Supplementary materials for “From roots to crowns: Global patterns of Phytophthora-associated symptoms in forest trees”
Description
This dataset accompanies the manuscript “From roots to crowns: Global patterns of Phytophthora-associated symptoms in forest trees”. It contains the publication-ready supplementary figures and tables together with detailed data summaries, model outputs, validation results and exact-profile co-reporting results supporting the analysis of 32,202 occurrence records and 3,948 eligible exact profiles from 721 sources. Supplementary Material 1 – Additional Figures contains Figure S1, which shows the distribution of primary exact profiles among the 721 contributing sources on a logarithmic scale. Supplementary Material 2 – Detailed Supplementary Tables contains Tables S1–S8: outcome support and conditional reporting estimates; predictor support; multiplicity-supported host-stage and hydrological-context contrasts; host and strict single-taxon Phytophthora variance components; repeated whole-source validation; nine prespecified sensitivity analyses; results for 12 prespecified symptom co-reporting pairs; and host and Phytophthora taxon levels meeting the support criterion. Supplementary Material 3 – Documented Host Symptom provides the documented associations among Phytophthora taxa, forest-tree hosts and recorded symptoms represented in the analysed collection. Supplementary Material 4.1 – Missingness and Coverage documents cohort construction, terminal dispositions, response-state missingness and coverage across sources, geography, time, taxonomy and analytical fields. Supplementary Material 4.2 – Results and Readiness provides descriptive profile-weighted and equal-source results, outcome and predictor support, source concentration and readiness assessments. Supplementary Material 4.3 – Contrasts and Model Estimates contains model-adjusted probabilities, host-stage and hydrological-context contrasts, model estimates, uncertainty intervals and multiplicity-adjusted results. Supplementary Material 4.4 – Model Diagnostics records model acceptance and numerical diagnostics, including convergence, rank, singularity, optimiser agreement, influence checks and host- and taxon-associated heterogeneity. Supplementary Material 4.5 – Validation and Sensitivity contains repeated source-blocked validation metrics, fold-level diagnostic results and complete outputs from the nine prespecified sensitivity analyses. Supplementary Material 4.6 – Exact-Profile Symptom contains the readiness assessments and constrained-null results for the 12 prespecified symptom pairs, including observed co-reporting, null distributions, observed-to-null ratios, excess co-reporting and multiplicity-adjusted probabilities.
Files
Steps to reproduce
1. Import the accepted closed relational workbook containing 32,202 occurrence records and the linked source, evidence-profile, host–taxon detection, manifestation and occurrence-crosswalk tables. 2. Join the tables using the supplied identifiers and verify one-to-one occurrence-to-profile mapping, profile coverage and agreement of source identifiers. Retain duplicate occurrence multiplicity for provenance. 3. Apply the prespecified eligibility criteria: detected evidence from a primary source; evidence class E2–E5; a forest-tree or tree host; an explicit forest context; direct host-tissue or host-associated-rhizosphere linkage; an eligible resolved Phytophthora class; and at least one analytically authorised manifestation response. 4. Collapse eligible records into equally weighted exact source–context–response profiles. The final cohort comprises 3,948 profiles from 721 sources; use source as the clustering unit. 5. Code the eight recorded plant-part outcomes, 19 detailed symptom codes, confirmed asymptomatic status and eight symptom-associated domains. Retain missing, non-applicable, conflicting and unconfirmed states outside populated-field denominators. Derive host stage, explicit hydrological context, sampled-material compartment, observation-method family and strict single-taxon status as described in the manuscript. 6. Calculate profile-weighted and equal-source conditional estimates, with 2,000 whole-source bootstrap replicates, and apply the prespecified outcome and predictor readiness criteria. 7. Fit the accepted source-random-intercept Bernoulli-logit models, calculate adjusted marginal probabilities and contrasts, apply Benjamini–Hochberg correction, and estimate supported host- and Phytophthora-associated variance components. 8. Perform ten repeats of five-fold source-blocked validation, the nine prespecified sensitivity analyses and fixed-margin symptom co-reporting randomisations using two independent chains with 9,999 saved matrices per chain. 9. Export the supplementary figure, detailed tables, documented associations, descriptive results, model estimates, diagnostics, validation, sensitivity and co-reporting outputs. Analyses were conducted in R 4.3.3 using the package versions reported in the manuscript.