Widespread loss of safe lake ice access in response to a warming climate

S3 Fig. Unsafe transition period empirical cumulative distribution functions. The empirical cumulative distribution functions (ECDF) of each combination of warming scenarios and ice quality scenarios. Represented are (a) the ECDFs of each warming scenario (colored lines for 1˚C, 2˚C, and 4˚C) with ice quality held constant. The left columns of both (a) and (b) show the ice formation transition period (i.e., “on”) and the right columns show the ice melt transi- tion period (i.e., “off”). (TIF) S4 Fig. Unsafe transition period empirical cumulative distribution functions. The same as in S3 Fig but represented are the ECDF of each ice quality scenario (100% black ice, 50% black ice– 50% white ice, and 100% white ice) with warming held constant. (TIF) S1 Table. Shapiro-Wilks result table. The results of the Shapiro-Wilks test for each ice quality and warming scenario. (PDF) S2 Table. Kruskal-Wallis results table comparing quality scenarios. The results of the Krus- kal-Wallis rank sum test, which compared the ice formation and melt transition period anom- alies by the lake ice quality scenario (i.e., 100% black ice, 50%black/white ice, 100% white ice). (PDF) S3 Table. Kruskal-Wallis results table comparing warming scenarios. The results of the Kruskal-Wallis rank sum test, which compared the ice formation and melt transition period anomalies by the warming scenario (i.e., 1˚C, 2˚C, 4˚C). (PDF) S4 Table. Comparing across ice quality and warming scenarios with Dunn’s test. The results from the Dunn’s test of multiple comparison using rank sums within each ice group

(black, white, etc.) across warming scenarios (i.e., 1˚C, 2˚C, 4˚C). The adjusted p value is pro- vided, using the Holm method for detecting the family-wise error rate. (PDF) S5 Table. Comparing across warming scenarios and ice quality with Dunn’s test. The results from the Dunn’s test of multiple comparison using rank sums within each warming scenarios (i.e., 1˚C, 2˚C, 4˚C) across ice group (black, white, etc.). The adjusted p value is pro- vided, using the Holm method for detecting the family-wise error rate. (PDF) S6 Table. Comparing warming scenarios across ice quality categories. The results from the asymptotic two-sample Kolmogorov-Smirnov test. The compared samples are within the same ice quality scenarios (i.e., 100% black ice, 100% white ice, and 50% white ice) and across warm- ing scenarios (i.e., 1˚C, 2˚C, 4˚C). (PDF) S7 Table. Comparing ice quality across warming categories. The results from the asymptotic two-sample Kolmogorov-Smirnov test. The compared samples are within the same warming scenario (i.e., 1˚C, 2˚C, 4˚C) and across ice quality scenarios (i.e., 100% black ice, 100% white ice, and 50% white ice). (PDF) PLOS ONE Changing ice conditions inhibit safe use PLOS ONE | https://doi.org/10.1371/journal.pone.0313994 December 11, 2024 15 / 19