Microbial communities across activated sludge plants show recurring species-level seasonal patterns

Microbial communities in activated sludge (AS) are the core of sanitation in wastewater treatment plants (WWTPs). Microbial communities in AS have shown seasonal changes, however, long-term experiments (>2 years) are rarely conducted, limiting our understanding of the true seasonal dynamics in WWTPs. In this study, we resolved the microbial seasonal dynamics at the species level in four municipal full-scale WWTPs, sampled every 7–10 days, during 3–5 consecutive years. By applying a new time-series analysis approach, we revealed that the seasonal pattern was species-specific, where species belonging to the same functional guild or genus may show different seasonal dynamics. Species could be grouped into cohorts according to their seasonal patterns, where seasonal cohorts showed repeatable annual dynamics across years and plants. Species were also grouped according to their net growth rate in the AS (i.e., growing species and disappearing species). Growing species were more prevailing in spring and autumn cohorts, while disappearing species, which were only present due to the continuous immigration from influent wastewater, were mostly associated with winter and spring cohorts. Most known process-critical species, such as nitrifiers, polyphosphate accumulating organisms and filamentous organisms, showed distinct species-specific patterns. Overall, our study showed that overarching seasonal patterns affected microbial species in full-scale AS plants, with similar seasonal patterns across plants for many dominant species. These recurrent seasonal variations should be taken into account in the operation, understanding and management of the WWTPs.


Table of contents
.

Figure S16. Classification of species within the same genus into seasonal cohorts.
Distribution and classification of species into seasonal cohorts within the same genus (for genera containing more than one species) in the four WWTPs.  Figure S18 visualises some of the parameters in Aalborg West ( Figure S18A) and Damhusåen ( Figure S18B). We do not have access to similar data for Randers and Aalborg E. Aalborg West and Damhusåen show a normal process performance throughout the years. Not all the WWTP measure the same parameters, with the same frequency or the same methods, which complicates cross-WWTPs comparisons. Time-series decomposition and simple harmonic fitting were used to analyse the seasonal component of the monitored parameters. The methodology used was the same as for the species evaluation (see section 2.3.2 in the main manuscript) except the measured parameters were standardised to mean of zero and variance of 1 prior to time-series decomposition.

Note 1: Species response to operational and environmental variables N.1.1 Main monitored parameters
The results of the analysis show that most of the monitored operational parameters fluctuate randomly during the year (see Supplementary file S4 for detailed analysis). Briefly, only temperature (both WWTPs), effluent nitrate concentration (both WWTPs), hydraulic retention time (HRT, Aalborg West), sludge volumetric index (SVI, Damhusåen), suspended solids in the tank (SS, Damhusåen), SS in the effluent (Damhusåen) dissolved oxygen (DO, Damhusåen) and influent insoluble nitrogen (Damhusåen). Wastewater temperature and effluent nitrate concentration were the only parameters that showed a similar pattern in both plants. Temperature can affect microbial growth rates, but also shift the biochemical equilibria and kinetic rates. For example, lower denitrification activity is well-known to vary depending on the organic load and the temperature in the anoxic reactor (Metcalf & Eddy et al., 2014). Higher nitrate concentration is usually observed when wastewater temperature is lower (winter in the northern hemisphere) due to lower denitrification kinetic rates (Dawson andMurphy 1972, Metcalf &Eddy et al., 2014). Moreover, lower temperatures increase oxygen solubility which makes it more difficult achieving anaerobic conditions in the anoxic tank, inhibiting denitrification (Oh and Silverstein, 1999). Other reasons for lower denitrification rates might be low organic concentration in the influent wastewater, although no systematic changes were observed for Aalborg West and Damhusåen. However, given the high diversity of potential denitrifying species, it is unclear if the lower denitrification activities are related to seasonal changes in microbial species or a kinetic response due to environmental factors where denitrification has a lower activity than nitrification.

N.1.2 Variance partitioning at species-level and seasonality of operational variables
Variance partitioning analysis was used to estimate the contribution of each of the monitored parameters to the species-observed variability. Variance partitioning analysis was carried in R using the package variancePartition (Hoffman and Schadt, 2016). In case of collinear monitoring parameters, only the most commonly used has been kept. For example, chemical oxygen demand (COD), biological oxygen demand (BOD) and total organic carbon (TOC) are proxies for organic matter, and in general, are linearly correlated. For this study, we chose COD since it was measured in both plants. To remove the collinearity from total nitrogen (TN) and ammonia, and total phosphorus (TP) and orthophosphate; ammonia was subtracted from TN and orthophosphates from TP. The resulting fractions named insoluble nitrogen and insoluble phosphorus were retained together with ammonia and orthophosphate. Figure S19 summarises the explained variance for all growing species in Aalborg West and Damhusåen. Figure S20 shows the variance partition explained by each monitored for the top 25 growing species in each WWTP.
Temperature was a major explanatory variable explaining up to 40% of the variance for some species in both Aalborg West and Damhusåen. As expected, it showed a repeatable seasonal pattern for both WWTPs. Some of the influent characteristics, such as the soluble COD (as a proxy for readily available organic matter), insoluble phosphorus or nitrogen (without ammonia) were also important (explaining ~18-40% of the variance for some species). However, those parameters appear to be WWTP-specific and display a random variation during the year ( Figure  S18, supplementary file S2-S3-S4). Some performance parameters, e.g., SVI in Damhusåen and effluent nitrate concentration (both WWTP), were associated to the variability of species. Both SVI (Damhusåen) and effluent nitrate concentration were seasonal (supplementary file S3-S4). However, these parameters are typically considered a response-variables from bacterial activity and whether the change in those parameters can affect back the microbial community is unknown. Nevertheless, for all growing species, between 18 -90% of the variance could not be explained by any of the monitored parameters. Currently, monitoring parameters are chosen to evaluate the process performance, aid with process control and comply with legislation. However, these bulk measurements appear inconclusive to evaluate their impact on species dynamics, especially to explain systematic variations. A harmonisation of measured parameters as well as detailed studies identifying the most relevant measurements may improve the understanding of microbial dynamics in engineered systems.