On 136,744 remeasured eastern US forest plots, sapling-tree latitude offsets of 89 species predicted a southward shift, but trees moved north in 2006–2021, mainly by tree loss in the south

Question: does the stage-for-time method predict tree range shifts? This method compares where saplings and adult trees grow in one inventory. We tested it on 136,744 USDA FIA plots in 37 eastern US states that were measured twice (T1 2002–2010, T2 2014–2026, mean interval 14.8 years). For 89 species, we compared the T1 offset between the sapling (1–4.9 in dbh) and tree (≥ 5 in) centroids with the later shift of the tree centroid on the same plots. We used a split-half design to remove shared sampling noise, a 1° grid-cell bootstrap, and nulls for the part that all species share. Results: at T1 the saplings were on average 14.3 km south of the trees [95% CI 6.9, 21.6] (species-specific part −17.8 km). But the trees moved 5.3 km north [3.0, 7.2] (species-specific part +6.3 km). At T2, the saplings were again 14.8 km south of the trees, and the offsets at T1 and T2 had a correlation of 0.91 across species. Thus the offset is a stable pattern, not a front that the trees follow. Between species, the offset predicted the realized shift only weakly (split-half r = 0.28 [0.15, 0.34], slope 0.10). The signs agreed for 56% of species. Offsets at the range edges did not predict the edge changes. We linked 8.0 million tree records between the visits. Trees that died were 27.6 km south and trees that were cut were 14.9 km south of the T1 tree centroid. Loss thus gave +5.0 km of the shift, and recruitment gave +0.5 km. The T1 saplings that reached 5 in were 19.2 km north of all T1 saplings, so the southern sapling surplus did not become trees. The results were the same with basal-area weights, other cell sizes, stricter species rules and a longer interval (16.9 years). Limits: one 15-year interval, one region, and saplings, not seedlings. Over many decades, the result can be different. Conclusion: a sapling-tree offset from one inventory did not show the direction of the near-term tree shift. Where trees die and are cut controlled the shift.

range shift · stage-for-time substitution · Forest Inventory and Analysis · tree migration · saplings · tree mortality · recruitment · eastern United States · remeasurement · climate change · species distribution