In Low Arctic tundra, thermal erosion of ice‐rich permafrost soils (thermokarst) has increased in frequency since the 1980s. Retrogressive thaw slumps (RTS) are thermokarst disturbances forming large open depressions on hillslopes through soil wasting and vegetation displacement. Tall (>0.5 m) deciduous shrubs have been observed in RTS a decade after disturbance. RTS may provide conditions suitable for seedling recruitment, which may contribute to Arctic shrub expansion. We quantified in situ seedling abundance, and size and viability of soil seedbanks in greenhouse trials for two RTS chronosequences near lakes on Alaska's North Slope. We hypothesized recent RTS provide microsites for greater recruitment than mature RTS or undisturbed tundra. We also hypothesized soil seedbanks demonstrate quantity–quality tradeoffs; younger seedbanks contain smaller numbers of mostly viable seed that decrease in viability as seed accumulates over time. We found five times as many seedlings in younger RTS as in older RTS, including birch and willow, and no seedlings in undisturbed tundra. Higher seedling counts were associated with bare soil, warmer soils, higher soil available nitrogen, and less plant cover. Seedbank viability was unrelated to size. Older seedbanks were larger at one chronosequence, with no difference in percent germination. At the other chronosequence, germination was lower from older seedbanks but seedbank size was not different. Seedbank germination was positively associated with in situ seedling abundance at one RTS chronosequence, suggesting postdisturbance revegetation from seedbanks. Thermal erosion may be important for recruitment in tundra by providing bare microsites that are warmer, more nutrient‐rich, and less vegetated than in undisturbed ground. Differences between two chronosequences in seedbank size, viability, and species composition suggest disturbance interacts with local conditions to form seedbanks. RTS may act as seedling nurseries to benefit many Arctic species as climate changes, particularly those that do not produce persistent seed.

This dataset is based on field sampling of soil seedbanks, in situ seedling counts and greenhouse germination trials. Material was harvested from 1x1 m plots along 50m transects. 4 transects per site, two sites consisting of thermokarst-disturbed lakes of different ages on Alaska's North Slope

There are no missing values in this dataset.