The following technical concerns are identified, each pointing to potential errors or limitations in the study design, data interpretation, or methodological approach:
1. The study use 5 replicates per treatment, but all replicates are within the same vineyard under the same management. The experimental design do not appears to include true spatial replication (e.g., multiple vineyards or geographically separated blocks). This raise the risk of pseudoreplication, where environmental or management confounders (e.g., microclimate, vine age, soil heterogeneity) are not adequately controlled.
2. The authors use FUNGuild to assign ecological functions (e.g., saprotrophs, pathogens) to fungal ASVs. However, FUNGuild is a predictive tool based on literature and databases, not experimental validation. Many fungal taxa have multiple ecological roles or are poorly characterized.
3. The authors construct co-occurrence networks from the top 50 ASVs and interpret network metrics (e.g., connectivity, stability) as indicator of microbial community health or function. However, correlation-based networks are highly sensitive to Compositional bias in sequencing data, Filtering thresholds (e.g., top 50 ASVs), and Lack of causal or functional inference.