Sapling offsets did not predict 15-year adult shifts of 118 tree species on 133,949 remeasured eastern US forest plots; a shared mortality gradient caused most of the northward adult shift

Question: does the position of tree saplings relative to adults of the same species (the juvenile-adult offset, a common indicator of tree migration) predict where the adults move? Approach: we used 133,949 USDA Forest Inventory and Analysis plots in 37 eastern US states with a first visit in 2002-2010 and a second visit 12-18 years later (mean 14.8 years). All plots were fully sampled at all visits. We followed each tree between visits. For 118 species we compared the T1 sapling-minus-adult centroid offset with the observed shift of the adult centroid, in latitude and in 1981-2010 plot temperature. We divided each shift exactly into ingrowth, natural mortality, harvest and species re-coding. A species x grid-cell model separated a field common to all species from the species-specific part. CIs come from a 1-degree cell block bootstrap. Results: saplings were 17 km south of adults on average (0.19 C warmer places). The offset thus predicts a southward shift (-3.9 km/decade), but adults moved north (+4.5 km/decade, 95% CI 2.9 to 6.0) and to colder places (-0.037 C/decade, about 9% of local warming). Across species the expected and observed shifts had r = 0.20 and a calibration slope of 0.15 (a correct indicator gives 1). The sign agreed for 57% of species. Range-edge percentiles gave the same result. Natural mortality gave +3.0 km/decade of the shift; a mortality field shared by all species, higher in warmer cells, explained +2.8 of this. The species-specific mortality part was near zero (+0.2, CI -0.6 to 0.7), except for a small part (+1.2) when we removed five pest-affected genera. Species-specific ingrowth gave +1.8 (CI 0.7 to 3.0): saplings that became adults were 27 km north of the sapling pool. Results were similar with 0.5 and 2 degree cells, common species only, and plots without harvest. Limitation: one 15-year interval; abundance centroids and percentiles on fixed plots, not colonization outside the range; the causes of the higher mortality in warm cells (climate, stand dynamics, pests) are not separated.

range shift · tree migration · juvenile-adult offset · Forest Inventory and Analysis · demography · tree mortality · recruitment · climate change · eastern United States · indicator validation · thermophilization