a. Precision of Estimates . The number of samples from intertidal, 

 3.7-meter, and 6.1-meter sampling stations needed to estimate population 

 values for abundance and species richness at 50- and 30-percent levels 



of precision (P%) at a 95-percent level of confidence was calculated 

 (Table 7) . These are the average number of samples needed to know that 

 95 percent of the time the true population means will lie within +P% of 

 the measured mean. These estimates from the sample data are derived 

 from properties of variance in the data and are corrected for nonnormal 

 distribution properties. Most of the contagiously distributed data sets 

 fit a negative binomial distribution. Types of distributions of sampled 

 data are presented in Table 8. 



In sampling design considerations, it was expected that a 

 greater heterogeneity of factors would influence distribution on the 

 beach more than offshore; therefore, 90 samples per survey were taken 

 at each intertidal station and subtidal assessments were based on 15 

 samples at each of two depths. Subsequent analysis indicated the 

 adequacy of this approach. Precision of estimates of species and 

 abundance population parameters were approximately equal from the 

 different depths using these discrepant sample numbers (Table 9] . The 

 number of samples taken at all depths was (on the average) sufficient 

 to estimate these population parameters at a +^30-percent precision 

 level. A 50-percent precision level may be reached by taking approxi- 

 mately half this number of samples within each depth habitat. This 

 fact is a useful one in relating level of information return to the 

 inherent cost factors on a given project. For example, in the present 

 study a 20-percent loss of precision would accompany a 50-percent 

 reduction in sample analysis time (Table 7) . 



b. Small-Scale (Within-Station) Variation on the Beach and 

 Comparison of Intertidal Transect Methods . At each intertidal station 

 no significant differences in variability or precision of estimate were 

 found between transects sampled either randomly or at fixed intervals 

 within strata (Fig, 2). An advantage of the line-point method is the 

 possibility of constructing regressions of variables with fixed positions 

 along the transect (see Fig. 10) and relating these to observed profile 

 features along the beach. 



Considering only a single transect along the beach, abundance 

 and species richness are estimated at respective precision levels of 

 about 27 and 45 percent. Grouping of the three transects at each inter- 

 tidal station increased precision by decreasing these estimates to 17 

 and 27 percent, respectively. Here again, cost optimization factors may 

 be considered in relation to precision criteria and sampling design. 



Single transects had mean values that deviated an average of 

 20.6 percent (median value = 13 percent) from the average of three tran- 

 sects combined; i.e. assuming that three transects located 50 meters 

 apart represent "true" population densities on the beach, then any 



48 



