The authors perform a detailed analysis of 32 pesticide residues across 20 sites and two soil depths. The interpretation of these data lacks contextualization with respect to key soil physicochemical parameters. Soil properties such as organic matter content, pH, texture, cation exchange capacity, and microbial activity play a foundational role in governing pesticide fate, including sorption, degradation, and mobility. Yet, the paper does not report or incorporate any measurements of these variables in its analysis. This omission raises concerns about the robustness of both the site-to-site comparisons and the clustering results (Figures 2–7), which assume homogeneity in soil behavior without evidence.
How did the authors account for the influence of key soil physicochemical properties, such as organic matter content, pH, texture, and microbial activity, on pesticide residue distribution and persistence across sampling sites, given that these parameters are critical to interpreting residue mobility, sorption, and degradation? Without reporting or normalizing for these variables, how can the site-to-site comparisons and cluster analyses be considered robust?
You mention that for each of the 20 sites, you collected three sub-samples per depth, mixed them to create one composite sample per site, and then selected three sub-samples from this composite for analysis.
My question is: Were these three sub-samples from the composite just technical replicates (i.e., you split the one composite and ran it through the instrument three times) or were they true analytical replicates (i.e., you performed three separate, complete extractions and analyses on the same composite sample)?
This is crucial. If they were only technical replicates, then you have only measured the instrumental variation for a single extraction per site. You haven’t accounted for the potentially large variability introduced during the extraction and clean-up process itself. The high recovery percentages in your table are from spiked samples, which don’t always reflect the real-world efficiency and consistency of extracting aged, bound pesticide residues from complex soil matrices.
If you didn’t perform independent extractions, then all the standard deviations you report for the pesticide concentrations are artificially low. They represent the precision of your GC-MS, not the accuracy and reproducibility of your entire method for these real soil samples. This means the actual uncertainty in your data is much larger than reported, which could invalidate your statistical clustering, your site-to-site comparisons, and the overall conclusion about which pesticides are “most abundant.”
Could you please clarify the replication procedure? Without true analytical replicates, the entire quantitative foundation of the paper becomes very shaky.