3.2 Principal Factors of Stress and Riparian Health Condition
PCA is widely used for data matrix transformation and allows the
researcher to understand the relationship between indicators by
condensing the dimensionality of the data using factor analysis. The
Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s
test of sphericity were used to check data legitimacy and consequently
determine PCA suitability. Both tests verified that our data were
statistically valid for PCA. The KMO tests resulted in scores of 0.858
and 0.731 for RHIs and stress indicators, respectively, which were
higher than a score of 0.6, whereas Bartlett’s tests appeared as 0.000,
which was lower than 0.05. PCA results for RHIs and stress indicators
are presented in Figures 6, respectively. Based on the initial
eigenvalues, three components (holding 14 out of 27 indicators and
accounting for 65.24% of the total variation) were extracted from the
riparian zone health indicators, as shown in Figure 6A. These were
confirmed as sufficient and authentic for parallel analysis using two
additional steps—screen plot and Monte Carlo PCA. The first component
(F1) was mostly condensed for PC1, H1, PC6, PC3b, R1, Er1 and Er3a,
while four RHIs (Ex4, Ex1b, Ex3, and Ex2) and three RHIs (Er3b, Er3c,
and Er3a) were shown in the second component (F2) and third component
(F3). Following the same selection criteria, three components (holding 7
out of 13 indicators and accounting for 70.90 % of the total variation
within the TGDR) were extracted from the stress indicators, as shown in
Figure 6B. The rotated component matrix for stress indexing indicated
that P8b, P9, P8a, and P3a, were strongly linked, based on similarly
high values, in the first component (F1), and two stress indicators (P7
and P10) were shown in the second component (F2), with one indicator
(P8c) in third component F3.
[Figure 6 to be inserted about here]