Originpro v2020

Statistical analysis was performed in R (v.4.2.1)¹⁰⁹ . For comparison between specific sets of datapoints, we used parametric and non-parametric testing procedures as indicated in the text. OriginPro v.2020 (OriginLab) was used for analysis of electrophysiological data. A multiple regression analysis using a GLM was performed analogous to that in ref. ^{110 (link)}. Briefly, we first calculated the abundance of environmental transcripts annotated with bacteriorhodopsin-like protein domains (Pfam: PF01036) and normalized these as a percentage of total transcripts mapped in each metatranscriptome sample. A GLM was then fitted using the glm function of the R stats package (v.4.2.1) using percentages of rhodopsin domains as response variable and temperature, salinity, hemisphere and the transformed nutrients (log(phosphate+0.01), log(nitrate+0.01), log(silicate+0.01) and log(iron+0.01)) as covariates. We then performed a backward elimination of factors by AIC using a stepwise algorithm with the step function of the R stats package. Regression summaries were exported using the jtools R package (v.2.2.0)¹¹¹ . Data analysis and visualization were performed in R tidyverse^{112 (link)} and using ggplot2 (ref.¹¹³ ) and ggtree¹¹⁴ .

Normal distribution and homogeneity of variances were tested by the Shapiro–Wilk and Levene tests, respectively. Student's t‐test was used to determine if the means of two datasets were significantly different from each other (P < 0.05). All statistical analyses used Python (v.3.7), Shapiro–Wilk test, Levene test and Student's t‐test were performed using the SciPy library. The piecewise function was fitted by linear and exponential goodness‐to‐fit regression (OriginPro v.2020; OriginLab, Northampton, MA, USA).

ROS imaging data were analyzed with EasyRatioPro (PTI, HORIBA Scientific) software and further processed with Excel (Microsoft, Redmond, WA, USA) and Igor Pro v8.0 (Wavemetrics, Lake Oswego, OR, USA) software. Protoplast images were processed with ImageJ (NIH). Figures were prepared with Origin Pro v2020 (Originlab, Northampton, MA, USA) and Adobe Illustrator v24.1 (Adobe, San Jose, CA, USA). Averaged data are presented as means ± SEM (N = number of protoplasts from 3–5 independent measurements). For comparisons with two groups such as basal ROS levels and ROS levels from DF M. sexta OS and tomato PF M. sexta OS, we used the non-parametric Mann-Whitney U test. For comparison with three groups, as depicted in Figure 4, for basal OS/tbH₂O₂ and NAC, we used a non-parametric Kruskal-Wallis test followed by Dunn’s pairwise post hoc comparisons. Non-parametric tests were used since data failed to meet normality assumptions after transformations. For all analyses, data from extractions were pooled to attain a sample size of 66–124 protoplasts and were repeated for at least three replications. All analyses were carried out using GraphPad Prism v9.0 (La Jolla, CA, USA).

Statistical differences between untreated D. magna control and LDPE-exposed samples were evaluated in one-way ANOVA. The data obtained for woodlice were visualized and analysed using OriginPro v2020 software (OriginLab, Northampton, MA, USA). For normal distributions and homoscedasticity of the data, one-way ANOVA was performed followed by Tukey tests; otherwise, a non-parametric Kruskal-Wallis test was used, followed by Mann-Whitney U-tests (Tables S2 and S3). p < 0.05 (*) was considered as significantly different.