Non-fire tree mortality on 201,087 unharvested US forest plot intervals rose 4% per decade in 2000–2025; without ash trees the national trend was zero, and stand ageing did not explain the change

Question: Did natural tree mortality increase in forests of the 48 conterminous US states between about 2000 and 2025? If so, did older and denser stands, or other causes, explain the change? Approach: We used USDA Forest Inventory and Analysis (FIA) plots that were measured two times or more (data downloaded on 23 October 2026). The main sample has 201,087 consecutive plot intervals on 102,154 plots with no harvest and no fire record. It contains 4.39 million records of trees with a diameter of 12.7 cm or more, and 374,907 non-fire deaths. We fitted complementary log-log binomial models with an offset for interval length, so the coefficients are ratios of annual hazards (HR) relative to interval midpoints in 2007-10. The models had state terms, and some also had stand-state terms (size, stand age, basal area, mean diameter, forest type, planted origin). The confidence intervals are clustered by county. Other checks were same-plot comparisons of consecutive intervals, removal of one genus at a time, and excess deaths by genus and by recorded cause. Results: The national crude rate was 1.39% per year. The trend was +4.1% per decade (HR 1.04, 95% CI 1.02-1.06), and most of it came after 2019 in the East. With ash (Fraxinus) removed, the national trend was 1.00 (0.98-1.02). In 2019 and later, the ash hazard was 6.5, 3.5 and 3.4 times its 2007-10 value in the Northeast, Midwest and South. Insects were the recorded cause for most ash deaths, which agrees with the spread of the emerald ash borer. Ash gave 104%, 72% and 34% of the excess deaths in these regions. In the Pacific states, non-fire mortality rose by 32% per decade, mostly in true firs and Douglas-fir. In the Interior West, it fell by 30% per decade because pine deaths from insects decreased after the bark-beetle outbreak of the 2000s. The stand-state terms changed the period HRs by 0.03 or less. On the same plots, the stand-state change predicted a change in hazard of -0.4% to +3%. Limitations: The cause codes are not consistent between regions. Intervals of 5-10 years cannot resolve single years, and the West has few midpoints after 2019. Intervals with midpoints before 2007 had higher rates for a reason that we did not find. The results are associations, not projections.

tree mortality · Forest Inventory and Analysis · emerald ash borer · Fraxinus · background mortality · stand development · forest health · bark beetles · drought · hazard ratio · remeasured plots · United States forests