Table 2. Summary of training and testing performance of Maxent models based AUC metrics, accuracy based on maximum test sensitivity and specificity decision boundary (logistic threshold), and action values with 95% confidence intervals.

If an upper bound of a confidence interval exceeds the maximum sampling value for a set of data, the maximum value is given.

Variables	Training AUC (± SE)	Testing AUC (±SE)	Accuracy	Action values (x) ¥ (95% CI)
Alkalinity $\left(\frac{mg}{L}\right)$	0.616 (0.003)	0.620 (0.006)	68.5	$x>129\frac{mg}{L}$ (125, 134)
BOD $\left(\frac{mg}{L}\right)$	0.572 (0.004)	0.554 (0.008)	60.6	$x<0.976\frac{mg}{L}$ (0.825,1.09) $x>6.19\frac{mg}{L}$ (4.51, 6.43)
Conductivity (µS)	0.628 (0.003)	0.638 (0.006)	65.6	x > 306 µS (287,315)
Dissolved Oxygen $\left(\frac{mg}{L}\right)$	0.635 (0.003)	0.640 (0.007)	67.7	$x<9.39\frac{mg}{L}$ (8.68, 10.6)
Discharge ( $\frac{m^3}{s}$)	0.556 (0.004)	0.553 (0.006)	63.8	*
Hardness $\left(\frac{mg}{L}\right)$	0.632 (0.003)	0.627 (0.006)	59.9	$x>132\frac{mg}{L}$ (122, 152)
NO3  $\left(\frac{mg}{L}\right)$	0.581 (0.004)	0.579 (0.007)	63.4	$x>1.78\frac{mg}{L}$ (1.63, 1.84)
pH	0.571 (0.003)	0.562 (0.006)	55.6	*
PO4 ( $\frac{mg}{L}$)	0.581 (0.004)	0.580 (0.008)	63.8	0.0642 $\frac{mg}{L}<x<7.80\frac{mg}{L}$ (0.0873, 0.766)˘ (6.27,9.01) ̑
Water Temperature (°C)	0.666 (0.003)	0.670 (0.005)	65.2	x > 12.4 °C (11.3, 15.5<x<20.0)
8-variable model	0.770 (0.002)	0.709 (0.005)	78.5
5-variable model	0.753 (0.002)	0.723 (0.006)	77.8
4 variable model	0.750 (0.002)	0.726 (0.005)	77.8

Notes.

^*Model was not significant.

^aValues of the variables that corresponded to impairment.

^b95% CI for the lower bound of the action value.

^c95% CI for the upper bound of the action value.