Additional information and results of statistical tests.

Sheet 1: Qualitative variables. Number of entries according to the qualitative variables here studied. Sheet 2: Pyrolysis temperature. Kruskal-Wallis rank sum test and Dunn’s post-hoc test results for all quantitative BPCA outputs as a function of pyrolysis temperature, categorised by low (≤ 300 °C, n = 68), mid (350 ≤ x n = 139), and high (≥ 700 °C, n = 29) temperature chars. The degrees of freedom for each Dunn’s test is 2. Temperature categories with the same letter are not statistically distinct (p Sheet 3: Precursor feedstock. Results of two-way ANOVA with the Benjamini-Hochberg Procedure on all quantitative variables for low, mid, and high temperature chars among the precursor feedstock categories of hardwoods, softwoods, and grasses. Sheet 4: Oxygen availability during pyrolysis. Results of two-way ANOVA with the Benjamini-Hochberg Procedure on all quantitative variables for low, mid, and high temperature chars among the air composition categories “0,” “20.5,” “atmospheric,” and “atmospheric (restricted oxygen).” Sheet 5: Chromatographic separation method. Summary of statistical tests to investigate the effect of chromatographic separation method (GC- or LC-BPCA) for all, low, mid, and high temperature chars. The statistical test used was automatically determined in each case by the R code according to the sample size and results of the variable for the Shapiro-Wilk and Bartlett tests of normality. Statistically significant p-values are indicated in bold. Sheet 6: Random Forest results. Summary statistics for random forest predictive models testing various numbers of principal components, trees, and mtry parameters with PCA method 1 (omission) and 2 (imputation) for the treatment of missing values. In each case, the model with highest accuracy is indicated in bold. (XLSX)