either spring sampling 

 period (spring 1980 or 1981), whereas average den- 

 sities of benthic macroinvertebrates were signifi- 

 cantly greater (P < 0.001 in fall 1980 and P < 0.005 

 in fall 1981) in the seagrass habitat in both of the 

 Octobers. 



Although the clam harvesting treatments did not 

 affect total density of benthic macroinvertebrates 

 in either habitat, species composition might still have 

 been altered. We identified all individuals in 16 cores 

 in each habitat from the spring 1980 pretreatment 

 sampling (4 cores randomly chosen from each con- 

 trol matrix and from each intense-kicking matrix) 

 and in 16 cores in each habitat from the spring 1981 

 posttreatment sampling (drawn equally from each 

 of the same matrices). This comparison holds season 

 constant and permits us to test for any gross shifts 



291 



