Supplementary figures S1-S5 from Temperature is a poor proxy for synergistic climate forcing of plankton evolution

S1: Study species Globoconella puncticulata (a) and Truncorotalia crassaformis (b). The scale bars represent 100 μm. S2: Map of study site IODP Site U1313 in the North Atlantic. S3: Autocorrelation plots for traits of Globoconella puncticulata showing autocorrelation in the original time series (a-c) but not in the first differences (d-f). S4: Autocorrelation plots for traits of Truncorotalia crassaformis showing autocorrelation in the original time series (a-c) but not in the first differences (d-f ). Only area in T. crassaformis still contained a small amount of autocorrelation in the first differences time series. However, comparison of two Generalized Least Squares models with and without autocorrelation (by setting the continuous auto-correlation parameter phi=0.3 and phi=0.1, which converges to 0 under maximum likelihood, respectively) show lower AIC values for the model without autocorrelation, implying that the model with temporal autocorrelation is actually a poorer fit to the data than that without. S5: ΔAIC values showing the added relative importance of each environmental parameter or combination of parameters to the total model.