Figure 5

Predicted changes in abundances in response to altered temperature. For each environmental factor, the maximum observed variation was used to modify the measured environmental factor over a 30-day period. The change was initiated at the beginning of each calendar month. Panel A: Examples of predicted abundance changes (red/green lines) over time for each SC were simulated with either the original water temperature (red) or a water temperature (green) that was modified during a 30-day period (grey bar). Observed values are shown as blue dots. The abscissa shows the days of a two-year period starting from 7/1/2010. The ordinate shows the relative abundances (percentages) of SC-1, SC-2 and SC-3. The maximum and minimum (grey) of the predicted values are shown. Panel B: Results were grouped by the month during which water temperature was modified. The water temperature was either increased (blue bars) or decreased (yellow bars) for 30 days corresponding to calendar months, and a modified LSQ was computed using the mean abundance of 10 simulations using 10 randomly chosen sets from among 84 estimated models. The results suggest that a 30-day change in water temperature between January and May has a stronger effect on SC abundances. This result was validated with observed data (p = 0.03). Panel C: Results were grouped by SCs. For each of the 14 SCs (abscissa), the modified LSQ was computed when the water temperature was either increased (blue bars) or decreased (yellow bars) for 30 days. The change in water temperature has a greater effect on some SCs (1, 2, 9, 11, 12) than on others. This observation was confirmed with observed data (p = 0.0046).