The Baltic Sea is the world's largest area suffering from
eutrophication-driven hypoxia. Low oxygen levels are threatening its
biodiversity and ecosystem functioning. The main causes for
eutrophication-driven hypoxia are high nutrient loadings and global warming.
Wastewater treatment plants (WWTP) contribute to eutrophication as they are
important sources of nitrogen to coastal areas. Here, we evaluated the
effects of wastewater treatment plant effluent inputs on Baltic Sea
planktonic communities in four experiments. We tested for effects of effluent
inputs on chlorophyll
The Baltic Sea has the largest area affected by eutrophication-driven hypoxia (Conley et al., 2011). Eutrophication is expanding in the Baltic Sea; from 2007 to 2011 the entire open Baltic was found to be eutrophic (Fleming-Lehtinen et al., 2015). A 10-fold increase in the hypoxic area has been recorded for the last 115 years, mostly related to increased nutrient inputs from land (Carstensen et al., 2014). The lack of oxygen in marine waters causes death of marine organisms and catastrophic changes in marine metazoan communities. Thus, hypoxia is emerging as a major threat to marine biodiversity (Vaquer-Sunyer and Duarte, 2008), although prokaryotic diversity can increase in oxygen minimum zones (Wright et al., 2012).
Municipal wastewater treatment plants (WWTPs) contribute to eutrophication
because they are a substantial source of nitrogen (N) to natural waters
worldwide (Seitzinger et al., 2005). To reduce the environmental impact of
WWTP effluent discharge, limits on the concentration of nitrogen have been
imposed. In the European Union, the “Urban Waste Water Directive”
(91/271/EEC) sets the discharge limit of effluents from urban wastewater
treatment plants for total nitrogen (TN) between 10 and 15 mg N L
Effluent from WWTPs includes both dissolved inorganic (DIN) and organic N
(DON). The conventional biological treatment (secondary treatment) combines
coupled nitrification/denitrification and can potentially reduce TN to
around 8–12 mg N L
Physicochemical parameters in coastal seawater for the different sampled seasons. Standard errors (SE) are derived from duplicate sample analysis. C : N ratio is calculated as the ratio DOC : DON (moles).
DON can play an active role in providing nutrition to both phytoplankton and
bacteria (Berman and Bronk, 2003) and affects planktonic metabolism in
areas receiving significant amounts of DON. Dissolved organic matter (DOM)
inputs to coastal areas can also affect metabolic rates and favour bacterial
processes (Berglund et al., 2007). Here, we investigated the effects of
wastewater treatment plant (WWTP) effluent inputs on planktonic metabolic
rates in the Baltic Sea. We did so on the basis of four experiments where WWTP
inputs were added to natural communities. We tested for effects of effluent
inputs on metabolic rates: gross primary production (GPP), net community
production (NCP), community respiration (CR) and bacterial production (BP);
on chlorophyll
Natural marine planktonic communities from the Baltic Sea Proper were
collected (sampling dates included in Table 1) 10 km off the east coast of
Öland, Sweden, at the Linnaeus Microbial Observatory (LMO;
56
Wastewater effluent was collected within 10 days prior to experiment
(sampling dates included in Table 2) from the wastewater treatment plant
(WWTP) in Kalmar for effluent enrichment. Samples from WWTP were filtered
using pre-combusted (450
Wastewater effluent nutrient content for the different seasons sampled. C : N ratio is calculated as the ratio DOC : DON (moles).
Four experiments were performed to cover all seasons – spring, summer, autumn
and winter – to be able to measure seasonal variation in both planktonic
communities and effluent characteristics under different environmental
conditions. Each experiment consisted of five different treatments, three of them with different additions: one with
WWTP addition in a proportion of
Changes in dissolved oxygen (DO) in closed bottles were assumed to result
from biological metabolic processes and to represent net community
production (NCP
Incubations were illuminated by artificial light (OSRAM L36W/865 Lumilux
Daylight), with a photosynthetically active radiation intensity of 1373.2
NCP was estimated as the changes in DO content during 24 h intervals
(dDO/d
As incubations were performed following a natural light regime to mimic natural conditions, results may differ from incubations performed at light and dark conditions in parallel. Both approaches assume equal respiration rates under light and dark conditions. This assumption may lead to underestimation of CR and GPP, as respiration rates are probably higher during daylight than at night (Grande et al., 1989; Pace and Prairie, 2005; Pringault et al., 2007), but it does not affect NCP estimates (Cole et al., 2000). In incubations performed under dark conditions, phytoplankton growth is suppressed, decreasing phytoplankton respiration contribution to community respiration.
BP was estimated by measuring incorporation of
Samples for chlorophyll
Chlorophyll
DOC was measured on a Shimadzu TOC V-CPN in non-purgeable organic carbon (NPOC) mode on acidified samples (HCl to pH < 2). The instrument was calibrated daily with potassium hydrogen phthalate. DOC concentrations were calculated from the average area of three injections, with an area covariance of less than 2 %.
Total dissolved nitrogen (TDN) was measured in duplicate after persulfate
oxidation. The method of persulfate oxidation was chosen instead of high-temperature combustion (HTC), as it has been demonstrated to be more
appropriate for eutrophic waters, such as the Baltic Sea, as well as coastal
areas (Bronk et al., 2000). Inorganic nutrient analyses (nitrate
(NO
Bacterial 16S rRNA gene fragments were amplified with bacterial primers 341F
and 805R (Herlemann et al., 2011) following the PCR protocol of Hugerth et
al. (2014) with some modifications. We thus performed a two-step PCR: (i) amplification with the main forward and reverse primers 341F-805R to amplify
the correct fragment within the V3–V4 hypervariable region of the 16S rRNA
gene and (ii) amplification using template from the first PCR to attach the
handles and indexes needed to run the Illumina Miseq run and for barcoding
individual samples. Amplification was carried out in duplicates for each
biological replicate using an annealing temperature of 58
Relationships between chlorophyll
Metabolic rates data from the four experiments were combined to test the
relationship between the given metabolic rates and physicochemical
parameters (Table 1) by mixed-effects models. Physicochemical parameters
were chosen avoiding collinearity. Selected variables were DOC, DON, nitrate
and phosphate concentration. We used DOC as a proxy for dissolved organic
matter (DOM). Variables were selected according to its significance.
Variables were removed from the model following its
Differences in community composition between treatments were tested using permutational analysis of variance (PERMANOVA) on Bray–Curtis distances. To test the correlation between absolute changes in environmental conditions, metabolic rates and absolute shifts in bacterioplankton community composition, we performed MANTEL tests. For alpha-diversity measures we subsampled each sample to 10 000 sequences. Analyses performed at the OTU level were based on selecting the top 200 most abundant OTUs. For OTU level analyses on Cyanobacteria we selected OTUs affiliated with Cyanobacteria among the top 200 most abundant OTUs. Taxonomic annotation from SINA/SILVA database was limited for cyanobacterial OTUs and we therefore extended the annotation by using BLASTn (NCBI). For all analyses on community composition we examined the following eight major phyla/classes: Actinobacteria, Bacteroidetes, Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Cyanobacteria, Planctomycetes, and Verrucomicrobia. All other phyla/classes were grouped together and defined as “Others”. All statistical tests were performed in R 3.0.2 (R Core Team, 2014) and using the package “vegan” (Oksanen et al., 2007). Graphical outputs were made using the package “ggplot2” (Wickham, 2009). Phylogenetic analyses using maximum likelihood trees were performed with MEGA 6.0.6 and the Tamura–Nei model (Tamura et al., 2011).
Treated wastewater nutrient content differed between seasons (Table 2). The
highest TDN values were measured in winter (600.1
Nutrient content in the seawater also differed between seasons (Table 1),
with the highest TDN value in autumn (21.0
Coastal waters showed a typical seasonal pattern (Vahtera et al., 2007),
with low chlorophyll
Chlorophyll
Statistics for the fitted models for the different metabolic rates
and the variables that explain its variability; to account for
pseudo-replication, incubation day nested to season (i.e. experiment) was
included as random factor.
Chlorophyll
Gross primary production (GPP) for natural communities in the experiments
varied from 2.03
GPP variability was explained by differences in DOC concentration (Table 3), with this variable explaining 84 % of its variability (Fig. 3a). GPP decreased with DOC concentration (Table 3).
Community respiration (CR) for natural waters in the experiments varied
between 5.30
CR was inversely correlated with DOC concentration, with this variable explaining the 84 % of CR variability (Table 3, Fig. 3b).
Gross primary production (GPP) in mmol O
Net community production (NCP) for natural communities in the experiments
varied between
NCP was dependent on DOC concentration, with this variable explaining the 79 % of its variability (Table 3, Fig. 3c). NCP significantly decreased with
DOC content (
Results of MANTEL tests (Pearson's
Comparison of actual values and values predicted by the mixed-effects model for
Bacterial production (BP) tended to increase in the treatment with the
higher addition of effluent (Fig. 6). Repeated measures MANOVA showed
significant differences in BP for different sampling days, for treatments
and for the interaction between sampling day and treatment for experiments
conducted in summer and autumn (
BP was positively correlated with DOC content in spring, summer and winter
(
The variables that best explained BP variability were phosphate, DOC, DON
and NO
Community respiration (CR) in mmol O
Net community production (NCP) in mmol O
Bacterial community structure showed two distinct clusters with summer
communities separated from spring and winter across all experiments (Fig. S1, Supplement). Community composition in each experiment
exhibited, in general, a temporal succession and an additional response to
different treatments. We carried out MANTEL tests to elucidate the influence
of environmental factors on community composition and metabolic rates.
Changes in temperature significantly explained absolute shifts in
bacterioplankton community composition across all experiments (Pearson
Alpha-diversity estimated from the Shannon index was relatively similar between
treatments in each experiment and ranged from 3.34 to 5.82
Bacterial production in
Differences in alpha-diversity, estimated from the Shannon index, between controls and nutrient amendment – i.e. all nutrient-amended treatments were binned and compared against all controls. Circles denote variation in alpha-diversity within the binned samples. Colours correspond to different treatments.
Betaproteobacteria, Bacteroidetes and Alphaproteobacteria dominated the
April experiment, where Betaproteobacteria displayed a marked increase in
relative abundance from T0 to T7 (Fig. 8). In general, few differences in
community composition between treatments were observed. Nevertheless,
Betaproteobacteria decreased in relative abundance by more than half in
controls until T7, while they maintained their abundance in the other
treatments. For the January experiment, differences between treatments were
more pronounced (Fig. 8). Bacterial groups other than the eight major
phyla/class (“Others”) had nearly 4-fold higher relative abundance in
the
Patterns in community composition indicated that effluent amendments had an
effect on bacterial population dynamics in our experiments coupled with the
concomitant changes in metabolic rates. Hence, we performed Pearson
correlation tests to determine links between environmental factors,
metabolic rates and shifts in relative abundances at phyla/class level.
Shifts in relative abundances of Cyanobacteria, Planctomycetes and
Verrucomicrobia were positively correlated with temperature (Fig. 9). In
contrast, Alphaproteobacteria, Bacteroidetes and Betaproteobacteria were
negatively correlated with temperature. Cyanobacteria, Planctomycetes and
Verrucomicrobia displayed a strong negative correlation with community
respiration but a positive correlation with bacterial production. These
three groups of bacteria were also negatively correlated with
PO
Relative abundances (i.e. percentage of total sequences) of major bacterial groups at phyla/class level in the different treatments and experiments. Colours denote specific groups.
Changes in relative abundance of particular bacterial populations typically
followed the overall pattern within each major phyla/class. For example,
To extend the analysis of the strong Cyanobacteria population dynamics
observed in the July experiment, we investigated particular OTUs and plotted
relative abundances of this group across all experiments (Fig. S4). For the
other experiments, cyanobacterial populations had, in general, low relative
abundance but were still more abundant in treatments with effluent and
nutrients amendments than without (except for the April experiment). Six
OTUs showed particularly high relative abundance in the July experiment
(Fig. S4). These cyanobacterial populations increased with time and at T7
both
Nitrogen-rich dissolved organic matter (DOM) from WWTP effluents had
significant impacts on Baltic Sea planktonic metabolic rates: DOM
significantly increased bacterial production, whereas it decreased gross and
net primary production and community respiration rates, as showed in the
results of the mixed-effects models, where DOC is used as a proxy for DOM.
Bacterial production was also positively correlated with DON concentration,
suggesting that DON can provide nitrogen nutrition to bacteria. BP was
negatively correlated with phosphate concentration, due to seasonal
variations, as phosphate content is higher in winter when BP is low. A
parallel increase in BP and decrease in bacterial respiration (BR) rates
results in an increase in bacterial growth efficiency
(BGE
Wastewater treatment plant effluent inputs to the Baltic Sea raised
bacterial production at the same time as it reduced primary production,
leading to more carbon being used by the microbial loop. This increase in
bacterial production parallel with a decrease in primary production moves
the ecosystem towards heterotrophy. This is supported by a higher BP : NCP
ratio in treatments with addition of WWTP effluent (mean
Correlations between shifts in relative abundances of major
bacterial groups at phyla/class level and environmental factors and
metabolic activity. The level of correlation is estimated from Pearson's
Effluent inputs decreased GPP and NCP, resulting in a reduction in photosynthetic rates, declining oxygen production in the photic layer. The Baltic Sea is already the largest eutrophication-driven hypoxic area in the world (Conley et al., 2011), and a decrease in biological oxygen production could further aggravate hypoxic conditions in this already affected area. The lack of oxygen is an important environmental problem is this area; it produces a reduction in marine benthic diversity as a result of the death of sensitive marine organisms and it affects biogeochemical cycles (Conley et al., 2009). Furthermore, it increases phosphorus fluxes from sediments into overlaying waters, changing redox conditions in the water column and reduces the ecosystem capacity of removing nitrogen, as a consequence of the reduction in the substrate needed for denitrification (nitrate) when sediments become more reducing (Conley et al., 2009).
Although several microbial taxa showed weak correlations with contemporary changes in environmental conditions and/or metabolic activity, specific opportunistic populations proliferated in effluent input treatments. In particular, verrucomicrobial and cyanobacterial populations responded in relative abundance to effluent inputs in summer. Thus, OTUs affiliated with Verrucomicrobia decreased in relative abundance in the treatments with effluent addition compared to controls. In contrast, the relative abundance of a few specific cyanobacterial populations increased upon enrichment (but less so in controls – i.e. the cyanobacterial growth was not only an effect of higher temperatures in the summer experiment). Generally, it is likely that the proliferation of cyanobacteria in the summer experiment is linked to the actual abundance of cyanobacteria, which is typically higher in summer, so that the “seeding” population for this taxon was higher. The Baltic Sea suffers from extensive Cyanobacteria blooms in summer that can easily be observed from space, primarily caused by eutrophication (Vahtera et al., 2007). The death and sedimentation of Cyanobacteria blooms, and the subsequent decay of this organic material, is a contributing mechanism for oxygen depletion in bottom waters. Consequently, Cyanobacteria blooms have been linked to hypoxia development and expansion in the Baltic Sea. Warming could further increase cyanobacteria blooms in the Baltic Sea (Paerl and Huisman, 2008; Paerl and Paul, 2012). Here, we found that relative abundances of Cyanobacteria were positively correlated with temperature.
Links between metabolic activity and compositional changes of bacterial
communities are frequently observed in aquatic ecosystems (Bell et al., 2005;
Allison and Martiny, 2008; Logue et al., 2016). Yet, in other cases, such
linkages are relatively weak and possibly confounded by environmental
complexity (Comte and Del Giorgio, 2011; Comte et al., 2013; Langenheder et
al., 2005, 2010). Our results showed that effluent inputs caused simultaneous
shifts in community composition coupled with changes in metabolic rates.
Changes in temperature were the major driver of community structure, but
phosphate also significantly explained variations in the relative abundance
of particular groups and taxa. This emphasises that changes in temperature
and nutrient availability can affect bacterioplankton community dynamics.
Similarly, differences in temperature and nutrient conditions lead to shifts
in community structure in, for example, mesocosm experiments with
Mediterranean and Baltic Sea microbial assemblages (Degerman et al., 2013;
Gomez-Consarnau et al., 2012; Pinhassi et al., 2006; von Scheibner et al.,
2014). More importantly, in these studies, compositional shifts occurred with
concomitant responses in community metabolic activity. Apart from the
influence of temperature in structuring the bacterial communities in the
present study, shifts in bacterioplankton community composition were highly
correlated with changes in phosphate concentrations. In agreement, previous
findings show that phosphate is a driver of shifts in community structure in
the southern Californian coast and Baltic Sea (Fuhrman et al., 2006;
Andersson et al., 2010). For example, Andersson et al. (2010) suggested that
limiting conditions due to a decline in phosphate during the summer
Cyanobacterial bloom promote selection in the bacterioplankton community
where specific OTUs can proliferate. Moreover, in an adjacent area of the
Baltic Sea Proper, opportunistic cyanobacteria,
including N
The so-called “bottle effect”, in which confinement of water causes shifts
in bacterioplankton community composition and physiological rates, is a
factor to consider in interpreting results from experiments with natural
microbial assemblages (Fuchs et al., 2000; Massana et al., 2001; Baltar et
al., 2012). Such effects are typically detected by rapidly increasing
proportions of fast-growing gammaproteobacterial populations and rate
measurements across all treatments (including controls; Pinhassi and Berman,
2003; Sjöstedt et al., 2012; Dinasquet et al., 2013). In our current
experiments, microbial community composition remained relatively similar to
in situ communities and we did not observe excessive increases in
opportunistic bacterial populations in the controls. Rather, increases and
decreases in relative abundance were observed among populations typical of
Baltic Sea Proper, such as
Inputs of WWTP effluent further stimulated bacterial production in summer, when it was already high due to elevated temperatures. Summer was the period of the year that responded sharply to effluent additions. Warming could also increase respiration rates to a larger degree than primary production, moving the system towards heterotrophy (Brown et al., 2004; Harris et al., 2006; Vaquer-Sunyer et al., 2015; Yvon-Durocher et al., 2010). Simultaneous warming and inputs from wastewater treatment plant effluents increased planktonic respiration rates and bacterial production faster than it increased planktonic primary production in the Baltic Sea (Vaquer-Sunyer et al., 2015), leading to higher biological oxygen consumption than production, which may lead to the depletion of the oxygen pool, further aggravating hypoxia in the Baltic Sea. Here, we found that WWTP effluent inputs increased bacterial production at the same time as they decreased net and gross primary production and community respiration. A parallel increase in bacterial production and decrease in primary production leads to more carbon being used by the microbial loop and may have consequences for food web transfer efficiency.
The current study showed that inputs of DOM from WWTP effluents were related to increased bacterial production and decreased primary production and community respiration, which could lead to an increase in BGE. DON concentration enhanced bacterial production, suggesting that bacteria can use DON as a nitrogen source. The increase in BP and decrease in CR could be caused by high lability of the OM that supported secondary bacterial production, without an increase in respiration. Seasonal changes in temperature were the most important factor for structuring community composition, but phosphate concentrations also significantly explained variations in the relative abundance of particular groups and taxa. In summer, the relative abundance of Cyanobacteria increased after effluent inputs (but less so in the controls). Cyanobacteria have been linked to hypoxia in the Baltic Sea, and an increase in their abundance could result in oxygen depletion of the Baltic bottom waters. Inputs from wastewater treatment plant effluent could further worsen hypoxic conditions in the Baltic Sea.
Reductions in the OM content in wastewater treatment plant effluents are needed to reduce its potential negative consequences. Effluent inputs resulted in a reduction in photosynthetic rates, moving the system towards heterotrophy, decreasing oxygen production in the photic layer in the Baltic Sea.
STRÅNG data used here are from the Swedish Meteorological and Hydrological Institute (SMHI), and were produced with support from the Swedish Radiation Protection Authority and the Swedish Environmental Agency.
Raquel Vaquer-Sunyer designed research and performed experiments. Markus V. Lindh, Jarone Pinhassi and Saraladevi Muthusamy analysed bacterial diversity samples and data. Heather E. Reader wrote the code for metabolic rates calculations. All authors were involved in the writing stage of the manuscript and collaborated on the analysis, interpretation, and discussion of the results.
The authors would like to thank Catherine Legrand, Emil Fridolfsson, Anders Månsson and Kristofer Bergström at the Linnaeus University for their help on Kalmar's WWTP effluent water and seawater sampling. We would also like to thank Carolina Funkey for help with nutrient analysis and Elena Baraza for statistical advice. Raquel Vaquer-Sunyer was supported by a Marie Curie Intra-European Fellowship (IEF). This research is a contribution to the projects “The role of dissolved organic nitrogen (DON) on the development and extent of eutrophication-driven hypoxia and responses to global warming”, funded by the FP7 Marie Curie IEF (grant number 299382), and “Managing multiple stressors in the Baltic Sea”, funded by FORMAS (grant number 217-2010-126). This work resulted from the BONUS COCOA project and the BONUS BLUEPRINT project, which were supported by BONUS (Art 185), funded jointly by the EU and the Swedish Research Council FORMAS. Edited by: G. Herndl Reviewed by: three anonymous referees