Weak evidence for heritable changes in response to selection by aphids in Arabidopsis accessions

The files are named as <phenotype > _<data set>.xlsx. For each phenotype, there are analyses with all three accessions (‘all’), only the two accessions with trichomes (‘NoWil3’), or with only one accession (‘OnlyWei0’, ‘OnlyMa0’, and ‘OnlyWil3’). For each of them, there is one version with all data included and one in which the extreme values were removed (‘.outliersRemoved’). For example: bladeArea_all.outliersRemoved.xlsx refers to the analysis of the blade area using all three accessions and removing extreme values. Within each workbook, there are three sheets for untransformed data (‘y’), log-transformed data (‘logy’), and square-root-transformed data (‘sqrty’). Within each sheet, the output of the two models tested either with the regular linear model (lm) or the random effects model (lmer) is given. In case of the regular lm output, alternative F- and p-values (F_altern and P_altern) refer to the tests against the correct error stratum (i.e., nested model). The columns F and P contain the F- and p-values from tests against the residuals. Regarding the model terms: ‘__Blocks’ is the five-level factor of experimental blocks that was split into two contrasts (‘EX’ and ‘BL’); ‘__Selection’ is the four-level factor with the selection treatments that was split into three contrasts (‘AN’, ‘CO’, and ‘SE’). ‘__Selection_X_accession/experiment’ is accordingly the interaction of the four-level factor with the selection treatments and the accession (‘ET’) or the ‘EX’ contrast. Abbreviations: Df: degrees-of-freedom, SS: sum of squares, MS: mean squares, F: F-value, P: p-value, F_altern: F-value of nested model, P_altern: p-value of nested model, perc_SS: % sum of squares explained by the term.

For each phenotype, power was estimated using all three accessions or only the two accessions with trichomes. In case of trichome density, only the latter is available. Each table contains the percentage of significant outcomes for a given simulation and factor/contrast of interest (based on 1,000 simulations). The simulations are named as <type > _<effect size>. There were five different types of simulations with effect sizes ranging from 1% to 20%: (‘step’) differences between all four groups with step sizes of 1%, 5%, 10%, 15%, and 20% between each group (with the level following AN<CO<BB<LE, note that the largest difference here is 60%); (‘onlyAn’) only different in ancestors; (‘onlyCo’) only different in controls; (‘onlySe’) different in both selection treatments; (‘onlyLe’) different in only one selection treatment.

Given are the sample IDs, line ID, treatment, generation, accession label (used by Lin et al., 2024), and the accession assignments based on the genotypes inferred from the RNA-seq data and the publicly available data from 1001 Arabidopsis accessions (this study).

Rosette diameter in mm, blade area in mm². Factors are labeled as described in ‘Materials and methods’.