Given the data and analyses of Menzies et al. (2022), how would you recommend refining the model of Rolsky and Kelkar (2021) (e.g., changes to conceptual model, model assumptions, equations, parameterization)?

I don’t see any major flaws in the model structure; the main uncertainty lies in the relatively low degree of confidence in many of the parameters used to estimate model inputs and PVA degradation rates, as well as assumptions about proportional pathways of PVA within the WWTP process. I’m not convinced that the results presented in Menzies et al. (2022) significantly improve the situation. They show that under lab conditions (22˚C throughout?) degradation is fairly rapid to around 80% loss after a lag period of 1-4 weeks, but that there is much lower and more variable degradation in river water samples (maybe at most 25-50% loss), which presumably are similar to conditions post-release for most WWTPs (and it is not clear, but likely they tested at 22˚C, not ambient temperatures, so these are likely inflated estimates). If the data on average retention times in WWTPs presented in Rolsky and Kelkar (2021) are correct, then the findings of Menzies et al. have little bearing on the proportions of PVA degraded in WWTPs – if anything they suggest the rates used by Rolsky and Kelkar (2021) are optimistic.
If the goal is to provide a better estimate of environmental leakage of PVA, then I would suggest the following as the key action points:
1. Obtain direct estimates of PVA use in laundry and detergent pods from industry production data (and ideally add data from other PVA uses). If spatial use/release is important, then obtain regional sales data of pods.
2. Test assumptions about the proportions of PVA going into solid vs liquid phase treatments in the primary and secondary clarifier stages in WWTPs.
3. Refine the data on WWTP handling protocols; i.e. obtain better information on WWTP procedures to better account for likely degradation rates in water entering treatment plants. Key parameters include retention time in ASP (in particular how much PVA remains dissolved vs in the sludge), and resultant degradation rates in the ASP process across a range of WWTP facilities and environmental conditions (summer vs winter). Levels of tertiary treatment also appear to be important in determining the amounts of PVA persisting after treatment, so having a better idea of the proportion of plants that have different tertiary treatment options would improve understanding of PVA leakage.
4. If feasible, screen receiving environments (WWTPs, rivers and marine systems for coastal outfalls) for the presence and activity levels of PVA-degrading bacteria. Menzies et al. (2022) show that mineralization rates vary considerably in river water inoculums, presumably at least in part due to the presence of appropriate bacterial strains. It might also be possible to use the presence of PVA-active bacteria as an indicator of PVA leakage (i.e. if they are largely constrained to environments where PVA is fairly abundant, then their presence could indicate the spatial extent of PVA contamination downstream from WWTP and untreated sewage outfalls).

The analyses proposed are difficult to be applied for the study of PVA in effluents. Respirometric measurements are not specific for PVA, and the presence of other organic compounds in the analyzed wastewater would mask degradation results.
The analysis proposed can be useful to evaluate the PVA degradation capacity of different activated sludge samples. These experiments could be done with PVA as only carbon source and using the different biomass samples. A more active sludge would be obtained if PVA are available in the wastewater typically treated.
The calculation and use of the correct PVA distributed in RAS/WAS would be required to obtain more reliable results. The PVA degradation rates could be used for the critical point that is the calculation of the PVA removal efficiency for the activated sludge in the aeration tanks.
The estimation of the untreated wastewater for each state should be made more carefully, specially using the real WWTP treatment capacity for each state. The fraction of untreated wastewater should be also estimated with a better study, not using data from India.
There is also the need to obtain better data of PVA removal in several parts of the plant under more real operational conditions. PVA has been reported to be biodegraded by some specific microorganisms, hence it is expected that some PVA-degraders could be developed if the effluents treated have continuous input of PVA.

start with parameterization

- Several different qualities (MW and DH) of PVA (dubbed "PVOH") were degraded extensively (with three qualities reaching ~90 %) within 60 days in OECD 301-type studies. These PVA polymers all satisfied the conditions to be labelled readily biodegradable (>60 % biodegradation within 28 days). While Rolsky & Kelkar may not agree with this statement, and it should not necessarily be considered as irrefutable proof, meeting the criteria for ready biodegradability is generally considered to be conclusive evidence that a material should not be considered persistent. As I stated before, the translation of lab-scale-data to WWTP-scale by Rolsky & Kelkar may have been too conservative, and the results obtained by Menzies et al. with different qualities of PVA tend to support a less conservative approach.

- PVOH79 was degraded extensively in the GMR environmental river water sample, although the concentration of PVOH79 was hypothesized to be orders of magnitude greater than the environmentally predicted concentration. More variability was observed in the MR study, although one replicate still reached 79 % indicating the source river for this environmental river water sample is still likely to contain competent degraders. These findings may also be reason to re-evaluate the relatively low biodegradation rates in the WWTP model presented by Rolsky & Kelkar. Furthermore, it could be argued the model presented by Rolsky & Kelkar might benefit from an (optional) additional compartment, in which removal efficiencies of PVA in surface water is taken into account. This could be a simple one-compartment model or more extensive, depending on quality and abundance of available data.

- In general, Menzies et al. observe that a shift from 301-type to 302-type studies, with one of the major differences being an improved biomass ratio in the 302-type studies, result in faster biodegradation rates but similar lag-phase. This was especially pronounced for the slower degrading polymers. This appears to tie in with my previous remarks that the translation of biodegradation studies to WWTP-scale by Rolksy & Kelkar did not adequately take into account the much higher biomass concentration and several orders of magnitude higher biomass-to-PVA ratio. One recommendation by Menzies et al. is to give more weight to studies performed with higher biomass concentrations, a recommendation which is in line with my personal scientific opinion as well as the way I perceive the current general opinion of the scientific community to be. The results obtained by Menzies et al. with different qualities of PVA tend to support a less conservative approach than the one adopted by Rolsky & Kelkar, and subsequently would lead to a model with lower PVA concentrations in sludge and effluent.

1. Rolsky and Kelkar (2021) should perform their own waste water samplings (to verify concentration of PVA and amount of PVA degraded in a WWTP) instead of just taking outdated data from the literatures.

2. An online survey of 527 respondents is hardly a meaningful representative of the general US population, which has a population of 330 millions (<1% of total population). Perhaps the authors would like to consider narrowing their scope to the laundry purchasing habit of a town or a city instead?

3. Authors should be more careful in presenting their calculations and equations. The amount of daily treated wastewater per facility should be 2.12 million gallons instead of 2.12 billion (Page 4 of 15). Equation (3) should read WWT,2015 = NF * 2.12 MGD (million gallons) instead of billion gallons.

4. Rolsky and Kelkar (2021) should also improve the conceptual model, assumptions, equations and parameterization so that data on degradation of PVA can be obtained. The authors should also be clear on the parameters used in the study.

5. Data and analyses of Menzies et al. (2022) is good example to evaluate the biodegradation of water soluble polymer. For example, Menzies et al (2022) has shown Biodegradation of water-soluble polymers in Fig. 1,2, 3,4, whereas Rolsky and Kelkar (2021) does not have any results/table/graphs on degradation of PVA.

6. It should be noted that the study conducted by Menzies et al. (2022) showed great differences with the study performed by Rolsky and Kelkar (2021). Menzies et al.’s work involved two test methods on the real water sample while Rolsky and Kelkar’s work adopted online survey plus existing literature for estimation. The study nature was different, so if recommendation is necessary to refine the current work by Rolsky and Kelkar, suggestions are as follows.

a. The existing model assumptions and equations can be remained to estimate the final outcome as reference value. Then, Rolsky and Kelkar may extend their work to the collection of real samples from pre-treatment until post-treatment to verify the estimated value. The scope of study may also expand by considering the data from other countries to verify the robustness of the proposed models.

b. Rolsky and Kelkar may extend their work by collection of real samples. Then, the OECD 302B Zahn Wellens Test method can be adopted since Menzies et al. concluded that this test method shows better prediction efficiency than the OECD 301B Ready Biodegradability Test.

The biggest weakness is the lack of representativeness of the online survey results for the U.S. population, and the data from that section of the study is questionable, which affects the assumptions about PVA loading into the wastewater treatment facilities. In addition, it is worth noting that the authors conducted this study during the COVID-19 pandemic, and they used pre-pandemic data on water use and wastewater treatment. There are many scholarly publications on this topic, for examples:

Li, D., Engel, R.A., Ma, X., Porse, E., Kaplan, J.D., Margulis, S.A. and Lettenmaier, D.P., 2021. Stay-at-home orders during the COVID-19 pandemic reduced urban water use. Environmental Science & Technology Letters, 8(5), pp.431-436.

Zechman Berglund, E., Thelemaque, N., Spearing, L., Faust, K.M., Kaminsky, J., Sela, L., Goharian, E., Abokifa, A., Lee, J., Keck, J. and Giacomoni, M., 2021. Water and wastewater systems and utilities: Challenges and opportunities during the COVID-19 pandemic. Journal of Water Resources Planning and Management, 147(5), p.02521001.

Kim, D., Yim, T. and Lee, J.Y., 2021. Analytical study on changes in domestic hot water use caused by COVID-19 pandemic. Energy, 231, p.120915.

The authors did not acknowledge this scenario at all,, and how the results of the online survey might be different.

It is also not clear how the authors differentiated "public" and "domestic" wastewater streams, and this distinctions could be refined.

I first of all would not change either the conceptual model or the model assumptions. The key issue would be the parameterization, especially with regard to biodegradation rates that are eventually needed. The data and the analyses of Menzies et al shows that in studies performed according to the OECD guidelines on biodegradation testing, and whilst using different analytical methods to confirm degradation in different ratios of test duration, concentrations of microbes, and PVA levels, PVA has the potential of being degraded to a higher extent than assumed in the modelling study of Rolsky and Kelkar. The data and underlying information available does show, however, that it is not straightforward to extrapolate the findings of the ' OECD-conform' degradation testing to realistic conditions in WWTPs or especially in the natural environment. In my opinion, this gap between OECD test results and rate constants in realistic systems (like WWTPs) needs to be filled in before re-parametrizing the model.
It would also be a major step forward when experimental effluent levels of PVA in WWTPs were available.

The Menzies et al (2022) study demonstrates that PVA can fully degrade given enough time and microbial community present, and some of this degradation will also happen in river water, which will impact the PVA in the untreated water modeling part. So, some degradation could be added to the untreated water, realizing that biodegradation is a question of time, so this would need to be taken into account, too. On the treated water side, given the short hydraulic retention time, and the observed lag times in Menzies et al, it is not obvious that major changes to the model should be made to the fraction of PVA that will flow through the WWTP. Arguably, the removal rate in sludge could be increased a little.