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

3.3 Spatial variation Transition periods show distinct regional variability throughout the Northern Hemisphere but overwhelmingly shift toward longer unsafe ice transition periods. Unsafe transition periods on average increased (Table 1; Fig 3) and transition periods tended to increase the most along coastal regions in North America as well as southern latitudes in North America, Europe, and Asia. For example, transition periods exceeding 10 days composed approximately 30% of grids (n = 661) for white ice conditions at 4˚C. On the other hand, grids shifting to longer safe ice periods consisted of less than 1% of grids (n = 14). More grids shift toward longer safe ice peri- ods at 1˚C under black ice conditions (18%, n = 496). These grids are primarily located in northern latitudes of North America, Europe, and Siberia as well as some grids along the west- ern coast of Greenland. 3.4 Distribution patterns While the transition period of both ice formation and ice melt was similar for 100% black ice, with increasing ratios of white ice within the ice column, the mean transition period became longer for the ice formation period than for the ice breakup period. The distributions of the transition period anomalies became more exaggerated and skewed when both warming and ice quality interacted, such that both the mean transition period and the distributions of the transition periods were significantly different between warming scenarios within the same ice quality category (p < 0.05) and between ice quality categories within the same warming sce- nario (p < 0.05).

Generally, the distribution of transition period anomalies was more constrained under black ice conditions, though it showed right skewing when warming reached 4˚C (Fig 4). However, when ice quality became 50% and 100% white ice, the right skewness of the distribu- tions became more and more exaggerated (i.e., the distribution includes larger anomaly val- ues). The distributions between ice quality categories are significantly different from one another. While all distributions for both ice formation and melt periods are significantly Table 1. Unsafe transition period anomalies. Warming Formation Transition Anomaly Formation Transition Standard Deviation Melt Transition Anomaly Melt Transition Standard Deviation Sample size 100% Black Ice 1˚C 1.1 2.3 0.8 2.4 2829 2˚C 1.5 2.7 1.4 2.6 2762 4˚C 2.6 3.7 2.4 3.6 2618 50% White Ice 1˚C 8.0 4.9 3.9 3.5 2743 2˚C 8.7 5.7 4.7 4.4 2674 4˚C 10.7 6.3 5.9 5.4 2512 100% White Ice 1˚C 15.3 8.1 6.3 4.9 2643 2˚C 16.6 8.6 7.2 5.5 2580 4˚C 19.8 8.9 8.7 6.6 2379 The duration anomaly of the “shoulder season” when ice is � 2 cm and � 10 cm (black ice), � 15 cm (50% white ice), or � 20 cm (white ice), relative to a base period of 1851–1880. Values in parentheses are standard deviation. These values are for the Northern Hemisphere (latitude > 40 & < 80). Warming is in degrees Celsius; transition periods and standard deviation are in days and sample size is the number of grids present in the analysis. https://doi.org/10.1371/journal.pone.0313994.t001 PLOS ONE Changing ice conditions inhibit safe use

PLOS ONE | https://doi.org/10.1371/journal.pone.0313994 December 11, 2024 8 / 19