PETERSON ET AL.: IMPACT OF MECHANICAL CLAM HARVESTING 



35' 



34''45 



40-- 



75° 40 



30' 



25' 



Figure 1.— The locations of the study sites in eastern North CaroHna, near Cape Lookout. BSS indicates the sand-fiat and BSG the 

 seagrass-bed locations. Tick marks on the margins of the figure denote minutes of N. latitude and W. longitude. 



Sound is a shallow marine lagoon with a lunar tide 

 of about 0.6 m range, little salinity variation (28- 

 34''/oo), and a wide seasonal temperature range 

 from a winter monthly minimum of 2°-4°C to a 

 summer monthly maximum of 29°-30°C. In Jan- 

 uary 1980, we selected in each habitat 6 square plots 

 (matrices) of 1,225 m^ area, each of which had a 

 virtually constant water depth of about 0.1-0.3 m 

 at low tide and homogeneous surface appearance. 

 Specifically, all seagrass matrices held a spatially 

 uniform cover of a seasonally varying mixture of two 

 seagrasses, eelgrass Zostera marina and shoalgrass 

 Halodule wrightii, whereas no sand-flat matrix con- 

 tained seagrasses. These seagrass matrices had been 

 continuously vegetated from at least 1974 until 1980 

 and the seagrass cover had not extended over the 

 sand flat during that same period (Peterson et al. 

 1984). 



Before harvest treatment, we subsampled all 6 

 matrices in each habitat to test whether there were 

 any initial differences among matrices in response 

 variables. This sampling occurred between 22 Feb- 

 ruary and 31 March 1980 in the sand flat and from 

 1 April to 6 May 1980 in the seagrass bed. A fixed 

 number (9 or 36) of uniformly distributed 0.25 m^ 

 subsamples was taken from each matrix to estimate 

 abundance of hard clams, bay scallops, and seagrass 

 (Table 1). A uniform sampling array was chosen to 

 reduce the field effort and to avoid risk of sampling 

 at or even near (<1 m) the same locations during 

 subsequent sampling. A grid of marked ropes at- 

 tached to equally spaced stakes was placed around 

 the circumference of each matrix and moved to a 

 new, randomly chosen set of positions for each new 

 sampling date, thus producing a "frame shift" of 

 the sampling template. 



283 



