How stochastic cell fate and endoreduplication yield non-random epidermal patterns

Abstract Pavement cells in the Arabidopsis thaliana epidermis span a wide range of sizes and ploidy levels, but rules that generate this heterogeneity across an organ remain unclear. Clark et al. identify a shared genetic pathway that promotes large, polyploid pavement cells in both sepals and leaves, then ask whether the familiar “scattered” distribution of giant cells is truly random. By combining whole-tissue imaging with two independent computational randomization approaches that regenerate tissues from segmented images while preserving cell size distributions and key boundary constraints, together with a stochastic cell-autonomous model, the authors show how an initially random pattern can later appear clustered relative to a changing random baseline as tissues grow and subdivide. The study provides a quantitative framework for testing spatial organization in cellular mosaics where point-based methods fail, and it shows how proliferation history can convert early stochastic fate decisions into a statistically non-random mature pattern.