upon the scale of the classification system as well as the classification system utilized (see Table C.l). 
The estuaries sampled in the Classification Study and NCA data set fell into a limited number of 
estuary classes (2-4); however, there was not a consistent pattern in the grouping of estuaries within a 
class among the different classifications. One exception was the Columbia River Estuary, which 
consistently was placed in a separate class for classifications based on geomorphology, susceptibility 
to nutrient pollution, and statistical clustering. 
Cumulative distribution functions (CDFs) were produced for each of the three data sets 
(Yaquina, Classification, and NCA). For this analysis, the data from the three data sets were divided 
into marine and riverine dominated regions (Zones 1 and 2, respectively). In addition, the Yaquina 
data set was further divided into wet and dry seasons. Only recent data (1998-2006) were used in 
creating the CDFs for Yaquina Estuary. The CDFs produced for the Yaquina and Classification data 
sets represent percentiles associated with the number of samples (i.e., not weighted by percentage of 
estuarine area). The NCA program typically computes CDFs using the appropriate sampling area 
weightings, which are based on areas of sampling strata determined from GIS (US EPA, 2004a). This 
allows estimation of the areal extent of Oregon’s estuaries associated with any value of an indicator 
variable. However, for the present study, estimates of percentiles for NCA data sorted by salinity zone 
were produced without use of area weightings and represent percentiles associated with the number of 
samples. This was done for consistency among data sets, and because area estimates of salinity zones 
were not available for all Oregon estuaries. An additional set of CDFs were produced for the NCA 
data set excluding the Columbia Estuary, which differs from the other Oregon estuaries in size, 
geomorphology, and other factors. 
4.5 Statistical Analysis 
Due to the non-normal distribution of the data, non-parametric statistical tests were used for all 
analyses. The Mann-Whitney Rank Sum test was used to determine whether there were significant 
differences in median values between zones or seasons. The Kruskal-Wallis one way analysis of 
variance on ranks was used to test whether there were significant differences in the median values 
between the Yaquina (dry season only), Classification Study, and NCA data sets. If there were 
significant differences (p<0.05), then Dunn’s test was used for pairwise multiple comparisons. For all 
tests, p values less than 0.05 were considered significant. 
32 
