Geoscience Reference
In-Depth Information
Table 10.3 Proportional land cover of each of the Tribs watersheds as determined by classification
of Landsat TM imagery, May 1992
polygon, which was necessary for further calculations. The proportion of each land
cover type within each buffer was calculated using Microsoft Excel. The result
was a condensed table with the proportion of each land cover type corresponding
to each sample point. These data were then joined to the original IBI and QHEI
data in ArcGIS thus assigning an IBI and QHEI score to the proportional land
cover composition of each buffered area. These data were imported into SPSS for
statistical analysis. After logarithmically transforming all data, multiple regression
analyses were conducted to assess the influence of the proportion of each land cover
type on IBI scores and QHEI scores.
Nine multiple regression models were run using SPSS 14.0. First, each depen-
dent variable (IBI and QHEI) were analyzed against the land cover data within
the 100 m buffer, then the same was analyzed for the 500 m buffer. The datasets
were divided and analyzed separately based on the water body type (riverine or
estuarine) of the IBI/QHEI sample locations. QHEI and IBI for the 100 and 500 m
buffer were analyzed for the divided dataset. Finally, a bivariate linear regression
analysis was conducted to determine possible relationships between IBI scores and
QHEI scores. All models used the stepwise regression method to eliminate covari-
ant variables and include only those statistically significant (
0.05).
The results of the multiple regression showed that within the 500 m buffer, agri-
culture, wet vegetation, bare ground and grass together explain 58.4% (adjusted
R
2
ΒΌ
a
0.05.) Agriculture alone explained
16.2% of variance. Water/wet vegetation explained 13.3%, bare ground 17.7%
0.584) of the variance in IBI scores (p
<
