Resources

Effects of imposed disturbance and SRT on QQ-based antifouling efficacyFigure 1 shows the fouling rate profiles of MBRs over time under different operating conditions, summarizing the average fouling rates of each phase. Two representative transmembrane pressure (TMP) profiles, the average fouling time, and the number of MBR runs at each phase are also provided in Supplementary Fig. 1 and Supplementary Table 1. In Phase 1, the average fouling rates for Reactors 1 and 2 were nearly the same, i.e., there was no statistically significant difference between the reactors. This confirms that the MBRs were in identical states, in terms of fouling behavior when operated in the conventional mode. In Phase 2, the average fouling rates of Reactors 1 and 2 were also almost the same, showing that the QQ effect on fouling mitigation appears to be insignificant. Unlike previous reports on the QQ effect on antifouling efficacy35, it was unclear whether QQ played a role in fouling control under Phase 2 conditions. One possible reason for the difference is SRT, which will be further investigated and discussed later. After disturbance (2 d starvation with a high shear rate of 103 s−1) was applied to both MBRs at the beginning of Phase 3, both MBRs experienced severe fouling phenomena, with sharp increases in the average fouling rate at >40 kPa/d. No QQ effect was observed in this phase either. The applied disturbance may have caused a drastic change in mixed liquor characteristics, probably including the microbial community structure, while aggravating fouling propensities, which will be discussed further in later sections.Fig. 1: Membrane-fouling rate.a Its variations with time for Reactor 1 (R1) and Reactor 2 (R2). Each data point indicates the average fouling rate for each MBR run. b Average fouling rates with t-test probability (p) values for each phase. Error bars indicate 1 SD. Details on the experimental conditions are provided in Table 2.Full size imageTo examine how SRT affects membrane fouling in the MBRs, we increased SRT from 50 to 75 d in Phase 4. The average fouling rate of Reactor 1 slightly decreased with vacant beads (which contain no QQ bacteria), whereas that of Reactor 2 decreased more significantly with QQ beads (corresponding to 47% of that of Reactor 1). With the longer SRT, it appears that QQ affected biofouling mitigation. When Reactor 1 was switched to conventional mode in Phase 5, its membrane fouling rate was slightly reduced compared to Phase 4. This implies that the vacant beads had no effect on fouling mitigation and in fact may have caused membrane fouling. A previous study also reported that membrane fouling increased when the media added were trapped inside the membrane fibers36. However, Reactor 2 exhibited a notably slower fouling rate, which corresponds to 55% of that of Reactor 1. Thus, the biofouling control due to QQ was evident at long SRT (75 d). The effect of SRT on QQ will be discussed further in later sections, along with the time-series data of MBR operational performance and microbial community.Effects of QQ, disturbance, and SRT on biopolymer productionFigure 2a–d show EPS and SMP variations during MBR operations with and without QQ at different SRT values. The EPS and SMP data normalized to mixed liquor suspended solids (MLSS) are also provided in Supplementary Fig. 2. During Phase 1, when the MBRs were operated in the conventional mode, the EPS-carbohydrate (EPS-C) and EPS-protein (EPS-P) levels were similar in both MBRs (~20 and 80 mg/L, respectively). Notably, however, there was only ~3% probability that the EPS-P level between Reactors 1 and 2 occurs by chance. One possible explanation is that the microbial communities in both the reactors should change as a result of the provision of synthetic wastewater, leading to alterations in metabolic products. A previous study also reported that fluctuations in EPS-P level at the beginning of MBR operation were observed due to bacterial acclimation to new environments13. In Phase 2, the EPS-C concentration decreased by ~13.5% in Reactor 1 compared with that of Phase 1, but decreased by 33.1% in Reactor 2. However, the EPS-P concentrations in both reactors remained virtually unchanged. It seemed that the lower EPS levels with QQ did not virtually contribute to fouling mitigation, possibly because its levels were still too high to make a perceptible reduction in membrane fouling. In Phase 3, the EPS-C concentration in Reactor 1 increased by ~6%, whereas it increased by ~39.3% in Reactor 2. This substantial EPS-C increase may have resulted in severe membrane fouling, even in the presence of QQ beads. A previous study reported that increased EPS production was strongly correlated with environmental stresses such as shear and starvation37. It was thought that the disturbance at the beginning of Phase 3 may have caused a similar phenomenon. When the SRT was increased to 75 d in Phase 4, the EPS levels in the two MBRs decreased. In Reactor 1, EPS-C and EPS-P concentrations declined by ~15.6% and 11%, respectively, whereas their concentrations decreased by ~74.4% and 21.65%, respectively, in Reactor 2. Similarly, previous studies also reported that EPS production was reduced with long SRT values11,38. In Phase 5, EPS-C and EPS-P contents continued to decrease in Reactor 2; however, no further decrease was observed in Reactor 1. The higher EPS content caused preferential attachment of biomass onto the membrane surface so as to form cake layers39. The reduced EPS production associated with the QQ strategy correlates with previous findings35,40. It is thus believed that membrane fouling could be mitigated by the presence of QQ media.Fig. 2: Biopolymer concentrations and mixed liquor characteristics according to phases in the two MBRs.a EPS-carbohydrates (EPS-C); b EPS-proteins (EPS-P); c SMP-carbohydrates (SMP-C); d SMP-proteins (SMP-P); e mixed liquor suspended solids (MLSS); and f floc size. The box is determined by the 25th and 75th percentiles, whereas the whiskers are determined by the 5th and 95th percentiles. The white square symbol inside each bar represents the average value of each parameter.Full size imageThe SMP-carbohydrate (SMP-C) and SMP-protein (SMP-P) levels were also similar in both reactors in Phase 1 (~5 and 4.5 mg/L, respectively). In Phase 2, there were slight changes in both. When disturbance was applied in Phase 3, the SMP levels in both reactors significantly increased. The SMP-C and SMP-P concentrations in Reactor 1 increased by ~45.3% and ~36.1%, respectively, and their respective increases in Reactor 2 were more significant at ~62.4% and ~110%. These results agree well with previous studies, which reported that the disturbance imposed on microorganisms induced the release of microbial polymeric substances28,37, in addition to substrate limitations41. When the SRT was increased to 75 d in Phases 4 and 5, the SMP-C and SMP-P concentrations started decreasing, and the decline was more significant with QQ. For instance, in Phase 5, Reactor 2 had the lowest SMP-C and SMP-P levels, at ~54.2% and ~62.4% lower than these respective values in Phase 4. In addition, the longer SRT contributed to decreased SMP levels when comparing Reactor 2 between Phases 2 and 5. This result is consistent with previous studies35,40, which reported that the presence of QQ media reduced soluble biopolymer contents in MBRs. Another previous study also reported that increasing the SRT in MBRs alleviated biofouling42. Notably, QQ caused the more substantive and immediate decrease of SMP-C than that of SMP-P in Phase 4. This could be associated with the inhibition of protease enzyme secretion in the presence of QQ enzymes leading to reduced degradation of soluble protein43. Overall, it can be concluded that the presence of QQ media reduced biopolymer production more significantly when the SRT was extended.Effects of QQ, disturbance, and SRT on mixed liquor characteristics and biological treatment efficienciesMixed liquor characteristics, such as MLSS and floc size, were monitored over time (Fig. 2e, f). The MLSS concentration in both MBRs from Phases 1–3 varied in the range of 2100–2250 mg/L. Disturbance (starvation with shear) at the beginning of Phase 3 caused a slight decrease in biomass concentration compared to that of Phase 2. At the longer SRT (75 d) in Phases 4 and 5, the MLSS concentration increased to 2650–2900 mg/L. It is natural that a longer SRT should increase MLSS levels at the same yield. It appeared that the MLSS levels were a bit higher with QQ than without it as observed from Phase 2 through 5. Microbial growth can be promoted if QS that requires carbon sources is inhibited. A recent finding pointed out that the QQ enzyme (acylase) may increase microbial yield, converting the resource (food) to more biomass44. Operational parameters such as QQ, SRT, and disturbance did not yield significant changes in floc size during the entire study, although fluctuations were possible45. In this study, there was a slight increase in floc size with QQ in Phase 5. The microbial floc size is a function of several factors, such as QS, QQ, nutrients, and operational conditions43. A recent study reported that there was a negative correlation between floc size and EPS level, because the excessive EPS played a role in reducing the hydrophobicity of flocs and, thereby, weakening the cells’ attachment46. It is thus seen that the reduced EPS content may help enhance the floc aggregation, possibly resulting in greater floc sizes.The biological treatment efficiencies of the two MBRs were evaluated in terms of removals of chemical oxygen demand (COD), total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) (Supplementary Fig. 3a–d). The effects of SRT and QQ on these removal efficiencies were almost negligible, although disturbance caused a slight decrease in organics removal. The result coincided with the increased SMP level with shear in Phase 3. In short, mixed liquor properties and biological treatment performances seemed to tolerate the effects of QQ and SRT, although they were slightly impacted by disturbance.Microbial community structure changeFigure 3a and Supplementary Table 2 show microbial community variations relative to phases, with clear microbial community structure shifts between phases. Two species were dominant in the seed sludge: Dokdonella immobilis (11.31%) and Sphaerotilus natans (14.91%). After inoculation in the laboratory MBRs, the dominance of these species diminished and other species, being adapted to the synthetic feed, flourished instead (Phase 1). In Phase 2, Thiothrix eikelboomii (15%–18%) and Panacibacter ginsenosidivorans (11%–14%), which were negligible in the seed sludge, became dominant in both reactors. In addition, the relative abundances of Kofleria flava and Flavitalea antarctica increased to 7.57% and 6.95%, respectively. These two species were more abundant in Reactor 1 than Reactor 2. Lastly, the major species of the seed sludge, such as D. immobilis, S. natans, and Terrimonas lutea, became minority species ( 0.7) with SMP-C, SMP-P, and the relative abundance of four individual microbial species (i.e., F. antarctica, P. glucosidilyticus, S. piscinae, and T. carbonis). SMP had a lot stronger correlations with fouling rates than EPS, although the actual amounts of the former were a lot smaller than those of the latter. The result indicates that the soluble biopolymers present in the bulk liquid play a more important role in membrane fouling, possibly due to their direct deposition onto the membrane surface. The strong, negative correlations of fouling rates with COD and TOC removal efficiencies support the above explanation. In particular, P. glucosidilyticus and S. piscinae, which had the highly strong correlations with membrane fouling (r  > 0.87), accordingly exhibited strong correlations with SMP. Notably, T. eikelboomii, which was the most abundant in Reactor 1 of Phases 4 and 5, had a relatively weak negative correlations (r = −0.32) with membrane fouling and so not as strong as did K. flava (r = −0.73). The decrease in the relative abundance of T. eikelboomii in Reactor 2 of Phase 5 should be associated with QQ, but the species might still have been contributing to membrane fouling, as discussed above. As expected, the microbial diversity indices between OTUs and Chao1, as well as Shannon and inverse Simpson, were found to be strongly correlated. However, the fouling rate did not have strong correlations with any of the microbial diversity indices (−0.05 ≤ r ≤ 0.55), although the microbial diversity was always higher in the presence of QQ (see Table 1).Fig. 4: Correlation analysis.Spearman’s correlation coefficients between all the MBR parameters determined in the study.Full size imageOn the other hand, the content of EPS-C and EPS-P had strong, negative correlations with MLSS levels. The biomass increase was accompanied with longer SRT, so the aged sludge produced less EPS amounts leading to the floc size decline (corresponding to a negative correlation, i.e., r = −0.52). Notably, the floc size had a strong, positive correlation (r = 0.84) with TN removal, suggesting that simultaneous nitrification and denitrification possibly occurred with larger biological flocs59. In addition, D. immobilis showed a strong positive correlation (r = 0.84) with TP removal, proposing its role as a potential phosphate uptake strain60. Overall, the relationships between MBR parameters (e.g., fouling rates, mixed liquor characteristics, biological treatment efficiencies, and microbial species dominance) helped better understand the fouling patterns and biological performances in the MBRs with and without QQ.In summary, the QQ effect on MBR antifouling efficacy was clearer when the SRT was extended from 50 to 75 d, although the disturbance (starvation with shear) aggravated membrane fouling, which counteracted the positive QQ effect. QQ yielded a significant biopolymer production decrease with the longer SRT. Accordingly, organic substance removal showed relatively strong, negative correlations with MBR-fouling propensity. MBR microbial communities showed dynamic responses to the feed change, QQ, disturbance, and SRT. With disturbance, F. antarctica, S. piscinae, and P. glucosidilyticus dominated the microbial community leading to substantive membrane fouling. However, the microbial community balance between T. eikelboomii and K. flava, whose relative abundances appeared to be affected by SRT and QQ, played a key role in fouling propensity under stabilized conditions. The correlation analysis showed strong positive relationships between membrane fouling rate and the abundance of several microbial species (F. antarctica, P. glucosidilyticus, S. piscinae, and T. carbonis). However, there was no strong correlation between T. eikelboomii and membrane fouling propensity, possibly due to the antagonism by K. flava, and vice versa. More

1.
The price of fast fashion. Nat. Clim. Change 8, 1 (2018).
2.
Shirvanimoghaddam, K., Motamed, B., Ramakrishna, S. & Naebe, M. Death by waste: fashion and textile circular economy case. Sci. Total Environ. 718, 137317 (2020).
CAS  Article  Google Scholar 

3.
Boriskina, S. V. An ode to polyethylene. MRS Energy Sustain. 6, E14 (2019).
Article  Google Scholar 

4.
Grigore, M. Methods of recycling, properties and applications of recycled thermoplastic polymers. Recycling 2, 24 (2017).
Article  Google Scholar 

5.
Ragaert, K., Delva, L. & Van Geem, K. Mechanical and chemical recycling of solid plastic waste. Waste Manag. 69, 24–58 (2017).
CAS  Article  Google Scholar 

6.
Zhang, Z. et al. Recovering waste plastics using shape-selective nano-scale reactors as catalysts. Nat. Sustain. 2, 39–42 (2019).
Article  Google Scholar 

7.
Tong, J. K. et al. Infrared-transparent visible-opaque fabrics for wearable personal thermal management. ACS Photonics 2, 769–778 (2015).
CAS  Article  Google Scholar 

8.
Hsu, P.-C. et al. Radiative human body cooling by nanoporous polyethylene textile. Science 353, 1019–1023 (2016).
CAS  Article  Google Scholar 

9.
Boriskina, S. V. Nanoporous fabrics could keep you cool. Science 353, 986–987 (2016).
CAS  Article  Google Scholar 

10.
Peng, Y. et al. Nanoporous polyethylene microfibres for large-scale radiative cooling fabric. Nat. Sustain. 1, 105–112 (2018).
Article  Google Scholar 

11.
Boriskina, S. V., Zandavi, H., Song, B., Huang, Y. & Chen, G. Heat is the new light. Opt. Photonics News 28, 26–33 (2017).
Article  Google Scholar 

12.
Higg Materials Sustainability Index (Higg Co., accessed 27 October 2020); https://msi.higg.org/

13.
Muthu, S. S., Li, Y., Hu, J. Y. & Mok, P. Y. Recyclability potential index (RPI): the concept and quantification of RPI for textile fibres. Ecol. Indic. 18, 58–62 (2012).
CAS  Article  Google Scholar 

14.
Allwood, J. M., Laursen, S. E., Rodríguez, C. M. de & Bocken, N. M. P. Well Dressed? The Present and Future Sustainability of Clothing and Textiles in the United Kingdom (Cambridge Univ. Press, 2006).

15.
van der Velden, N. M., Kuusk, K. & Köhler, A. R. Life cycle assessment and eco-design of smart textiles: the importance of material selection demonstrated through e-textile product redesign. Mater. Des. 84, 313–324 (2015).
Article  Google Scholar 

16.
Steinberger, J. K., Friot, D., Jolliet, O. & Erkman, S. A spatially explicit life cycle inventory of the global textile chain. Int. J. Life Cycle Assess. 14, 443–455 (2009).
CAS  Article  Google Scholar 

17.
Shimel, M. et al. Enhancement of wetting and mechanical properties of UHMWPE-based composites through alumina atomic layer deposition. Adv. Mater. Interfaces 5, 14 (2018).
Article  Google Scholar 

18.
Yousif, E. & Haddad, R. Photodegradation and photostabilization of polymers, especially polystyrene: review. SpringerPlus 2, 398 (2013).
Article  Google Scholar 

19.
Hawkins, W. L. in Polymer Degradation and Stabilization (ed. Harwood, H. J.) 3–34 (Springer, 1984).

20.
Abusrafa, A. E., Habib, S., Krupa, I., Ouederni, M. & Popelka, A. Modification of polyethylene by RF plasma in different/mixture gases. Coatings 9, 145 (2019).
Article  Google Scholar 

21.
Princen, H. M. Capillary phenomena in assemblies of parallel cylinders. II. Liquid columns between horizontal parallel cylinders. J. Colloid Interface Sci. 30, 359–371 (1969).
Article  Google Scholar 

22.
Zhang, J. & Han, Y. Shape-gradient composite surfaces: water droplets move uphill. Langmuir 23, 6136–6141 (2007).
CAS  Article  Google Scholar 

23.
Wallenberger, F. T. The effect of absorbed water on the properties of cotton and fibers from hydrophilic polyester block copolymers. Text. Res. J. 48, 577–581 (1978).
CAS  Article  Google Scholar 

24.
Shamey, R. in Polyolefin Fibres: Structure, Properties and Industrial Applications (ed. Ugbolue, S. C. O.) 359–388 (Woodhead Publishing, 2017).

25.
Lozano, L. M. et al. Optical engineering of polymer materials and composites for simultaneous color and thermal management. Opt. Mater. Express 9, 1990–2005 (2019).
Article  Google Scholar 

26.
Cai, L. et al. Temperature regulation in colored infrared-transparent polyethylene textiles. Joule 3, 1478–1486 (2019).
CAS  Article  Google Scholar 

27.
Workman, J. J. Jr in Encyclopedia of Analytical Chemistry (eds Meyers, R. A. & Provder, T.) https://doi.org/10.1002/9780470027318.a2021 (John Wiley & Sons, Ltd, 2006).

28.
Geyer, R., Jambeck, J. R. & Law, K. L. Production, use, and fate of all plastics ever made. Sci. Adv. 3, e1700782 (2017).
Article  Google Scholar 

29.
A New Textiles Economy: Redesigning Fashion’s Future (Ellen Macarthur Foundation, 2017); https://www.ellenmacarthurfoundation.org/publications/a-new-textiles-economy-redesigning-fashions-future

30.
Bisinella, V., Albizzati, P. F., Astrup, T. F. & Damgaard, A. Life Cycle Assessment of Grocery Carrier Bags (Danish Environmental Protection Agency, 2018); https://www.researchgate.net/publication/326735612_Life_Cycle_Assessment_of_grocery_carrier_bags

31.
Ni, G. W. et al. A salt-rejecting floating solar still for low-cost desalination. Energy Environ. Sci. 11, 1510–1519 (2018).
CAS  Article  Google Scholar 

32.
Alberghini, M. et al. Multistage and passive cooling process driven by salinity difference. Sci. Adv. 6, eaax5015 (2020).
CAS  Article  Google Scholar 

33.
Lal Basediya, A., Samuel, D. V. K. & Beera, V. Evaporative cooling system for storage of fruits and vegetables – a review. J. Food Sci. Technol. 50, 429–442 (2013).
Article  Google Scholar 

34.
McLain, V. C. Final report on the safety assessment of polyethylene. Int. J. Toxicol. 26, 115–127 (2007).
Google Scholar 

35.
Suhardi, V. J. et al. A fully functional drug-eluting joint implant. Nat. Biomed. Eng. 1, 0080 (2017).

36.
Halden, R. U. Plastics and health risks. Annu. Rev. Public Health 31, 179–194 (2010).
Article  Google Scholar 

37.
Terinte, N., Manda, B. M. K., Taylor, J., Schuster, K. C. & Patel, M. K. Environmental assessment of coloured fabrics and opportunities for value creation: spin-dyeing versus conventional dyeing of modal fabrics. J. Clean. Prod. 72, 127–138 (2014).
Article  Google Scholar 

38.
Bombgardner, M. Greener textile dyeing. C&EN Glob. Enterp. 96, 28–33 (2018).

39.
Carroll, B. J. Accurate measurement of contact angle, phase contact areas, drop volume, and Laplace excess pressure in drop-on-fibre systems. J. Colloid Interface Sci. 57, 488–495 (1976).
CAS  Article  Google Scholar 

40.
Kralchevsky, P. A., Paunov, V. N., Ivanov, I. B. & Nagayama, K. Capillary meniscus interaction between colloidal particles attached to a liquid–fluid interface. J. Colloid Interface Sci. 151, 79–94 (1992).
Article  Google Scholar 

41.
Masoodi, R. & Pillai, K. M. Wicking in Porous Materials: Traditional and Modern Modeling Approaches (CRC Press and Taylor & Francis, 2012).

42.
Princen, H. M. Capillary phenomena in assemblies of parallel cylinders. I. Capillary rise between two cylinders. J. Colloid Interface Sci. 30, 69–75 (1969).

43.
R Core Team R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2019). More

Shape analysis of 3038 daily flow patterns (2016–2020) for WWTP1 and WWTP2
An example effluent flow pattern for a 10-day period at WWTP1 is shown in Fig. 1. EDM detected spilling intervals are overlaid to demonstrate the flattening effect of spilling on the profile of the flow pattern.
Fig. 1: WWTP1: example effluent flow pattern for 10 days annotated with EDM confirmed spilling intervals.

A 24-h (midnight to midnight) daily flow pattern of 96 15-min-interval average flow rates (litres/second) of treated effluent is shown in blue. The black horizontal, linear annotations represent EDM recorded intervals denoting a discharge from a storm tank (i.e., consented spill or potentially unconsented spill of untreated sewage), the shortest being 15 min and the longest over 24-h. Total daily rainfall (mm/d) is provided in green. The first two days, with no detected spills, show diurnal patterns of low flow between midnight (previous day) and the first peak after mid-morning, followed by a lull until a second, smaller peak in the evening. The next seven days (15/12/18 through 21/12/18) involve spill intervals of various length (black EDM line), showing a flattening of flow, which is typical of storm discharge during heavy rainfall. The last day shows elevated flows and a partial return to a diurnal flow pattern with no spills reported.

Full size image

For WWTP1 (resp. WWTP2), EDM data were available for 446 (resp. 471) consecutive days for 2018–2020 during which untreated sewage spill intervals of varying lengths had been recorded. For each day, spill intervals were aggregated to the total number of hours of discharge. Of the days used for machine learning for WWTP1 (resp. WWTP2), 339 (resp. 346) involved no EDM recorded spilling incidents and 107 (resp. 125) days had spills with various lengths of which over a third were for 24-h. For WWTP1 (resp. WWTP2), 97 (resp. 117) days with an aggregated ‘spill’ length of at least 3-h were labelled as ‘spill’ and 349 (resp. 354) with an aggregated ‘spill’ length of below 3-h as ‘normal’. A 3-h aggregation period was selected because it guaranteed a reasonable number of ‘spill’ days on which to base the supervised learning and preliminary attempts to predict spilling hours per day were weakest for aggregated daily spills under 3-h. Where no EDM data was available, days were labelled as ‘unknown’. The average ‘normal’ (blue line) and ‘spill’ (black line) daily flow patterns as a proportion of storm overflow rates (red line) are shown in Fig. 2 for each WWTP. The storm overflow rates mark the minimum flow that should be treated before untreated sewage spills can be made in compliance with EA permits to discharge to watercourses.
Fig. 2: Average daily flow patterns.

a WWTP1: black curve for ‘spill’ days (n = 97) and blue curve for ‘normal’ days (n = 349); b WWTP2: black curve for ‘spill’ days (n = 117) and blue curve for ‘normal’ days (n = 354).

Full size image

Separate shape models were generated for flow patterns from 2016 to 2020 for WWTP1 (n = 1511) and WWTP2 (n = 1527). The first principle component of shape variation, PCA1, in both models, is associated with magnitude, and temporal shifting of morning flow peak (see Supplementary Video 1.mp4) as well as “seasonal” changes related to daylight saving, public holidays and vacation periods (Supplementary Fig. 1). Despite differences in the population served by WWTP1 and WWTP2, Fig. 3a, b shows similar distributions for scatter plots of PCA1 vs PCA2 for 2121 flows for 2016–2018 without EDM data. Analogous plots of PCA1 vs PCA2 for 917 flows for 2018–2020 with EDM data (Fig. 3c, d) suggest that, for both WWTPs, PCA2 is correlated with shape difference between ‘normal’ flow (open circles) and ‘spill’ affected flow (filled triangles). This spill-related flattening is illustrated by morphing the overall average daily flow pattern for WWTP1 between −1 and +1 standard deviations of PCA2 (Supplementary Video 2.mp4). Interestingly, the area under the receiver-operating characteristics curve associated with using PCA2 alone for ‘normal’/’spill’ discrimination is 0.88 and 0.91 for WTTP1 and WWTP2, respectively (this is the estimated probability of correctly classifying a pair of flow patterns selected randomly, one each, from the ‘normal’ and ‘spill’ labelled subsets).
Fig. 3: PCA1 vs PCA2 for daily flow patterns.

Unknown spill status (grey filled circles); spill confirmed by EDM (filled black triangles); confirmed as normal by EDM (unfilled grey circle) 2016–2018 without EDM data a WWTP1 (n = 1065); b WWTP2 (n = 1056); 2018–2020 with EDM data c WWTP1 (n = 466); d WWTP2 (n = 471).

Full size image

Supervised learning of the effect of sewage spills on 917 effluent flow patterns
The performance of 20-folded cross-validation of supervised learning for labelled flow patterns for WWTP1 and WTTP2 is shown in Supplementary Tables 3 and 4 for 20 support vector machine (SVM) variations while retaining up to 15 PCA modes for flow pattern synthesis. The number of PCA modes retained for shape synthesis affects the validity of the reconstruction of each daily flow pattern and hence classification accuracy. For the three best-performing algorithms, Supplementary Fig. 2 shows the variation in classification accuracy of daily flow patterns for different numbers of retained PCA modes estimated as the average area under the 20 receiver-operating characteristic curves associated with the cross-validation folds. For the optimal classifiers, the average area under the receiver-operating characteristic curve was 0.97 for WTTP1 and 0.96 for WWTP2.
For verification, prior to wider application, the optimal ML classifiers defined for each WWTP were used to reclassify the flow patterns used in their derivation. Figure 4 shows these flow patterns in contiguous temporal sequence with annotations for each day reflecting EDM detected spill intervals (horizontal black segments) and ML confirmation of ‘spill’ (unfilled gold circles). During this period there were 97 (resp. 117) days with an EDM confirmed aggregated spill of at least 3-h at WWTP1 (resp. WWTP2). The agreement between optimal ML classification and spill day labels derived from EDM data was extremely high (WTTP1: sensitivity = 0.91, specificity = 0.95; WTTP2: sensitivity = 0.98, specificity = 0.98), as would be expected for such “training” data.
Fig. 4: Daily effluent flow patterns and event duration monitor (EDM) detected spill intervals at WTTP1 and WWTP2 used as training data (Dec’2018–Mar’2020).

The daily flow and EDM spill data are measured at 15 min intervals. Flow is coloured (orange/blue/pink) to distinguish different years. Black horizontal lines delimit EDM detected spill intervals. Daily flows of aggregated spill length of at least/less than 3-h are labelled as ‘spill’/‘normal’ prior to the supervised learning. Gold circles indicate days classified as ‘spill’ following the training of the machine learning (ML) algorithms to produce an optimal classifier for each WWTP. The grey dashed line represents the storm overflow which defines the minimum sewage flow that should be treated even during storm filling or overflow. Additional annotations are telemetry alarms provided by the operator. These alarms have the potential to corroborate ML predictions of ‘spill’ days for the unseen flow patterns from 2009 to 2018 for which there is no EDM data. Similar charts showing the unseen ML classification of the 2009–2018 daily flow patterns overlaid with rainfall and river level data are provided in Supplementary Figs 5–10.

Full size image

Figure 4 also includes data from other alarms related to untreated sewage discharges that have the potential to corroborate ML flow pattern classification for historical periods without EDM data. For WWTP1, there is near-perfect agreement (Cohen’s kappa: 0.81–1.00) between the EDM, STO (Storm Tank Overflow) and COL (Consented Overflow Level) alarms and ML classification for Feb ‘19–Feb ‘20 (Fig. 4 and Table 2). For just two months, Dec ‘18 and Jan ‘19, the EDM and COL devices concur with near-perfect agreement (Cohen’s kappa = 0.95), the STO device was largely at odds (Cohen’s kappa ≤ 0), and the ML classifier flagged incidents detected by all three. These results suggest that the STO is a good candidate and the COL alarm is an excellent candidate for corroborating ML detected putative spills at WWTP1 when EDM data is unavailable.
Table 2 Agreement of ML classification, EDM, COL and STO alarms for the supervised learning.
Full size table

For WWTP2, there is almost perfect agreement between EDM and COL alarms (Cohen’s kappa = 0.87) and with ML classification (Cohen’s kappa = 0.78) (Fig. 4 and Table 2). No STO alarm data were provided for 2020 and between Dec ‘18 and Dec ‘19 STO showed only chance agreement with other devices and the ML classifier (Cohen’s kappa  More

1.
Wang, W. X. Interaction of trace metals and different marine food chains. Mar. Ecol. Prog. Ser. 243, 295–309 (2002).
CAS  Article  Google Scholar 
2.
Dautremepuits, C., Paris-Palacios, S., Betoulle, S. & Vernet, G. Modulation in hepatic and head kidney parameters of carp (Cyprinus carpio L.) induced by copper and chitosan. Comp. Biochem. Physiol. Part C 137, 325–333 (2004).
Google Scholar 

3.
Kalay, M., Ay, P. & Canil, M. Heavy metal concentration in fish tissues from the northeast Mediterranean Sea. Bull. Environ. Contam. Toxicol. 63, 673–671 (1999).
CAS  Article  Google Scholar 

4.
Ashraf, W. Accumulation of heavy metals in kidney and heart tissues of Epinephelus microdon fish from the Arabian Gulf. Environ. Monit. Assess. 101, 311–316 (2005).
CAS  Article  Google Scholar 

5.
Khaled Abdel-Halim, Y. Biomarkers in ecotoxicological research trails. J. Forensic Sci. Toxicol. 1, 1005 (2018).
Google Scholar 

6.
Javed, M., Ahmad, M. I., Usmani, N. & Ahmad, M. Multiple biomarker responses (serum biochemistry, oxidative stress, genotoxicity and histopathology) in Channa punctatus exposed to heavy metal loaded waste water. Sci. Rep. 7, 1675 (2017).
Article  CAS  Google Scholar 

7.
WHO/IPCS. Environmental Health Criteria 155, Biomarkers and Risk Assessment: Concepts and Principles (IPCS, World Health Organization, 1993).

8.
Varô, I., Navarro, J. C., Amat, F. & Guilhermino, L. Characterization of cholinesterase and evaluation of the inhibitory potential of chlorpyrifos and dichlorvos to Artemia salina and Artemia parthenogenetica. Chemos 48, 563–569 (2001).
Article  Google Scholar 

9.
Eggen, R. I., Behra, R., Burkhardt-Holm, P., Escher, B. I. & Schweigert, N. Challenges in ecotoxicology. Environ. Sci. Technol. 38, 58–64 (2004).
Article  Google Scholar 

10.
Moore, M. N., Depledge, M. H., Readman, J. W. & Paul Leonard, D. R. An integrated biomarker-based strategy for ecotoxicological evaluation of risk in environmental management. Mutat. Res. 552, 247–268 (2004).
CAS  Article  Google Scholar 

11.
Ferrando, S. et al. Gut morphology and metallothionein immunoreactivity in Liza aurata from different heavy metal polluted environments. Ital. J. Zool. 73, 7–14 (2006).
CAS  Article  Google Scholar 

12.
Au, D. W. The application of histo-cytopathological biomarkers in marine pollution monitoring: a review. Mar. Pollut. Bull. 48, 817–834 (2004).
CAS  Article  Google Scholar 

13.
Van der Van der Oost, R., Beyer, J. & Vermeulen, N. P. E. Fish bioaccumulation and biomarkers in environmental risk assessment: a review. Environ. Toxicol. Pharmacol. 13, 57–149 (2003).
Article  Google Scholar 

14.
Turan, F., Eken, M., Ozyilmaz, G., Karan, S. & Uluca, H. Heavy metal bioaccumulation, oxidative stress and genotoxicity in African catfish Clarias gariepinus from Orontes river. Ecotoxicology 29, 1522–1537 (2020).
CAS  Article  Google Scholar 

15.
Maurya, P. K. et al. Bioaccumulation and potential sources of heavy metal contamination in fish species in River Ganga basin: possible human health risks evaluation. Toxicol. Rep. 6, 472–481 (2019).
CAS  Article  Google Scholar 

16.
Alshkarchy, S. S., Raesen, A. K. & Najim, S. M. Physiological effect of some metals on blood parameters of common carp Cyprinus carpio, reared in cages and wild in the Euphrates river, Babil, Iraq. Life Sci. Arch. 6(1932), 1907 (2020).
Google Scholar 

17.
Dane, H. & Sisman, T. Effects of heavy metal pollution on hepatosomatic ındex and vital organ histology in Alburnus mossulensis from Karasu River. Turk. J. Vet. Anim. Sci. 44, 607–617 (2020).
CAS  Google Scholar 

18.
Javed, M., Ahmad, I., Usmani, N. & Ahmad, M. Bioaccumulation, oxidative stress and genotoxicity in fish (Channa punctatus) exposed to a thermal power plant effluent. Ecotoxicol. Environ. Saf. 127, 163–169 (2016).
CAS  Article  Google Scholar 

19.
Stern, B. R. Essentiality and toxicity in copper health risk assessment: overview, update, and regulatory considerations. Toxicol. Environ. Health 73, 114–127 (2010).
CAS  Article  Google Scholar 

20.
Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile for Copper (Centers for Disease Control, 2002).

21.
Kumar, D., Malik, D. S. & Gupta, V. Fish metallothionein gene expression: a good bioindicator for assessment of heavy metal pollution in aquatic ecosystem. Int. Res. J. Environ. Sci. 6, 58–62 (2017).
CAS  Google Scholar 

22.
Irato, P., Santovito, G., Piccinni, E. & Albergoni, V. Oxidative burst and metallothionein as a scavenger in macrophages. Immunol. Cell Biol. 79, 251–254 (2001).
CAS  Article  Google Scholar 

23.
Kumar, N. et al. Validation of growth enhancing, immunostimulatory and disease resistance properties of Achyranthes aspera in Labeo rohita fry in pond conditions. Heliy 5, e01246 (2019).
Article  Google Scholar 

24.
Chakrabarti, R. et al. Effect of seeds of Achyranthes aspera on the immune responses and expression of some immune-related genes in carp Catla catla. Fish Shellfish Immunol. 41, 64–69 (2014).
CAS  Article  Google Scholar 

25.
BIS (Bureau of Indian Standard). Drinking Water Specification IS: 10500: 1992 (BIS New Delhi, 1992).

26.
UNEPGEMS (United Nations Environment Programme Global Environment Monitoring System/Water Programme). Adapted for Water Quality and Ecosystem Health (UNEPGEMS (United Nations Environment Programme Global Environment Monitoring System/Water Programme), 2006).

27.
Pengfei, L. et al. Heavy metal bioaccumulation and health azard assessment for three fish species from Nansi Lake, China. Bull. Environ. Contam. Toxicol. 94, 431–436 (2015).
Article  CAS  Google Scholar 

28.
Ahmed, A. S. S. et al. Bioaccumulation of heavy metals in some commercially important fishes from a tropical river estuary suggests higher potential health risk in children than adults. PLoS ONE 14, e0219336 (2019).
CAS  Article  Google Scholar 

29.
Fernandes, C. et al. Heavy metals in water, sediment and tissues of Liza saliens from Esmoriz-Paramos lagoon Port. Environ. Monit. Assess. 136, 267–275 (2008).
CAS  Article  Google Scholar 

30.
Islam, A. et al. Assessment of heavy metals concentration in water and tengra fish (Mystus vittatus) of Surma River in Sylhet region of Bangladesh. Arch. Agric. Environ. Sci. 4, 151–156 (2019).
Article  Google Scholar 

31.
Carvalho, C. D. S. & Fernandes, M. N. Effects of copper toxicity at different pH and temperatures on the in vitro enzyme activity in blood and liver of fish, Prochilodus lineatus. Mol. Biol. Rep. 46, 4933–4942 (2019).
Article  CAS  Google Scholar 

32.
Barisic, J., Cannon, S. & Quinn, B. Cumulative impact of anti-sea lice treatment (azamethiphos) on health status of Rainbow trout (Oncorhynchus mykiss, Walbaum 1792) in aquaculture. Sci. Rep. 9, 16217 (2019).
Article  CAS  Google Scholar 

33.
Sokmen, B. B., Tunali, S. & Yanardag, R. Effects of vitamin U (S-methyl methionine sulphonium chloride) on valproic acid induced liver injury in rats. Food Chem. Toxicol. 50, 3562–3566 (2012).
CAS  Article  Google Scholar 

34.
Oztopuz, O. et al. Melatonin ameliorates sodium valproate-induced hepatotoxicity in rats. Mol. Biol. Rep. 47, 317–325 (2019).
Article  CAS  Google Scholar 

35.
Zorriehzahra, M. J., Hassan, M. D., Gholizadeh, M. & Saidi, A. A. Study of some hematological and biochemical parameters of Rainbow trout (Oncorhynchus mykiss) fry in western part of Mazandaran province, Iran. Iranian. J. Fish. Sci. 9, 185–198 (2010).
Google Scholar 

36.
Parvathi, K., Palanivel, S., Mathan, R. & Sarasu, Sublethal effects of chromium on some biochemical profiles of the fresh water teleost, Cyprinus carpio. Int. J. Appl. Biol. Pharm. Technol. 2, 295–300 (2011).
Google Scholar 

37.
Luckoor, P., Salehi, M. & Kunadu, A. Exceptionally high creatine kinase (CK) levels in multicausal and complicated rhabdomyolysis: a case report. Am. J. Case Rep. 18, 746–749 (2017).
Article  Google Scholar 

38.
Oitani, Y. et al. Interpretation of acid α-glucosidase activity in creatine kinase elevation: a case of Becker muscular dystrophy. Brain Dev. 40, 837–840 (2018).
Article  Google Scholar 

39.
Javed, M. & Usmani, N. Stress response of biomolecules (carbohydrate, protein and lipid profiles) in fish Channa punctatus inhabiting river polluted by Thermal Power Plant effluent. Saudi J. Biol. Sci. 22, 237–242 (2015).
CAS  Article  Google Scholar 

40.
Palanisamy, P. G., Sasikala, D., Mallikaraj, N. B. & Natarajan, G. M. Electroplating industrial effluent chromium induced changes in carbohydrates metabolism in air breathing cat fish Mystus cavasius (Ham). Asian J. Exp. Biol. Sci. 2, 521–524 (2011).
CAS  Google Scholar 

41.
Parvathi, J. & Karemungikar, A. Leucocyte variation, an insight of host defenses during hymenolepiasis and restoration with praziquantel. Indian J. Pharm. Sci. 73, 76–79 (2011).
CAS  Article  Google Scholar 

42.
Sharma, J. & Langer, S. Effect of manganese on haematological parameters of fish, Garra gotyla gotyla. J. Entomol. Zool. Stud. 2 3, 77–81 (2014).
CAS  Google Scholar 

43.
Gupta, N. Trypanosome Parasites of Some Fishes of Aligarh. Ph.D. Thesis, Aligarh Muslim University (1981).

44.
Lawrence, D. A. Heavy metal modulation of lymphocyte activities: I. In vitro effects of heavy metals on primary humoral immune responses. Toxicol. Appl. Pharmacol. 57, 439–451 (1981).
CAS  Article  Google Scholar 

45.
Dalmo, R. A., Ingebrigtsen, K. & Bøgwald, J. Non-specific defence mechanisms in fish, with particular reference to the reticuloendothelial system (RES). J. Fish. Dis. 20, 241–273 (1997).
CAS  Article  Google Scholar 

46.
Secombes, C. J. The nonspecific immune system: cellular defences. In The Fish Immune System: Organism, Pathogen and Environment (eds, Iwama, G., Nakanishi, N.) 63–103 (Academic Press Inc., 1996).

47.
Sakai, M., Taniguchi, K., Mamoto, K., Ogawa, H. & Tabata, M. Immunostimulant effect of nucleotide isolated from yeast RNA on carp, Cyprinus carpio L. J. Fish. Dis. 24, 33–38 (2001).
Article  Google Scholar 

48.
Rombout, J. H. W. M., Huttenhuis, H. B. T., Picchietti, S. & Scapigliati, G. Phylogeny and ontogeny of fish leucocytes. Fish. Shellfish Immunol. 19, 441–455 (2005).
CAS  Article  Google Scholar 

49.
Kono, Y. & Fridovich, I. Superoxide radicals inhibit catalase. J. Biol. Chem. 257, 5751–5754 (1982).
CAS  Article  Google Scholar 

50.
Ahmad, Z. et al. Accumulations of heavy metals in the fish Orecochromis niloticus, and Poecilia latipinna and their concentration in water and sediment of Dam Lake of Wadi Namar, Saudi Arabia. J. Environ. Biol. 36, 295–299 (2015).
Google Scholar 

51.
Ameur, W. B. et al. Oxidative stress, genotoxicity and histopathology biomarker responses in Mugil cephalus and Dicentrarchus labrax gill exposed to persistent pollutants. A field study in the Bizerte Lagoon: Tunisia. Chemosphere 135, 67–74 (2015).
CAS  Article  Google Scholar 

52.
Livingstone, D. R. Oxidative stress in aquatic organisms in relation to pollution and agriculture. Rev. Vet. 154, 427–430 (2003).
CAS  Google Scholar 

53.
Hermenean, A. et al. Histopathological alterations and oxidative stress in liver and kidney of Leuciscus cephalus following exposure to heavy metals in the Tur River, North Western Romania. Ecotoxicol. Environ. Saf. 119, 198–205 (2015).
CAS  Article  Google Scholar 

54.
Lopez, E. L., Sedeño-Díaz, JacintoElías, Claudia, S. & Liliana, F. Responses of antioxidant enzymes, lipid peroxidation, and Naþ/Kþ-ATPase in liver of the fish Goodea atripinnis exposed to Lake Yuriria water. Fish. Physiol. Biochem. 37, 511–522 (2011).
Article  CAS  Google Scholar 

55.
Francisco, P. et al. Oxidative stress responses and histological hepatic alterations in barbel, Barbus bocagei, from Vizela river, Portugal. Rev. Int. Contam. Ambie. 29, 29–38 (2013).
Google Scholar 

56.
Sunjog, K. et al. Heavy metal accumulation and the genotoxicity in barbel (Barbus barbus) as indicators of the Danube River pollution. Sci. World J. 2012:351074, 1–6 (2012).
Article  CAS  Google Scholar 

57.
Ahmed, M. K. et al. Genetic damage induced by lead chloride in different tissues of fresh water climbing perch Anabas testudineus (Bloch). Environ. Monit. Assess. 182, 197–204 (2011).
CAS  Article  Google Scholar 

58.
Ameur, W. B. et al. Oxidative stress, genotoxicity and histopathology biomarker responses in mullet (Mugil cephalus) and sea bass (Dicentrarchus labrax) liver from Bizerte Lagoon (Tunisia). Mar. Poll. Bull. 64, 241–251 (2012).
Article  CAS  Google Scholar 

59.
Romeo, M., & Giamberini, L. History of biomarkers. In Ecological Biomarkers, Indicators of Ecotoxicological Effects (eds Amiard-Triquet, C., Amiard, J. C., Rainboe, P. S.) (CRC Press Taylor and Francis Group, 2013).

60.
Javed, H. et al. Efficacy of engineered GO Amberlite XAD-16 picolylamine sorbent for the trace determination of Pb (II) and Cu (II) in fishes by solid phase extraction column coupled with inductively coupled plasma optical emission spectrometry. Sci. Rep. 8, 17560 (2018).
Article  CAS  Google Scholar 

61.
American Public Health Association (APHA), Standard Methods for the Examination of Water and Wastewater Analysis, 21st 442 edn. 289 (American Water Works Association/Water Environment Federation, 2005).

62.
Marklund, S. & Marklund, G. Involvement of the superoxide anion radical in the autoxidation of pyrogallol and a convenient assay for superoxide dismutase. Eur. J. Biochem. 47, 469–474 (1974).
CAS  Article  Google Scholar 

63.
Aebi, H. Catalase in vitro. Methods Enzymol. 105, 121–126 (1984).
CAS  Article  Google Scholar 

64.
Buege, J. A. & Aust, S. D. Microsomal lipid peroxidation. Methods Enzymol. 52, 302–310 (1978).
CAS  Article  Google Scholar 

65.
Jollow, D. J. et al. Bromobenzene-induced liver necrosis. Protective role of glutathione and evidence for 3,4-bromobenzene oxide as the hepatotoxic metabolite. Pharmacology 11, 151–169 (1974).
CAS  Article  Google Scholar 

66.
Habig, W. H., Pabst, M. J. & Jakoby, W. B. Glutathione S-transferases. The first enzymatic step in mercapturic acid formation. J. Biol. Chem. 249, 7130–7139 (1974).
CAS  Article  Google Scholar 

67.
Singh, N. P. et al. A simple technique for quantitation of low levels of DNA damage in individual cells. Exp. Cell Res. 175, 184–191 (1988).
CAS  Article  Google Scholar 

68.
Yildiz, S., Gurcu, B., Basimoglu, Y. K. & Koca, S. Histopathological and genotoxic effects of pollution on Anguilla anguilla in the Gediz River (Turkey). J. Anim. Vet. Adv. 9, 2890–2899 (2010).
CAS  Article  Google Scholar 

69.
Nhiwatiwa, T., Barson, M., Harrison, A. P., Utete, B. & Cooper, R. G. Metal concentrations in water, sediment and sharptooth catfish Clarias gariepinus from three peri-urban rivers in the upper Manyame catchment, Zimbabwe. Afr. J. Aquat. Sci. 36, 243–252 (2011).
CAS  Article  Google Scholar 

70.
Ankur, K., Siddiqui, N. A. & Gautam, A. Assessment of heavy metals and their interrelationships with some physicochemical parameters in ecoefficient rivers of Himalayan Region. Environ. Monit. Assess. 185, 2553–2563 (2013).
Article  CAS  Google Scholar 

71.
Neeratanaphan, L. et al. Genotoxicity and oxidative stress in experimental hybrid catfish exposed to heavy metals in a municipal landfill reservoir. Int. J. Environ. Res. Public Health 17, 1980 (2020).
CAS  Article  Google Scholar  More

In Lake Hallwil, the contribution of oxic methanogenesis to overall diffusive CH4 emissions has been estimated to be 90%6 or 63–83%5, but we show here that NOMC ~ 17%.
In the mass balance of the SML extending from 0 to 5 m water depth5,6, Günthel et al.5 used an average sediment flux of Fsed = 1.75 mmol m−2 day−1, averaging flux estimates of Donis et al.6 from two sediment cores, one collected at 3 m and the other at 7 m water depth. The δ13C of the CH4 in the pore water of these two cores differ substantially6, indicating differences in production and oxidation of CH4 between the sediments in the SML and at 7 m water depth. The estimate of Fsed in the SML should therefore be based on the core collected at 3 m water depth. Using the approach of Donis et al.6, the correct Fsed derived from the data of this core is Fsed = 2.8 mmol m−2 day−1 (Peeters et al.9, see Supplementary Note 2.1 for details).
Günthel et al.5 and Donis et al.6 apparently have erroneously used gas transfer coefficients instead of proper CH4 fluxes to calculate emissions. This conclusion is demonstrated by the perfect agreement between the values published erroneously as CH4 fluxes, Fsurf, by Günthel et al.5 and the values of the gas transfer coefficients of CH4 at 20 °C, kCH4, calculated by us (Table 1). The values published by Donis et al.6 as CH4 fluxes are very similar to these kCH4 and therefore also do not represent CH4 fluxes but gas transfer coefficients (for details, see Supplementary Note 2.2).
Table 1 Average CH4 surface fluxes from Lake Hallwil and gas transfer velocities.
Full size table

The gas transfer coefficient of CH4 must be multiplied by the difference between the surface concentration (0.3 mmol m−3, ref. 6) and the atmospheric equilibrium concentration of CH4 (CH4,equ = 0.003 mmol m−3 at 20 °C9), i.e. by ~0.3 mmol m−3, to obtain Fsurf. Fsurf is therefore ~3.3 times smaller than the values of the gas transfer coefficients erroneously taken by Günthel et al.5 and Donis et al.6 as CH4 fluxes (Table 1 and details in Supplementary Note 2.2).
Donis et al.6 and Günthel et al.5 used values obtained from measurements with floating chambers to calculate emissions, but these values claimed to represent Fsurf appear to be in fact values for transfer coefficients, suggesting the same mistake as in the case of the wind models. Donis et al.6 stated: “Average flux (April–August 2016) is equal to 0.8 ± 0.2 mmol m−2 d−1 from MacIntyre relationship for positive buoyancy and to 0.6 ± 0.3 mmol m−2 d−1 from chamber measurements. The latter, not significantly different from the wind-based relationship, was used for the mass balance”. Günthel et al.5, co-authored by D. Donis, claim that the “MacIntyre relationship for positive buoyancy”10 provides an average value of 0.7 for Fsurf, but in fact 0.7 is the average value for kCH4 in unit m day−1 (0.7 m d−1, see Table 1) and Fsurf for this model is 3.3 times smaller (0.21 mmol m−2 d−1, see Table 1). The value by Donis et al.6 for the MacIntyre relationship10 is even slightly larger than 0.7 and therefore clearly incompatible with Fsurf but is rather a gas transfer coefficient as is obvious in the case of Günthel et al.5. The good agreement between the value for the gas transfer coefficient obtained from the MacIntyre model for positive buoyancy flux10 and the values from the chamber measurements suggests that the values from the chamber measurements are not gas fluxes but also gas transfer coefficients (see Supplementary Note 2.2 for more details).
Donis et al.6 derived from their chamber measurements the wind-based model “Hallwil relationship” specifically for Lake Hallwil. The establishment of this Hallwil relationship required that Donis et al.6 calculated gas transfer coefficients from their chamber measurements. In their Supplementary Fig. 4, Donis et al.6 show that the values from their chamber measurements agree well with those from the Hallwil relationship (Supplementary Fig. 2 and Supplementary Note 2.2). Note, however, that the values for the Hallwil relationship are in fact gas transfer coefficients and not Fsurf, supporting that also the values from the chamber measurements represent gas transfer coefficients and not Fsurf (Supplementary Fig. 2 and Supplementary Note 2.2 for more details). This conclusion implies that the values from the chamber measurements by Donis et al.6 must be multiplied by ~0.3 mmol m−3 to give proper CH4 fluxes, which are then ~3.3 times smaller than the CH4 fluxes used in the mass balances of refs. 5,6.
Because there are only four chamber measurements available for 2016 and one of them was exceptionally low (see ref. 6 and Supplementary Note 2.2), the Hallwil relationship is considered here to provide the most reliable estimate of the average k600 in Lake Hallwil and therefore applied to calculate the average surface CH4 flux for April to August 2016, i.e., Fsurf = 0.24 mmol m−2 d−1 (see Table 1 and Supplementary Note 2.2). The reliability of the Hallwil relationship was confirmed by Günthel et al.5 and by Hartmann et al.11 comparing different estimates of surface fluxes in the South Basin of Lake Stechlin.
With Fsed = 2.8 mmol m2 day−1 and Fsurf = 0.24 mmol m2 day−1, NOM = 416 mol day−1 and the contribution of NOM to total emissions is NOMC = 17% (Supplementary Table 1 in Supplementary Note 2.3 includes also additional estimates of NOMC). The low value of NOMC suggests that most of CH4 in the SML originates from the sediments, which is consistent with the δ13C isotopic composition of CH4 in Lake Hallwil9. The uppermost CH4 in the sediment core from the SML is characterized by δ13C about –59‰, which corresponds very closely to the δ13C of the CH4 in the open water of the SML ranging from −62‰ to −58‰ (Figs. 4 and 5 both in ref. 6). Thus the δ13C values suggest that the CH4 from the uppermost pore water in the sediment of the SML is the source of the CH4 in the open water and do not indicate a reduction of the δ13C expected in case of substantial CH4 production. More

The ongoing COVID-19 pandemic had its first confirmed case in the United Kingdom in late January 2020, but transmission increased rapidly leading to the government imposing a lockdown on the whole population, banning all “non-essential” travel and contact with people outside one’s home on 23 March 2020.1 Globally, the lockdown has caused households to change their typical consumption behaviours drastically across a variety of major categories, resulting in an initial sharp increase in spending, especially in essentials and food items.2 Studies dedicated to the impact of COVID-19 on water consumption focused on aggregate demand and general demand peaks. For example, in Germany, a significant shift in aggregate demand peak was observed from 07:10 pre-lockdown to 09:40 during lockdown.3 In a Waterwise4 report, certain regions in the UK saw a 35% increase in peak daily consumption during the lockdown. In Brazil, analysis of data from 26 days pre-lockdown and 26 during lockdown has revealed an 11% increase in household water consumption attributable to the lockdown.5 Although this rise can be generally attributed to an increase in diurnal consumption owing to people remaining at home, increase in preventive behaviours such as hand-washing1 also became contributory factors.
Household water demand in England and Wales is already at an all-time high, constituting 55% of the 32 Cubic Gigametres per year (Gm3/yr) total UK household water consumption footprint, with southeast England having the highest per capita consumption (PCC) and being already declared as severely water-stressed.6,7,8 The impacts of the extended time people stayed at home under the lockdown and the ensuing changes in behaviour arising from this have been an increase in household water demand, exacerbating existing pressure on network water supply.
Water utility companies are increasingly searching for ways to understand the full nature of household water use, how to improve network demand forecasting and achieve effective water efficiency interventions. By presenting a data-driven detailed characterisation of household clusters, including their unique patterns, we have demonstrated how the understanding of the impact of these unique patterns of behaviour on network demand can help in the design of demand forecasting and intervention that targets households on the basis of their shared cluster characteristics. Many demand strategies have relied on existing socioeconomic (SE) and sociodemographic (SD) household variables (e.g., ACORN9) and self-reported behaviours through surveys and water use diaries.10,11 Our work not only significantly enhances the precision of forecasting and intervention when enriched with SE and SD variables, but also provides a scalable framework for the inclusion of ordinary-metered and unmeasured households that share SE/SD characteristics peculiar to particular clusters.
We analysed the weekly water consumption data, at an hourly resolution, for January to May 2020 of 11,528 smart-metered households. We then classified the households according to their temporal patterns of consumption, highlighting their unique characteristics and their respective shares of relative and absolute consumption before and during the COVID-19 lockdown.
All households in the study are from a single water provider, collected across two geographical areas 50 miles apart consisting of 24 District Metering Areas (DMAs). As the aim of this study was to quantify the impact of the Covid-19 lockdown on aggregate water demand while highlighting household clusters’ underpinning temporal demand patterns, only anonymised smart meter data was utilised. Data on SD/SE or occupancy variables of the participating households were not available.
Overall temporal water consumption patterns
The analysis revealed an average consumption of 3256 cubic metres per day (m3/d) for the 11,528 households across the network for the period before the COVID-19 lockdown, equating to a per household consumption (PHC) of 284 litres per day (l/h/d), as per the UK average.12 Consumption remained even between the first week of January (J1) and the first week of February (F1) averaging 350 m3/d (291 l/h/d), followed by a 20% decline in February week 2–3 (F2–F3), before returning to average values in February week 4 (F4) to March week 3 (M3) as in Fig. 1b. A sharp increase was recorded in March week 4 (M4), the week of the COVID-19 lockdown, to 3756 m3/d (326 l/h/d), a rise of 10% on the previous week, reaching 4747 m3/d (411 l/h/d) in May week 4 (MY4), some 46% above pre-lockdown average. The cause of the 20% drop in consumption in the second week of February 2020 remains unknown. The water utility did report the loss of four days of data in that period owing to equipment power outage. The absence of any other plausible cause is suggestive that this may have resulted from “Storm Ciara”,13 which brought heavy rain and very strong winds to the region on 9 February 2020, causing widespread power issues.
Fig. 1: Households’ consumption patterns and trends before and during the COVID-19 lockdown in the UK.

a Differences in per household consumption (PHC) for January–May 2019 and 2020. b Weekly average 24-hour consumption for all households–January week 1 (J1) to May week 4 (MY4)—showing normal consumption trend, anomaly due to data loss and increase in consumption during lockdown period. c Hourly consumption patterns, showing households’ average proportion of hourly consumption to their daily average. d Households’ hourly mean and standard deviation consumption in litres. e Boxplots illustrating the comparison between pre-lockdown and lockdown total consumption in cubic metres (m3). Value at the top of whisker is the maximum consumption; bottom of whisker is the minimum consumption; top bound of the box is the upper quartile value; bottom bound of the box is the lower quartile value; the line in the centre of the box is the median and the x in the centre of the box is the mean ((bar x)). f Weekly cluster consumption trend showing how much each of the four clusters consumes per week in m3. The error bars indicate standard deviation (σ). g Weekly number of households per cluster. The error bars indicate standard deviation (σ).

Full size image

Comparison between this study and similar data from the previous year, January–May 2019, for the same households revealed similar patterns of consumption and cluster behaviours. However, analysis of the data revealed a respective rise in PHC across the network of 13%, 22%, and 29% in March, April and May 2020 (Fig. 1a).
To examine the temporal (hourly) consumption patterns, four quartiles (Q1–Q4) were assigned to the values between the lowest and highest consumption range, revealing a consistent 24-hour pattern throughout the period irrespective of the volume of consumption (Fig. 2a–d). Households Q1 represents 1–2% (per hour) of daily consumption and occurs invariably between 00:00 and 06:00. Q2, representing 3–4% of daily consumption, occurs principally between 14:00 and 15:00 and Q3, 5–6%, occurring at different times, particularly 12:00–13:00 and 21:00. The Q4 (peak) occurs at 9:00–11:00 and 19:00–20:00. The daily mean network water demand was 27% higher during lockdown than pre-lockdown, median 43% higher and Q4 26% higher. Figure 1e presents a comparison of consumption before and during the lockdown.
Fig. 2: Clusters’ hourly consumption patterns and comparison of clusters’ share of consumption before and during the COVID-19 lockdown in the UK.

a Cumulative pattern and percentage of hourly consumption for households in the “Evening Peak (EP)” cluster. Consumption is in (m3). b Cumulative pattern and percentage of hourly consumption for households in the “Late Morning Peak Peak (LM)” cluster. Consumption is in (m3). c Cumulative pattern and percentage of hourly consumption for households in the “Early Morning Peak (EM)” cluster. Consumption is in (m3). d Cummulative pattern and percentage of hourly consumption for households in the “Multiple Peak (MP)” cluster. Consumption is in (m3). e Average daily consumption per cluster pre-lockdown. f Average daily consumption per cluster during the lockdown. e, f Value at the top of the whisker is the Maximum consumption; bottom of the whisker is the minimum consumption; top bound of the box is the upper quartile value; bottom bound of the box is the lower quartile value; the line in the centre of the box is the median and the x in the centre of the box is the mean ((bar x)). g Clusters’ share of total hourly consumption pre-lockdown. h Clusters’ share of total hourly consumption during the lockdown.

Full size image

According to our findings, households’ proportion of total hourly water demand depends upon the clusters they belong to (Fig. 2g, h), although the ratio of their hourly consumption to their daily demand is largely consistent irrespective of the time of year or volumes consumed (Fig. 2a–d).
Household cluster characterisation before and after lockdown
The results reveal four distinct clusters of household water consumers characterised by unique diurnal and night-time consumption patterns. The clusters are named Evening Peak (EP), Late Morning (LM), Early morning (EM), and multiple peak (MP).
EP
Households in EP typically use ~6% of their daily consumption between 07:00 and 08:00 but their most significant consumption occurs between 19:00 and 20:00, which invariably constitutes ~10% of their daily demand in just 1 hour. Their Q2 constitutes ~4% of their relative daily consumption per hour and occurs between 09:00 and 16:00, with Q1, occurring between 00:00 and 05:00, representing ~1–2% (Fig. 2a). During the pre-lockdown weeks, this cluster constituted ~30% of the households across the network and has been responsible for over 50% (~76 m3/hr) of the relative consumption between 19:00 and 22:00 (Fig. 2g) and ~33% (1065 m3/d) of the total daily consumption, with a mean ((bar x)) of 39 m3/hr, standard deviation (σ) of 25 m3/hr and maximum (max) of ~92 m3/hr (Fig. 2e).
During the lockdown, the percentage of households in EP dropped to a network average of 25% and their dominance of consumption between 19:00 and 22:00 decreased to an average 45%, being ~135 m3/hr (see Fig. 2h) along with their share of the total daily consumption, which also decreased to 26% (~1087 m3/d). Lockdown hourly mean, standard deviation and maximum for EP were, respectively, (bar x) 63 m3/hr, σ 44 m3/hr and max 165 m3/hr (Fig. 2f).
LM
LM describes households whose peak (Q4) occurs typically at ~10:00, representing ~11–12% of their relative daily water consumption in just 1 hour, with their next significant water use activities (~5% of daily consumption/hour) occurring at 19:00. Q2 for this cluster constitutes ~4% of their relative daily consumption per hour and occurs between 14:00 and 17:00, with a Q1 being identical to EP and EM (Fig. 2b). On average, this cluster has the highest relative consumption between 10:00 and 12:00, constituting 38% (~63 m3/hr) pre-lockdown (Fig. 2g.), was represented by ~30% of households and had a 25% (808 m3/d) share of the total daily consumption, with (bar x) 28 m3/hr, σ 19 m3/hr and max 76 m3/hr (Fig. 2e).
The percentage of households in LM increased to an average of 37% across the network during the lockdown weeks but their consumption between 10:00 and 12:00 remained at an average of 38%–~74 m3/hr (see Fig. 2h). Their share of the total daily consumption increased to ~31% (~1281 m3/d). Lockdown being respectively (bar x) = 51 m3/hr, σ = 36 m3/hr and max = of 134 m3/hr (Fig. 2f).
EM
Households in EM have the fewest instances of peaks which constitute ~12–13% in 24 hours and occurs between 07:00 and 08:00. Q3 for this cluster, ~7% of their relative daily consumption, occurs at 19:00, Q2 between 10:00 and 17:00, constituting ~3–4% of their relative daily consumption per hour, and Q1 identical to EP and LM (Fig. 2c). On average, this cluster, made up of 26% of household, was responsible for 40% (~59 m3/hr) of pre-lockdown consumption occurring between 07:00 and 08:00 (Fig. 2g.) and 22% (723 m3/d) of the total daily consumption, with (bar x) of 26 m3/hr, σ of 17 m3/hr and max of 73 m3/hr (Fig. 2e).
EM experienced the sharpest decrease in the number of households during the lockdown period—an average of 12% across the network, resulting in a significant drop in their share of relative consumption between 07:00 and 08:00 to from 40% to 20%–~38 m3/hr (see Fig. 2h). Their share of the total daily consumption also fell to 12% (433 m3/d). Lockdown being, respectively, (bar x) 17 m3/hr, σ 10 m3/hr and max of 40 m3/hr (Fig. 2f).
MP
MP has the highest instances of Q4s within 24 hours (about seven instances of 6–7% of their relative daily consumption). They also have multiple instances of Q3s and Q2s at 5% and 4% of relative daily consumption, respectively. Their Q1, like the other clusters, resides between 00:00 and 06:00, constituting ~2–3% of relative daily consumption (Fig. 2d). During the pre-lockdown weeks, this cluster represented 14% of the households across the network. MP dominates consumption between 00:00 05:00—at an average of 32% (~8 m3/hr) (Fig. 2g) and about 20% (661 m3/d) of the total daily consumption, with (bar x) 19 m3/hr, σ 8 m3/hr and max 29 m3/hr (Fig. 2e).
MP experienced the most significant increase in the number of households during the lockdown period—a 93% increase between M3 and M4 maintaining an average of 26% of all households during the lockdown period. This has resulted in an increase in their share of hourly consumption between 00:00 and 07:00 to an average of 45%–~24 m3/hr; between 12:00 and 17:00 to an average of 39%–~98 m3/hr and 23:00 to 36%–~60 m3/hr (see Fig. 2h). Their share of the total daily consumption also rose to ~32% (~1326 m3/d). Lockdown being, respectively, (bar x) 67 m3/hr, σ 34 m3/hr and max of 110 m3/hr (Fig. 2f).
In another study,14 segmentation was based on heterogeneous micro-component consumption patterns and behaviour regularities and temporal characteristics. This work, unlike our study, performed a disaggregation of sub-minute smart meter data into end-use events, subsequently clustering households based on their end-use similarities. One difference of this study to the one here reported, however, is the consumer household sample size. In our study, some 11,528 households were assessed, and currently, sub-minute smart meters are unavailable across such a large region. Our segmentation was derived from normalised hourly smart meter data, being based on temporal patterns of consumption. The silhouette coefficient value, when ‘t-distributed stochastic neighbour embedding’ (t-SNE)15 was used for dimensionality reduction (as opposed to PCA), improved slightly from 3.9 to 4.1 for n_cluster = 4. However, this improvement only marginally enhanced the k-means results (by  More

Microbial community analysis
An overall microbial community analysis is presented as PCA plots and a dendrogram in Fig. 1. The cluster analysis showed good agreement between sample replicates, which clustered most closely. WWTP influent (i.e., untreated sewage) collected in the post-monsoon season clustered most closely with water samples from S4, S5 and S6 collected at same time, while WWTP influent from the monsoon season clustered with water samples from S6 collected in the same season. The WWTP effluent from both monsoon and post-monsoon season clustered together. The PCA plot with data from all the sampling times (Fig. 1b) generally showed a separation of downstream and WWTP influent water samples from the upstream and WWTP effluent samples along principal component 1, with only a few exceptions. Genera mostly found in the human gut microbiome15,26 like Streptococcus, Trichococcus, Lactobacillus, Enterococcus, Prevotella and Arcobacter, were highly prevalent in downstream and WWTP influent water samples, which separated these samples from the upstream water samples in the PCA. Among the three factors analysed (i.e., location, sampling time and water sample types), locations and sampling time had a significant effect on the similarity of the samples in the ANOSIM, although with relatively low R values (ANOSIM; Location: R = 0.29, p value = 0.001 and Sampling time: R = 0.16, p value = 0.01). ANOSIM further indicated no statistically significant differences between the microbial communities in water from locations S4, S5 and S6 and the wastewater influent (ANOSIM; (1) S4 and Inf: R = 0.0309, p value = 0.357: (2) S5 and Inf: R = 0.0617, p value = 0.369 and (3) S6 and Inf: R = 0.0123, p value = 0.3690).
Fig. 1: Cluster and PCA analysis at rank genus for 16S rRNA gene sequencing reads.

a Cluster analysis [all seasons], PCA plot b all seasons, c monsoon [June 2019] and d post monsoon [August 2019]. Arrows in the PCA plots indicate the ten variables with the highest loadings (vector lengths) in the PC1 and 2 space.

Full size image

An interesting picture emerged when a separate PCA (Fig. 1c, d) and cluster analysis (Supplementary Fig. 1) was conducted for water samples from the monsoon and the post-monsoon season. In both seasons, there were substantial, but seasonally distinct, contributions of genera found in the human gut microbiome to the variance among water samples along principal component 1: in the monsoon season Arcobacter, Aeromonas, Streptococcus and Prevotella had significant PC1 loadings; in the post-monsoon season Enterococcus, Acinetobacter, Streptococcus and Trichococcus had significant PC1 loadings. Separation of wastewater treatment plant effluent (WWTP Effluent) samples along PC1 away from the WWTP influent (WWTP Influent) samples in both sampling events signified the benefits of wastewater treatment, because human gut-associated genera became less predominant in treated wastewater microbiomes, as expected27. Accordingly, there was a clear separation of the most upstream water samples from the most downstream water samples along PC1 in both events, with the downstream water samples becoming more similar to WWTP Influent (Fig. 1c, d). Evidently, as the Bagmati River flowed into more densely populated areas, the characteristics of its water microbiome changed from a composition more similar to treated, to a composition more similar to untreated urban sewage, but the composition of the urban sewage was variable for the monsoon and post-monsoon season.
Abundance of human gut and putative pathogenic bacteria in the water microbiomes
A more detailed breakdown of the microbial community composition in the Bagmati River for the monsoon and post-monsoon season is reported in Table 1, and Supplementary Tables 1 and 2, which compare the total percentage relative abundance of putative human gut28 and pathogenic29 bacteria at genus and species level for different sampling sites in the Bagmati River, and the WWTP influent and effluent (Refer to Supplementary Tables 3–5 for more detailed lists of bacteria). Based on our previous findings24, species identities are not always reliable due to the limited read accuracy of the MinION sequencing reads, but the overall trends are nonetheless indicative of changes in microbial composition. For all sampling events, the water collected at the most upstream site S1 and S2 showed the lowest relative abundance for both human gut and putative pathogenic bacteria, whereas the highest relative abundance was observed in the water collected at the most downstream sites S4–S6 (Table 1, Supplementary Tables 1 and 2). The microbial water quality of water samples collected at site S1 can be considered as baseline data, as this watershed is distant from the densely populated Kathmandu Valley and has the minimal influence of human and urbanisation activities. Figure 2, and additional figures in Supplementary Information (Supplementary Fig. 2–9) show how the abundance of human gut and putative pathogenic genera changed in space and time along the Bagmati River. As the river flowed downstream, the abundance of some of these groups of bacteria increased, and the most drastic and significant increase was observed at the sites S4, S5 and S6 downstream of the Pashupatinath Temple as compared to site S1 (Two-sample t test, p value  More