Supplemental Material Title Measuring the Contribution of Agricultural Conservation Practices to Observed Trends and Recent Condition in Water Quality Indicators in Ohio, USA Author Robert J. Miltner Ohio Environmental Protection Agency, 4675 Homer-Ohio Lane, Groveport, Ohio 43125. [email protected](614) 836-8796 12 Pages 3 Tables 6 Figures
13
Embed
Supplemental Tables and Figures - … · Supplemental Material Title Measuring the Contribution of Agricultural Conservation Practices to Observed Trends and Recent Condition in Water
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Supplemental Material
Title
Measuring the Contribution of Agricultural Conservation Practices to Observed Trends and
Recent Condition in Water Quality Indicators in Ohio, USA
Supplemental Table S2. Summaries of water quality indicators by decade and ecoregion for Ohio headwaters included in the present study. Values for water chemistry parameters are geometric means, in mg L-1, plus one standard deviation. Values for the biological and habitat measures are arithmetic averages, plus one standard deviation.
Water Quality Measures
Biological and Habitat Measures
Decade NH3 TKN NOx TP TDS TSS N IBI QHEI Substrate N
Supplemental Table S3. Comparison of models evaluating variance covariate terms in models fit by generalized least squares (gls), and linear mixed models that include the covariance term suggested by the gls results. Fixed effects in the mixed models are ecoregion, decade, and decade nested within ecoregion, and the random effect is station (sampling location). Generalized Least Squares Models df AIC Linear Mixed Models df AIC NH3~1 random=~1|station, weights=fixed 3 3513 NH3~decade 4 4452 NH3~ecoregion, random=~1|station, weights=fixed 7 3389 NH3~decade, fixed variance covariate 4 3319 NH3~decade, random=~1|station, weights=fixed 5 3291 NH3~decade, constant variance covariate 6 3980 NH3~decade/ecoregion, random=~1|station, weights=fixed 17 3093
Supplemental Table S3. Continued - comparison of linear mixed models for the IBI, QHEI and substrate metric . Fixed effects in the mixed models are ecoregion, decade, and decade nested within ecoregion, and the random effect is station (sampling location). Linear Mixed Models df AIC IBI~1 random=~1|station 3 13627 IBI ~ecoregion, random=~1|station 7 13460 IBI ~decade, random=~1|station 5 13540 IBI ~decade/ecoregion, random=~1|station 17 13363 QHEI~1 random=~1|station 3 15021 QHEI ~ecoregion, random=~1|station 7 14784 QHEI ~decade, random=~1|station 5 15020 QHEI ~decade/ecoregion, random=~1|station 17 14794 Substrate~1 random=~1|station 3 10776 Substrate ~ecoregion, random=~1|station 7 10565 Substrate ~decade, random=~1|station 5 10779 Substrate ~decade/ecoregion, random=~1|station 17 10576
S7
Figure S1. Partial dependence plots of TKN concentrations fitted to individual predictors in the BRT model. The fitted line is locally weighted smoother that follows the central tendency of the fitted values over the range of predictor values. Fitted values were centered (i.e., the zero on the y-axis) by subtracting the mean of the fitted value (over each respective predictor).
S8
Figure S2. Partial dependence plots of TP concentrations fitted to individual predictors in the BRT model. The fitted line is locally weighted smoother that follows the central tendency of the fitted values over the range of predictor values. Fitted values were centered (i.e., the zero on the y-axis) by subtracting the mean of the fitted value (over each respective predictor).
S9
Figure S3. Partial dependence plots of NOx concentrations fitted to individual predictors in the BRT model. The fitted line is locally weighted smoother that follows the central tendency of the fitted values over the range of predictor values. Fitted values were centered (i.e., the zero on the y-axis) by subtracting the mean of the fitted value (over each respective predictor).
S10
Figure S4. Partial dependence plots of TSS concentrations fitted to individual predictors in the BRT model. The fitted line is locally weighted smoother that follows the central tendency of the fitted values over the range of predictor values. Fitted values were centered (i.e., the zero on the y-axis) by subtracting the mean of the fitted value (over each respective predictor).
S11
Figure S5. Partial dependence plots of substrate metric scores fitted to individual predictors in the BRT model. The fitted line is locally weighted smoother that follows the central tendency of the fitted values over the range of predictor values. Fitted values were centered (i.e., the zero on the y-axis) by subtracting the mean of the fitted value (over each respective predictor).
S12
Figure S6. Plots of fitted surface responses of a) TP, b) TSS and c) TKN in relation to the QHEI and percent agricultural land use from BRT models. Z-axis units are log 10 of ug/l.