f. Correlation Analyses . A correlation matrix was developed, using 

 the physical parameters and the major species. Twenty-nine variables were 

 used, producing 406 nonredundant correlations (Table 27). The correla- 

 tion coefficients from the comparison of major species with season, 

 transect, site, and horizontal distance from shore, generally agreed with 

 the results of the variance analysis. The correlation between 

 E. talpoida, Donax sp., A. virginiana, L. americanus , or M. sczelkowii 

 and temperature was significant, but no species showed a significant 

 correlation with salinity. Temperature and salinity changed gradually 

 with increased distance from shore, thus the correlations between a 

 species and temperature reflect the seasonal effect. 



Emerita talpoida had a strong negative correlation with mean grain 

 size, indicating an affinity for the coarse sediments of the upper swash 

 zone. Emerita talpoida correlated with sorting, skewness, and kurtosis 

 because of the partial correlation of these variables with mean grain 

 size. The burrowing amphipods, P. longimerus and B. quoddy ensis , had a 

 strong, positive correlation with mean grain size, indicating an affinity 

 for the fine sediments of the deep sites. 



A principal factor analysis with iteration and varimax rotation [Nei, 

 et al . , 1975) was performed on the correlation matrix. The extracted 

 factors, eigenvalues, and the percent of variance explained are pre- 

 sented in Table 28. Nine factors were extracted, but the first five, 

 interpreted below, explained over 80 percent of the variance: 



(a) Factor 1. This factor loaded heavily with month of 

 collection (loading 0.94165), indicating that the largest source of 

 variance was seasonal variation which accounted for about one-third of 

 the variance. 



(b) Factor 2 . This factor loaded heavily and positively with 

 site, horizontal distance from shore, mean grain size, and kurtosis. 



It loaded heavily and negatively with vertical distance from MSL and the 

 sorting, skewness, and carbonate content of the sediments. Since the 

 other variables were strongly correlated (positively or negatively) with 

 site, this factor depended on location on the transect, and accounted 

 for 20 percent of the variance. 



(c) Factor 5 . The factor loaded heavily with two selective 

 surface deposit- feeding polychaetes, Spiophanes hombyx and Megalona 

 vosea. These species occurred sporadically in the collections. Their 

 occurrence was very highly correlated, but they did not correlate 

 strongly with other variables. Scolelepis squamata, another selective 

 surface deposit-feeding polychaete, was abundant on most sites, but did 

 not fall into this factor. 



(d) Factor 4 . The factor loaded with the burrowing amphipod, 

 P. longimerus and with an omnivorous fast-burrowing polychaete, 

 Nephytes buaera. It is also loaded with a third variable, the kurtosis 

 of the sediments, and with a second burrowing, omnivorous polychaete. 



50 



