In 35,851 natural US forest plots, the tree richness–productivity elasticity of 0.30 comes from tree number: it is 0.02–0.05 with tree-number control, and rarefied-richness values stay within a null

Question: Liang et al. (2016, Science) reported that forest productivity P increases with tree species richness S as P ∝ S^θ, with θ = 0.26. They stated that sampling issues had little effect. Does this elasticity stay in US Forest Inventory and Analysis (FIA) plots when we remove the part that comes from tree number? Approach: We used 35,851 natural, undisturbed, single-condition FIA plots in 48 states, with first visits in 1996–2022 and second visits in 2001–2026. P is the gross stem-volume increment (survivor growth plus ingrowth) of trees ≥ 12.7 cm. We fitted log P on log S with 1° cell fixed effects and several control sets. A species-label permutation null shuffles species among trees in each cell. It keeps each tree's plot, size and growth, so it has no species-identity effect. Rarefied richness (ES5) is independent of tree number. Results: Without covariates, θ = 0.30 (95% CI 0.26–0.34). All of it comes through tree number: the elasticity of tree number to S is 0.37 and that of productivity per tree is −0.07. With basal area and elevation as covariates (Liang et al. type), θ = 0.13, but the permutation null gives θ = 0.49 for the same model. Thus a θ above 0.26 can occur with no species effect. When log tree number is in the model, θ = 0.02–0.05, only 0.03–0.04 above the null. Rarefied richness gives θ = 0.01–0.12, inside the null range. The result is the same for basal-area and biomass increment, 0.5° cells, earliest intervals and ES3/ES8. In the West and in some forest types (spruce–fir, Douglas-fir, lodgepole pine), θ = 0.11–0.35 remains after control of tree number. In loblolly–shortleaf pine and oak–pine forests, θ is negative. Limitations: The data are observational. Tree number can mediate a mixture effect, but rarefied richness does not need that control. We did not repeat the geospatial random-forest method or the climate covariates of the original study.

biodiversity-productivity relationship · forest inventory · FIA · species richness · sampling effect · permutation null model · rarefaction · replication · forest productivity · tree diversity