Oxidative stress after pollutant exposure depends strongly on experimental design and pollutant properties: a meta-analysis

Measurements of reactive oxygen species (ROS) are often performed to assess a species’ general sublethal stress response to a pollutant. However, ROS bioassays often produce seemingly ambiguous results, and the drivers that lead to these differences are largely unknown. To approach this gap, we conducted a meta-analysis on ROS generation, ROS-associated damage products, enzyme activities, and gene expression levels in response to exposures to two groups of pollutants, nano- and microplastic particles (NMP) and neonicotinoid insecticides (neonics). Based on 2294 ROS-related measurements extracted from 45 studies, we show that measured effects vary substantially with a strong overlap of measured effects with zero. As likely drivers of this variance, we identified multiple parameters of experimental design and pollutant properties. Finally, we performed data simulations and power analyses to investigate how well single experiments are able to detect ROS-related effects. We show that 21 out of 27 ROS markers achieve sufficient power (80%) to demonstrate effects with sample sizes smaller 20. Given the pollutant-dependent variability in ROS related responses and the low power of some markers, conclusions derived from single studies with low sample sizes (smaller than 5) are however at risk of being less informative than previously assumed.