Work

Relative density thresholds for gross growth, net growth, and mortality across forest types and ecoregions of the conterminous United States: code, aggregated estimates and figure data

Year: 2026
Type: dataset
Abstract: Analysis scripts, aggregated breakpoint estimates and figure data for Chivhenge et al., "Relative density thresholds for gross growth, net growth, and mortality across forest types and ecoregions of the conterminous United States" (submitted to Forest Ecology and Management). An annualized segmented mixed model was fitted to 276,984 USDA Forest Service Forest Inventory and Analysis remeasurement intervals on 85,821 plots to estimate relative density (RD) breakpoints for gross and net aboveground carbon growth, mortality incidence and mortality severity, nationally and within forest type by EPA Level III ecoregion cells pooled with a crossed random-effects meta-analysis. National breakpoints with projected RD were 0.42 (gross growth), 0.34 (net growth), 0.45 (incidence) and 0.50 (severity). Pooled local thresholds averaged 0.443 for gross and 0.447 for net growth, with 95% prediction intervals of 0.37 to 0.52 and 0.35 to 0.55. Supplementary Data S1 gives thresholds with 95% intervals for all 2,020 forest type by ecoregion combinations with FIA intervals. Files: graphical abstract (lead), Supplementary Data S1 (CSV), rdgrowth_data_v1.2.zip (aggregated tables), rdgrowth_code_v1.2.zip (scripts), rdgrowth_figures_v1.2.zip (figures under the revised manuscript numbering), README, data dictionary, CITATION.cff and the pre-publish gate reports. No plot coordinates, plot identifiers, plot-level records or credentials are included; ecoregions were assigned from public perturbed FIA coordinates. Version 1.1.0 (October 15, 2026) adds the analyses for the revised manuscript (subplot gap tests, SIMEX correction for plot RD error, ladder step 8, pooling stress tests and coverage checks, full-data starting-RD fits and the mortality incidence profile of Figure 2), their scripts, and the revised figure set, including a new map of gross and net breakpoints with their standard errors. Pooled thresholds and Supplementary Data S1 are unchanged. Version 1.2.0 (October 26, 2026) adds the October 19 to 26 rounds: the attrition by RD and RD error simulation tables, the October 21 and 26 figure passes with the figure data tables, the Table S12 recheck, the interval filter counts and the reanalyses R1 to R4, with the revised figure set (Figures 1 to 5 and S1 to S16). Version 1.1.0 was built and gated on October 15 but was never deposited, so this version also carries its additions. Pooled thresholds and Supplementary Data S1 are unchanged. The scripts match commit 3720ba28b261 of github.com/holoros/rdgrowth-conus (branch closeout-20261026, merged into main on October 26, 2026 as 6b3c4db0aec8). October 28 to 29, 2026 (before this version was published): adds rdgrowth_reanalysis3_v1.2.zip with the reanalyses run for the reviewer report of October 24, 2026 (projected RD path validation, alternative SDIMAX refits, slope bootstrap intervals, variance partition conventions, consecutive-interval and expansion-factor fits, threshold spread across density measures, undisturbed and ingrowth fits, harvest censoring weights, direct per-unit breakpoints, AIC size and power simulation, ecoregion error models), and the figure set is now Figures 1 to 4 and S1 to S19 (Figure 4 gross growth only with the extrapolated share, Figure S13 net growth, S17 workflow, S18 and S19 direct per-unit breakpoints). Version 1.3.0 (November 6, 2026) adds rdgrowth_round26to28_v1.3.zip with the Rounds 26 to 28 work: the universality maps (pooled thresholds beside the direct ecoregion fits, 60 km hexagon shares, the common RD window test, latest-visit stocking by region and type group), the merchantable volume rebuild (VOLCFNET) and its national fits, the spline boundary knot sensitivity of the hinge versus spline comparison, the net growth size and power simulation, the mortality share on consecutive intervals, the independent refit of the covariate-adjusted severity, the region threshold table, and the manuscript build scripts of Rounds 26 to 28. The figure set is now Figures 1 to 5 and S1 to S19 under the Round 26 numbering (Figure 1 curves with the Long 1985 density management zone, Figure 2 pooled and direct thresholds with open-ended end classes, Figure 3 where forests stand, Figure 4 pooled thresholds, Figure 5 breakpoint spread). Pooled thresholds and Supplementary Data S1 are unchanged. The scripts match commit 17f2ff89b1c5 (tag round28-20261106) of github.com/holoros/rdgrowth-conus. Version 1.4.0 (November 11, 2026) adds rdgrowth_round30to32_v1.4.zip with the Rounds 30 to 32 work: the optimal relative density analysis (the RD at which net growth levels off, from penalized regression splines of annual growth on the interval mean of projected RD with a whole-plot bootstrap, nationally, by region, and by Level III ecoregion), its sensitivity and stress tests (basis size, weighting, consecutive visits, starting RD, interval length, forest type group and region terms, interpolated RD), the 60 km hexagon maps of the pooled thresholds with their standard errors and of latest plot RD against the local threshold and the optimal RD, with the data behind them, and the manuscript build scripts of Rounds 30 to 32. All outputs are aggregated; plots are placed in hexagons from public coordinates. Pooled thresholds, Supplementary Data S1, and the other files are unchanged from version 1.3.0. The scripts match commit c373dbf2744a of github.com/holoros/rdgrowth-conus. Version 1.5.0 (November 11, 2026) adds rdgrowth_round33to35_v1.5.zip: the optimal relative density by measurement period (the optimum is found on intervals centered before mid-2013 and not on later intervals; period is confounded with interval length and region) and the manuscript build scripts of the Round 32 close-out and Rounds 33 to 35. The companion manuscript is now titled "Optimal relative density for forest dynamics attributes across the conterminous United States". All other files are unchanged from version 1.4.0. The scripts match commit 73b84711ff27 of github.com/holoros/rdgrowth-conus. Version 1.6.0 (November 11, 2026) adds rdgrowth_round36to38_v1.6.zip: the design-based forest land area of the 48 conterminous states from FIADB by ownership, origin, reserved status and stand age (state and national aggregates), used to describe the study domain, and the manuscript build scripts of Rounds 36 to 38. All other files are unchanged from version 1.5.0. The scripts match commit 8772b964e42d of github.com/holoros/rdgrowth-conus. Also in version 1.6.0: rdgrowth_round39to41_v1.6.zip, with the Round 39 to 41 manuscript builders and the optimal relative density refitted on intervals with and without recorded natural disturbance (Supplementary Table S46).
Language: English
Subfield: Forestry
Open Access status: green
OpenAlex ID: W7222175261
Cites:
Cited by:
Related to: 8