Longitudinal Wastewater-Based Epidemiology Reveals the Spatiotemporal Dynamics and Genotype Diversity of Diarrheal Viruses in Urban Guangdong, China

Comments 1: Sampling design is inadequate for temporal inference: monthly, non-composite influent sampling is insufficient to resolve short-term viral fluctuations and can misrepresent both the magnitude and timing of peaks illustrated in Figures 2 and 3. Established WBE practice indicates that at least twice-weekly sampling is required for trend detection. As such, the seasonality and outbreak inferences made from these figures could likely suffer from temporal aliasing.

Response 1:We thank the reviewer for their valuable comments regarding our sampling frequency. We acknowledge that a higher frequency, such as twice-weekly sampling, represents the gold standard for detecting real-time trends and capturing short-term fluctuations.

However, logistical and resource limitations during the initial two-year surveillance period restricted us to a monthly sampling protocol. The principal aim of this research was not to serve as an early warning system for outbreaks. Instead, it was intended to establish a long-term epidemiological baseline and to systematically analyze the seasonal distribution characteristics of various diarrheal viruses across three major cities over a two-year period.The monthly sampling strategy we selected is consistent with this specific objective and aligns with longitudinal WBE studies on enteroviruses conducted in this and other regions for similar purposes, as exemplified by "Capturing Noroviruses Circulating in the Population: Sewage Surveillance in Guangdong, China (2013–2018)". As we have stated in the manuscript, this precedent supports our methodological choice (see Page 15, Lines 430-432, Reference 36).Regarding this part of the discussion, we have added it to the discussion section, as”Thirdly, the monthly sampling is insufficient for capturing short-term dynamics. The observed patterns should therefore be viewed as broad seasonal trends, and future work requires higher-frequency sampling for more precise epidemiological tracking. "(page12,line312-314)

Comments 2: Sequencing lacks reproducibility and transparency: only mapped reads, not raw FASTQ files nor complete pipeline parameters, were deposited. This prevents independent verification of the genotypes reported in Figures 4 and 5. For studies making genotype-level claims, full raw data deposition is essential.

Response 2:  We appreciate your valuable comments on the manuscript.The raw FASTQ file has been uploaded to the GSA database, as stated in the article.”All sequencing reads mapped to the diarrheal viruses have been deposited to the GSA database of National Genomics Data Center (https://bigd.big.ac.cn/) with submission number PRJCA046343. ”（page6,line 147-148）

Comments 3: No process controls were used to quantify extraction or concentration efficiency, and no internal amplification controls were included to assess inhibition. Consequently, the detection frequencies in Figure 1 cannot be interpreted as biological differences across viruses; they may instead reflect workflow biases.

Response 3: We apologize for the omission of this information. Although a process control was not added to each sample to quantify the concentration efficiency, we did perform a systematic validation of our concentration step using Pepper mild mottle virus (PMMoV) before the study commenced. This validation was conducted on six authentic wastewater samples, yielding recovery rates between 34.4% and 53.6%. Furthermore, to assess potential PCR inhibition, an internal control (IC) included in the commercial extraction kit was used for every sample. We analyzed the IC Ct values from 71 samples and found them to be highly consistent, with a standard deviation (SD) of less than 0.5 Ct and a coefficient of variation (CV) of less than 2%. This information has now been added to the manuscript as follows:“The magnetic bead-based viral enrichment method was validated pre-study using PMMoV, yielding a recovery of 34.4%–53.6%. An internal control (IC) co-processed with each sample confirmed  PCR inhibition (Ct SD < 0.5; CV < 2%). "(page5,line109-111)

Comments 4: The study lacks virus-specific LODs and recovery data that prevents fair cross-virus interpretation (Figure 1). Without reporting analytical sensitivity and recovery for each target, differences such as 100% Astrovirus detection versus slightly lower Norovirus GII rates may be methodological rather than epidemiological. This is a fundamental omission for any comparative detection analysis.

Response 4: We apologize for the omission of this information. The viral-specific detection limit (LOD) of each kit we used was 500 copies/mL. Regarding recovery efficiency, we acknowledge that in this study, we did not perform viral-specific recovery rate assays for each of the seven targets individually, which is an objective limitation. However, to control the overall efficiency of our entire experimental workflow (including nucleic acid extraction and purification), we used PMMoV as a process control for preliminary experiments. PMMoV is a stable, non-enveloped RNA virus widely present in human feces. Our data show that PMMoV exhibited high recovery rates across different samples. We are now adding this information to the text, as”According to the manufacturer, the quantitative PCR kit has a reported typical limit of detection (LOD) of approximately 500 copies/mL for each target.”(page5,line 107-109)

Comments 5: The tNGS confirmation is limited to a small number of samples from one city, yet RT-PCR positivity is assumed accurate across all sites and months, including the 100% detection claims. Broader sequencing confirmation is necessary to exclude false positives and cross-reactivity.

Response 5: Thank you for your valuable feedback.The design rationale of this study is as follows: tNGS was employed as an orthogonal method validation. We selected a representative city with high viral prevalence and conducted parallel testing on samples collected over 12 consecutive months. The core objective of this step was to validate the high specificity and reliability of our designed RT-qPCR primer-probe combination in real-world complex environmental matrices, enabling effective differentiation between the target virus and other microorganisms. The validation results confirmed the robustness of our methodology. In addition to tNGS validation, stringent quality control procedures were implemented throughout the study to ensure the accuracy of data from all sites and months. For instance, no-template controls (NTCs) were established for each batch to monitor contamination, and positive controls were set to ensure reaction efficiency. These measures minimized the likelihood of false positives and cross-reactivity.Meanwhile, as you pointed out, for weakly positive samples with very high Ct values (>35), although we have excluded obvious issues through strict quality control, there remains a theoretically minimal risk of nonspecific signals. To enhance the rigor and transparency of our conclusions, we have adopted your valuable suggestion and added the following sentence in the discussion section of the manuscript to elaborate on this limitation:”Maintaining comprehensive probe coverage across highly variable viruses remains a challenge for tNGS, as indicated by the disparity between RT-PCR positive EV samples and undetectable EV sequences. For samples with high Ct values (>35), the potential for non-specific signals, although minimized by stringent quality controls, cannot be completely ruled out. ”(page 13 ,line 316-319 )

Comments 6:The 3.95-fold spring increase in Norovirus GII (Figure 3) is presented without normalization for influent flow, population equivalents, or fecal strength markers (e.g., PMMoV or crAssphage). Without normalization, dilution effects cannot be ruled out as contributors to the observed patterns.

Response 6: Thank you for your valuable feedback. We agree on the importance of normalization and acknowledge that the lack of such data is a limitation of our study. The discussion section in the text illustrates this limitation, and the phrase 'the 3.95-fold spring increase in Norovirus GII' in the abstract and conclusion has been revised and supplemented, as” Finally, several technical factors inherent to wastewater surveillance must be considered, including flow rate variations, dilution effects, and environmental inhibitors.”(page 12,line 314-366) “We revised the language in the Abstract and Conclusion, shifting the focus from the precise "3.95-fold increase" to a more cautious description of a "significant upward trend in spring."(page 2,line 37; page 7,line 177-179) We wish these revisions address your concern. Thank you again for your suggestion.

Comments 7: No exogenous recovery control was included to account for seasonal method variability. Month-to-month variation in solids content, temperature, and inhibitor load can shift recovery efficiency and mimic the changes observed in Figures 2 and 3. Recovery controls are required to distinguish methodological noise from true epidemiological signal.

Response 7: Thank you for your valuable feedback regarding our manuscript.The limitation regarding recovery efficiency is duly noted. We observed that the seasonal trends, especially the norovirus winter peak, show a notable similarity to established epidemiological patterns in the region (References 34, 37). This alignment may serve as an indirect validation of our findings.To mitigate downstream analytical variability, we incorporated a stable internal control (IC) to monitor for potential PCR inhibition, a known issue in wastewater matrices (Reference 16). The consistent performance of our ICs suggests that our quantitative results were not significantly skewed by inhibitors.Taking these factors into account, while direct recovery efficiency is absent, the combination of our internal controls and the corroboration from epidemiological data leads us to believe that the observed trends likely represent genuine viral dynamics.

Comments 8: Genotype profiles are derived almost exclusively from Foshan 2024 samples (Figures 4 and 5), yet the manuscript extrapolates these patterns to interpret multi-city seasonal peaks shown in Figure 3. Such generalization is not justified without sequencing from Dongguan and Zhuhai and across both years.

Response 8:I apologize for the misunderstanding in my previous expression.We acknowledge that the findings presented in our initial manuscript regarding Figure 3 (Seasonal Peaks Across Multiple Cities) and Figures 4/5 (Genotyping in Foshan) are logically distinct. These constitute two interrelated yet independent discoveries, rather than a causal relationship. We have revised the conclusions of the genotyping section from a "universal explanation" to a "case study." We have made revisions to the relevant content to clearly differentiate between our direct findings and prospective hypotheses. The primary modifications are as follows:“To gain deeper insights into the molecular epidemiology behind the observed viral load dynamics, we conducted a detailed case study using targeted metagenomic sequencing. A custom probe enrichment panell—targeting 38 human viruses, including all diarrheal viruses detected in this study—was applied to 12 wastewater samples collected specifically from Foshan during its 2024 peak season.”(page 9 ,line 214-218 ) ”Our case study in Foshan provides a potential explanation for this phenomenon: the spring peak in that city coincided with the predominance of the GII.17 genotype. Given the geographical proximity and high interconnectivity, we hypothesize that similar genotypic shifts may have driven the concurrent peaks in Dongguan and Zhuhai. However, this remains a hypothesis that requires direct molecular surveillance in those cities for validation. This regional synchrony was less pronounced in the sporadic dynamics of Enterovirus (EV) and Norovirus GI (Supplementary Figure S1), suggesting that community-level outbreaks are also substantially influenced by independent local factors.”(page 11 ,line 283-290) “Secondly, the genotype analysis was confined to Foshan in 2024, making any extrapolation of these specific viral patterns to other cities or time periods speculative without further targeted sequencing.”(page 12 ,line 310-312 )

Comments 9: Norovirus recombination is common, yet the genotyping workflow does not describe dual polymerase–capsid typing, contig assembly, or recombination screening. Consequently, the genotype assignments in Figure 5 and Supplementary Figure S3 may conflate recombinant and parental strains.

Response 9: We thank the reviewer for these important technical points regarding our genotyping analysis. Our genotyping was performed using the EsViriru workflow (v0.1.1), which works by aligning sequencing reads to a reference database. This read alignment-based strategy inherently supports dual-genotype typing. We only identify a dual genotype (e.g., GII.P16-GII.4) when reads from a single sample can be aligned to both the GII.P16 polymerase region and the GII.4 capsid region in our reference library with sufficient coverage and confidence. This process effectively distinguishes known recombinant genotypes from parental strains or simple mixed infections based on the genomic distribution of aligned reads. This approach provides a robust and appropriate method for accurately identifying prevalent genotypes in communities, including common recombinants. We believe this method is suitable for the monitoring objectives of this study. The relevant content has now been supplemented in the text as follows” This read-mapping approach assigned dual genotypes  based on  the capsid region (VP1 for enteroviruses, ORF2 for noroviruses), allowing for the accurate identification of known recombinant strains.”(page 5,line 125-127)

Comments 10: Relative abundance estimates in Figure 5 do not appear corrected for probe density, genome length, or recombination-aware alignment. Without these corrections, the dataset may overrepresent some genotypes and underrepresent others.

Response 10: We apologize for the omission of this information. Regarding probe density, it is uniformly distributed. In Figure 5, relative abundance is normalized by dividing the original reads per genotype by the length of its reference genome. Subsequently, the relative abundance of a genotype is calculated as the ratio of its normalized reads to the total normalized reads of all detected genotypes in the sample. Relevant information will be supplemented in the text, expressed as”To correct for genome length bias, the relative abundance of each viral genotype was calculated as the proportion of its Reads Per Kilobase (RPK) value relative to the total RPK sum of all genotypes within the sample.”(page 6 ,line 129-131 )

Comments 11: The inference that specific genotypes “drive” seasonal peaks (Figure 3) is based solely on monthly data from one city (Figure 5) without formal attribution modeling, cross-correlation analysis, or multi-site replication. Such causal statements are premature.

Response 11: We apologize for any confusion regarding the specific genotypes “drive” seasonal peaks descriptions.Due to limitations in research resources, our deep sequencing analysis was concentrated on Foshan as a representative city, failing to cover Dongguan and Zhuhai. We acknowledge that this limitation restricts the generalizability of our conclusions. Added qualifying language (e.g., "appears to drive," "likely contributed to") to our causal statements throughout the manuscript. Revised the key sentence in the Discussion to frame the connection as a compelling but unconfirmed explanation, and explicitly stated that "definitively establishing this causal link would require multi-site replication and formal attribution modeling."This information has been added to the article, as”The synergy between genotypic and quantitative data is evident. For instance, the spring peak coincided with the emergence of GII.17, suggesting a potential link between genotype shifts and seasonal dynamics (Figure 5). This integration highlights the potential for WBE to not only track transmission intensity but also to elucidate the predominant viral strains during outbreaks (e.g., NoV GII.17, AstV HAstV-1). Such molecular insights could be valuable for informing vaccine development, evaluating targeted interventions, and optimizing public health responses.”(page 12,line 293-299)

Comments 1: Please add something about antibiotic resistance genes in the viruses you cited, and add some recent references about this aspect, eg. “Presence of Potentially Infectious Human Enteric Viruses and Antibiotic Resistance Genes in Mussels from the Campania Region, Italy: Implications for Consumer’s Safety, Venuti et al. 2025”

Response 1: Thank you for your valuable suggestions. I have incorporated them into the Introduction section of the article: “Moreover, wastewater serves as a critical reservoir for antibiotic resistance genes (ARGs), making it a powerful tool for monitoring antimicrobial resistance (AMR) [22]. Although integrated surveillance of viruses and ARGs is an ideal public health strategy, the viral component is often overlooked. Therefore, establishing a reliable baseline for viral pathogens is a fundamental step.”(page 4,line 72-76)

Comments 2: It is not clear if you used internal control for reactions, and if so, specify which one you used.

Response 2: We apologize for the omission of this information. Although a process control was not added to each sample to quantify the concentration efficiency, we did perform a systematic validation of our concentration step using Pepper mild mottle virus (PMMoV) before the study commenced. This validation was conducted on six authentic wastewater samples, yielding recovery rates between 34.4% and 53.6%. Furthermore, to assess potential PCR inhibition, an internal control (IC) included in the commercial extraction kit was used for every sample. We analyzed the IC Ct values from 71 samples and found them to be highly consistent, with a standard deviation (SD) of less than 0.5 Ct and a coefficient of variation (CV) of less than 2%. This information has now been added to the manuscript as follows:“The magnetic bead-based viral enrichment method was validated pre-study using PMMoV, yielding a recovery of 34.4%–53.6%. An internal control (IC) co-processed with each sample confirmed  PCR inhibition (Ct SD < 0.5; CV < 2%). "(page5,line105-107)