2. Sampling Effectiveness . 



Three sampling techniques are recommended: coring, trenching, and 

 suction dredging. The area sampled is a function of the number and size 

 of samples taken. The depth of the samples and the sieye mesh must be 

 consistent. To be comparable to the greatest number of other studies, 

 samples should be taken to a depth of 8 inches (20 centimeters) and sieved 

 through a 0.5 -millimeter sieve. If time and cost constraints require, the 

 depth of sampling may be reduced to 4 inches (10 centimeters) and the 

 sieve increased to 1.0-millimeter mesh rather than reduce the number of 

 sampling stations. 



3. Sampling Strategy . 



Number of Sample Replicates . Three commonly used parameters to de- 

 scribe the benthic infauna are: the number of species, the total number 

 of individuals, and the number of individuals per species. The number 

 of samples initially taken to determine the number of replicate samples 

 needed is always greater than the number that would be taken during the 

 field survey at any one station. Although this sampling need not be 

 strictly random, individual samples should be chosen at random from the 

 combined group for the determination of the number of replicates needed. 



(1) Number of Species . Species are recorded in two ways : the 

 total number of species collected in all samples and the average number 

 of species per sample. If a species cannot be identified in the field 

 or laboratory, assign the unknown species a code number and proceed with 

 the identification. 



The figures and table used in the following sections are from a report 

 by Oliver and Slattery (1976), as an example of one approach to developing 

 a sampling strategy. Figure 2 is an example of an accumulation rate of 

 new species as the area sampled is increased (Oliver and Slattery, 1976) . 

 The replicates are averages of three random samples of twenty-eight 0.022- 

 square yard (0.018 square meter) samples. Considering all 28 samples as 

 100 percent of the sampling area, 16 samples represent 57 percent of the 

 area and contain 92 percent of the species, 8 samples represent 30 per- 

 cent of the area and contain 76 percent of the species, and 4 samples 

 represent 14 percent of the area and contain 57 percent of the species. 



To examine the relationship between the number of replicates and the 

 number of species per sample, replicates should be drawn at random and 

 the means of confidence limits computed for progressively larger sample 

 sizes. The average number of species per sample (Table 3) changed little 

 with increasing sample size above about eight replicates. The confidence 

 limits will continue to narrow as the number of replicates increase. 

 The number of replicates taken will depend on the sampling objectives and 

 the risk involved with making the wrong decision. 



13 



