that was more than double the estimated loss caused 

 by 6 min of pea digger use (Table 3). Both 

 aboveground and belowground components of the 

 seagrass demonstrated this statistically significant 

 difference between rakes. The bull rake also pro- 

 duced a greater estimated loss of seagrass biomass 

 per unit area raked, an effect that was also significant 

 for both aboveground and belowground components 

 (Table 3). An estimated 87% of the initially present 

 seagrass dry weight was removed by the bull rake in a 

 1 m 2 area that was completely raked. The magnitude 

 of this effect was similar for components both above 

 (89%) and below (83%) ground. In contrast, the pea 

 digger removed only an estimated 47% of seagrass 

 dry weight per unit area completely raked, with the 

 impact falling less heavily on roots and rhizomes 

 (37% decline) than on shoots (55% decline). The two 

 rakes did not differ significantly in estimated sea- 

 grass biomass removed per legal-sized clam collect- 

 ed, although the estimated loss of belowground dry 

 weight per clam collected by the bull rake was almost 

 double the estimated loss caused by the pea digger 

 (Table 3). 



Discussion 



By use of replicated field trials, we compared the ef- 

 fectiveness of two clam rakes in two contrasting ways. 

 We estimated in each of two habitats the rate of hard 

 clam capture per unit time, as would be appropriate if 

 harvest time were limiting. We also converted our 

 data into estimates of harvest per unit area raked, as 

 would be appropriate if suitable clamming habitat — 

 rather than time — were limited. We view these 

 measures as endpoints in a spectrum of possibilities 

 with the first more appropriate for managers of clam 



TABLE 2. — Hard clam capture rate per unit time, per unit area raked, 

 and capture efficiency of two clam rakes from six paired replicate 

 plots in a seagrass bed. F-tests revealed no significant difference 

 between treatments in variance, except for area raked which re- 

 quired a log transformation prior to performing the <-test. 



Seagrass bed 



resources that are abundant relative to the intensity 

 of harvest, and the second more relevant to clam 

 resources subjected to very intense harvest pressure. 

 By examining both endpoints, we hope to bracket 

 actual prevailing conditions. 



Our harvest data imply that habitat strongly in- 

 fluences the relative effectiveness of these two clam 

 rakes. In unvegetated sandy sediments, the pea dig- 

 ger captured significantly more legal-sized hard 

 clams per unit time than the bull rake (Table 1). In a 

 seagrass bed, the relative effectiveness was reversed 

 (Table 2). The difference between rake effectiveness 

 was not a consequence of greatly differing efficien- 

 cies of clam capture within raked areas, but rather of 

 differing rates of areal coverage. Because of approx- 

 imately equal efficiencies of clam capture, the rakes 

 did not differ significantly in hard clam capture per 

 unit area raked in either habitat. 



We suspect that the pea digger's advantage in un- 

 vegetated sandy sediments was dependent upon two 

 confounded factors: 1) The relatively low densities of 

 both living and dead hard clams, and 2) the absence 

 of living seagrass. In areas with low hard clam den- 

 sities, the pea digger will glide over unproductive bot- 

 tom without creating frequent contacts that require 

 excavation. Thus, more area can be covered than with 

 a bull rake, which must be pulled more deeply 

 through the sediments regardless of the scarcity of 

 clams. Entanglements with roots of living seagrass- 

 es may tend to slow the progress of the pea digger 

 which must plow through mats of seagrass, whereas 

 the greater inertia of the moving bull rake is less in- 

 fluenced by encountering a small obstacle. Because 

 these two factors (clam abundance and sea-grasses 



TABLE 3. — Comparison of environmental impacts on seagrass of two 

 different clam rakes used in seven paired replicate plots. F-tests 

 revealed no significant difference between treatments in variance 

 for any comparison. 



Seagrass bed 



Average ±1 SD 



Estimated impact 



Bull rake 



Pea digger Mest 1 





2 >2.54 cm thick. 



3 Back-transformed mean of arcsin-transformed percents of clams captured. 



4 n = 5 for this comparison, because one plot had no illegal-sized clams. 



'* = P<0 05; ** = P<0 01 ; ns = P>0.05 in a two-tailed paired Mest. 

 2 Average seagrass dry weight (g/m 2 ±1 SD) in the 7 control (1 m 2 ) plots: shoots  

 67.7 (±19.8); roots- 48.2 (±15.7); total - 1 1 6.0 (±32.4). 



432 



