Koontz, J., D.G. Huggins, C.C. Freeman, D.S. Baker - Central Plains ...

a priori Groups and Metric Selection
The stressor-response metrics were selected using a Pearson correlation matrix (i.e. Pearson
product-moment correlation coefficient) and linear regression test, except no single reference or
random group was established a priori. In this study, a priori „high‟ and „low‟ groups were
established for parameters that showed consistent significant responses to multiple
macroinvertebrate metrics using the 25 th and 75 th percentile, since significant variability in
response existed among landscape, plant community, and water quality measures.
Macroinvertebrate metrics were placed in a correlation matrix along with floristic quality
measures, water quality parameters, and surrogate spatial and temporal variables. All significant
(p ≤ 0.05) Pearson correlations were tested with linear regression and retained if still significant.
Relationships were commonly found between various multiple macroinvertebrate metrics and
one water quality measure, floristic quality metric, or other variable. Groups were created as
„least disturbed‟ or „degraded‟ condition with samples having parameter values equal to and
lower or higher than the 25 th or 75 th percentile value, respectively. The macroinvertebrate
metrics that were significantly related to the other environmental parameters through linear
regression analyses were assessed using the two-sample t-test method described by Stoddard et
al. (2008), resulting in 39 macroinvertebrate metric responses to 11 groups, with two groups
eliminated in this process. Many metrics also responded to various groups in the t-test analysis,
thus it was necessary to define each metric by its greatest t-score, further eliminating many a
priori groups.
Twenty-six metrics were retained, the greatest numbers of which were found in the Number of
Herbicides Detected group, Native Plant Richness group, and Maximum Depth group, with a
small representation of other groups having metrics with significant t-scores. Five
macroinvertebrate metrics having the lower t-score between high and low a priori groupings
were eliminated due to redundancy (Pearson R > 0.70) with another macroinvertebrate metric.
Only the Native Plant Richness, Number of Herbicides Detected, and Maximum Depth groups
were further evaluated because they had the greatest response from macroinvertebrate metrics
when metrics also responded to other parameters and groups. These three groups represented
hydrological and floristic wetland qualities as well as anthropogenic disturbance. The remaining
21 metrics were two sample t-tested in these groups.
T-test values remained significant for three metrics in the native plant richness group: Shannon‟s
Diversity Index (+), Percent Burrowers (-), and Count Intolerant Taxa to Suspended Solids and
Sediments (SSS) (+) (Table 12). Four completely different metrics in the maximum depth „high‟
and „low‟ groups were found to be significant in t-test scores: Percent Hydroptilidae (+), Count
ETO taxa (+), Percent Sprawler Taxa (+), and Percent Intolerant based on mean Regional
Tolerance Values (+). The metrics having significant t-test scores between the low and high
Number of Herbicides Detected group were Percent Non-Insect Taxa (-), Percent Burrowers (-),
Intolerant Taxa to Heavy Metals (+), and Count Intolerant Taxa to Suspended Solids and
Sediments (+). These metrics were not significantly (p