Increased microbial activity in a warmer and wetter climate enhances the risk of coastal hypoxia

Authors

  • Anna Nydahl,

    1. Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Satya Panigrahi,

    1. Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden
    2. Environment & Safety Division, Indira Gandhi Centre for Atomic Research, Kalpakkam, India
    Search for more papers by this author
  • Johan Wikner

    Corresponding author
    1. Umeå Marine Sciences Centre, Umeå University, Hörnefors, Sweden
    • Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden
    Search for more papers by this author

Correspondence: Johan Wikner, Umeå Marine Sciences Centre, Umeå University, SE-905 71 Hörnefors, Sweden. Tel.: +46 (0)90 7867980; e-mail: johan.wikner@umf.umu.se

Abstract

The coastal zone is the most productive area of the marine environment and the area that is most exposed to environmental drivers associated with human pressures in a watershed. In dark bottle incubation experiments, we investigated the short-term interactive effects of changes in salinity, temperature and riverine dissolved organic matter (rDOM) on microbial respiration, growth and abundance in an estuarine community. An interaction effect was found for bacterial growth, where the assimilation of rDOM increased at higher salinities. A 3 °C rise in the temperature had a positive effect on microbial respiration. A higher concentration of DOM consistently enhanced respiration and bacterial abundance, while an increase in temperature reduced bacterial abundance. The latter result was most likely caused by a positive interaction effect of temperature, salinity and rDOM on the abundance of bacterivorous flagellates. Elevated temperature and precipitation, causing increased discharges of rDOM and an associated lowered salinity, will therefore primarily promote bacterial respiration, growth and bacterivore abundance. Our results suggest a positive net outcome for microbial activity under the projected climate change, driven by different, partially interacting environmental factors. Thus, hypoxia in coastal zones may increase due to enhanced respiration caused by higher temperatures and rDOM discharge acting synergistically.

Introduction

Coastal zones are the most productive marine areas and are also the marine environments that are the most exposed to environmental drivers associated with terrestrial activities and human pressures (Gazeau et al., 2004; Smith & Benner, 2005). For example, elevated precipitation increases the supply of different chemical compounds through riverine discharge, apart from the influence of freshwater per se on salinity and hydrography (Temnerud & Weyhenmeyer, 2008). In addition, the ratios of nutrients and other quality factors related to discharged chemicals may change. A better understanding of the factors controlling important ecological processes, such as biomass production and respiration in the coastal zone, is therefore necessary.

Climate-change projections suggest that precipitation will increase by approximately 20% by the end of this century. This will lead to increased riverine discharge to the sea, potentially resulting in higher inputs of riverine dissolved organic matter (rDOM) and a lower salinity in the coastal zone (Meier, 2006; Raymond & Saiers, 2010). In addition, an air temperature increase of approximately 2–5 °C is projected for northern Scandinavia. This situation motivates research of the interactive effects of these climate-related land-sea drivers on microbial activity to determine their relative importance, as specified below.

Temperature is generally found to promote community respiration in estuarine environments, but chlorophyll also contributes to explain the variance in this parameter (Hopkinson & Smith, 2005). Some field studies have shown a positive effect of temperature on both bacterial growth and bacterial respiration (Vaque et al., 2009; Kritzberg et al., 2010), while other studies have suggested that bacterial growth is less dependent on temperature (Li & Dickie, 1987; Wikner & Hagström, 1999). The importance of temperature is further supported by mesocosm experiments, which have shown a positive effect of temperature, primarily on respiration but also on the timing of peak bacterial secondary production (Hoppe et al., 2008; Wohlers et al., 2009). Studies addressing the direct effect of carbon substrates on respiration are less common, but those that are available also suggest a positive effect (Pomeroy & Wiebe, 2001). For bacterial growth, temperature and DOM have been attributed equal importance, at least in cold waters (Kirchman et al., 2005). Therefore, the temperature and substrate concentration may influence bacterial growth and respiration in an interactive way, motivating simultaneous examination of these factors.

Salinity has also been observed to influence the transformation of rDOM. Bacterial utilisation of allochthonous dissolved organic material has been found to be enhanced under estuarine conditions compared with limnic conditions when measured using either carbon- or nitrogen-based methods (Wikner & Hagström, 1999; Stepanauskas et al., 2002; Smith & Benner, 2005). Most of these experiments have been performed in dark incubations, suggesting that the increased salinity per se promotes the bioavailability of terrigenous DOM (i.e. without photolytic activity being involved). In addition, pre-exposure to light will further enhance the bioavailability of DOM (Bertilsson et al., 1999; Smith & Benner, 2005). Salinity has also been shown to reduce the efficiency of estuarine bacterial growth and shift the taxonomic composition of the bacterial community towards more marine species, such as Pseudomonas and Bacteroidetes (Langenheder et al., 2003). It is therefore important to consider salinity when studying the bacterial transformation of rDOM in coastal zones.

In this study, we investigate the interactive influence of changes in salinity, temperature and rDOM on microbial respiration, growth and abundance in an estuarine community. The levels for the different treatments were chosen from a climate change perspective and correspond to the predicted changes for this area under a moderate reduction of anthropogenic CO2 emissions (Meier et al., 2006). The experimental design for the microcosms was fully factorial to allow statistical analysis of interaction effects. Although temperature and autochthonous DOM have been investigated in many previous studies, our study design is unique, in which we also examine riverine (allochthonous) DOM, salinity and the relative importance of these factors when varied simultaneously.

Materials and methods

Experimental setup

The experiment was performed using a full factorial design including three factors (rDOM, salinity and temperature) with two levels each, resulting in eight different treatments (Table 1). Each treatment was conducted with six replicates, but dissolved organic carbon (DOC) and nutrients were only measured in three replicates to match resources. Phytoplankton and flagellates were measured in two replicates per treatment to facilitate the calculation of their potential contribution to respiration.

Table 1. Experimental design, including the environmental factors and their levels
TreatmentrDOM (μM)Salinity (g kg−1)Temperature (°C)
  1. The in situ temperature 12 °C, 300 μM DOC and 3 g kg−1 salinity represent current levels.

1300112
2300115
3300312
4300315
5600112
6600115
7600312
8600315

Sampling and preparation of the media and inoculum

A defined artificial brackish water medium (BWM) with a salinity of 1 or 3 g kg−1 was prepared from a 100 × stock solution [NaCl: 79.48 g, KCl: 2.236 g, Na2SO4: 9.943 g, MgCl2 × 6H2O: 180 cm3 (203.9 g dm−3), CaCl2 × 2H2O: 30 cm3 (145.4 g dm−3)]. A total of 40 dm3 of river water was collected from the Öre River 2 km upstream of the river mouth (latitude 63.56216, longitude 19.69879). The river water was filtered via tangential flow through a 0.2-μm filter (suspended screen channel; Pall Corporation, Ann Arbor, MI) to obtain rDOM. This filtrate was further concentrated via tangential flow filtration (10 000 kd, suspended screen channel; Pall Corporation) to a volume of 1.0 dm3. The lower treatment level for rDOM (300) aimed at a concentration of 300 μM carbon, corresponding to the natural level of DOC expected in the Bothnian Sea. Eight different treatments were created by mixing rDOM with brackish water media of different salinities according to Table 1, followed by incubation at the in situ temperature (12 °C) or 3 °C elevated temperature (15 °C). rDOM 300, a salinity of 3 g kg−1 and the in situ temperature are considered to represent the current situation in the study area (i.e. control treatment), while rDOM 600, a salinity of 1 g kg−1 and the elevated temperature mimic potential effects of climate change. These values were chosen from a relevant high precipitation year (1998), when there was a 20% increase in the river flow compared with the long-term average (Wikner & Andersson, 2012).

The bacterial inoculum was created by prefiltering 20 dm3 of the estuarine water sample through a 200-μm mesh net. The prefiltered sample was then concentrated to a volume of 1 dm3 by filtering it through several 0.2-μm-pore-size polycarbonate filters (∅ 47 mm, Poretics; Osmonics Inc., CA) and resuspending the bacteria, protozoa and phytoplankton cells retained on the filter surface in the remaining sample water (1 dm3) with a Pasteur pipette. An aliquot of 20 cm3 of the microbial concentrate was added to each of 48 incubation flasks, together with 680 cm3 of the BWM (35× dilution of the bacterial inoculum). Following this procedure, we obtained a natural concentration of microbial plankton of < 200 μm, together with a defined chemical composition of the growth medium. The slightly higher dilution of the inoculum was chosen to keep the microbial abundance below the carrying capacity and promote net growth. The addition of rDOM contributed carbon substrates as well as nitrogen and phosphorous to the different treatments.

Incubation

The samples were incubated in 48 tissue culture flasks with a volume of 700 cm3 each (Polystyrene; Sarstedt Inc., NC). The temperature of the samples was controlled by incubating in two separate incubator cabinets (Prebatem 36 l; JP Selecta S.A., Spain) set at the in situ temperature (12 °C ± 0.01 °C) and in situ temperature plus 3 °C (15 °C ± 0.01 °C). The incubator cabinets were placed in a temperature-controlled room to minimise potential temperature fluctuations. Samples (a total volume of 215 cm3 per sampling occasion) were drawn from each of the 48 flasks at 1, 24 and 74 h with pipettes, without replacing the water. As the experiment was performed in the dark, no photosynthesis could occur.

Respiration

Respiration measurements were taken with a dynamic luminescence quenching-based system (Sensor Dish Reader, SDR2; PreSens GmbH, Germany) to allow continuous measuring. This system has been used previously in a similar way (Warkentin et al., 2007, 2011; Amin et al., 2012). Two plates containing 24 samples of 5 mL each were incubated at each temperature, and respiration measurements were performed for 12 h to obtain a sufficient oxygen change. Six replicate vials, one from each of the 48 incubation flasks, were used in each treatment. The Q10 values for respiration were estimated according to Sherr & Sherr (1996).

Bacterial growth

Bacterial growth was measured in triplicate for each of the 48 incubation flasks at each sampling based on the incorporation of radioactively labelled 3H-thymidine (20 nM final concentration, 86.0 Ci mmol−1; Amersham Biosciences UK Ltd., UK) in a 1-h incubation using the microcentrifuge method (Smith & Azam, 1992). A thymidine conversion factor (TCF) of 1.4 × 10−18 per mole of incorporated thymidine was applied based on a data set compiled from the Baltic Sea (Riemann et al., 1987; Autio, 1998, 2000; Wikner & Hagström, 1999).

Bacterial abundance

A 50-mL sample from each flask and time point was preserved in formaldehyde (2% final concentration) to estimate the bacterial abundance. A 5-mL aliquot of the samples was filtered through black polycarbonate filters (0.2 μm pores, 25 mm diameter, Poretics; Osmonics Inc.), followed by staining with acridine orange as described by Hobbie et al. (1977). Five microscopy fields for each filter (> 580 cells filter−1) were analysed using epifluorescence microscopy (63×/1.4 oil plan-Apochromat objective, Axiovert 100; Zeiss GmBH, Germany) with the image-analysis program Labmicrobe (Blackburn et al., 1998). The carbon density for each cell was calculated using published volume-to-carbon density functions (Norland, 1993). The applied method measures the members of both the Bacteria and Archea domains, and the term ‘bacteria’ is used hereafter for both groups.

Protozoans and phytoplankton

Counts of protozoans and phytoplankton were performed to obtain an estimate of their abundance after filtration in the samples and to calculate the potential contribution to dark respiration from these organisms. To enumerate the heterotrophic flagellates present, 15 mL water samples from two of the replicates used for the bacterial counts were filtered through black polycarbonate filters (0.6 μm pores, 25 mm diameter, Poretics; Osmonics Inc.) and stained with DAPI (Sherr et al., 1992). A total of 20 visual fields on each filter were counted manually using a microscope (100×/1.4 oil plan-Apochromat objective, Axiovert 100; Zeiss GmBH, Germany). Protozoan respiration was estimated based on protozoan cell volume according to Fenchel & Finlay (1983).

Samples (50 mL) from two incubation flasks from each treatment were preserved in Lugol's solution and used to perform phytoplankton counts. After settling in incubation chambers (25 mL) for 24 h, the phytoplankton were counted and identified manually under a microscope according to Andersson et al. (1996), using a Nikon Eclipse Ti–S with 100× and 400× objectives, respectively.

Nutrients and DOC

Total nitrogen and phosphorus were measured in samples oxidised via autoclaving (121 °C, 1.2 bar) in acidic potassium peroxodisulphate, analysing orthophosphate and nitrate, respectively, using a four-channel auto-analyser (QuAAtro marine; Bran & Luebbe®, Sweden) according to the Swedish Standards Institute and Grasshoff et al. (1983). The extended measurement uncertainty varied between 8.4% and 22% depending on the examined substance.

The dissolved organic carbon contents were determined in filtered samples (0.2-μm Supor filters; Pall Corporation) using a Shimadzu TOC-5000 (Shimadzu Corporation, Kyoto, Japan) high-temperature catalytic oxidation instrument with nondispersive infrared (NDIR) detection (Sugimura & Suzuki, 1988; Norrman, 1993). The samples were acidified to pH 4 and purged with an inert gas for 10 min prior to analysis. The carbon concentrations were calculated using potassium hydrogen phthalate as a standard.

Statistical analysis

Differences between the treatments, focusing on bacterial respiration, abundance and production, were tested with a repeated linear modelling procedure in spss 20.0 software (IBM SPSS Statistics 20 Algorithms, 2011; IBM Corporation). To achieve a normal distribution, the respiration data were transformed using the natural logarithm. The measures were repeated over time within the treatments, using time, rDOM, salinity and temperature as fixed factors. The partial eta-squared statistic (partial η2) was used to assess the effect size, which indicates the variation explained by a particular factor that was not explained by other variables. Significant values with partial η2-values > 0.3 were considered large. According to Mauchly's test, the assumption of sphericity was met for all of the variables, except bacterial growth, for which the Huynh–Feldt correction was applied.

Results

Microbial respiration was found to be primarily positively influenced by the increase in temperature (P < 0.001; partial η2 = 0.61) (Table 2). The positive effect of temperature on respiration was highest during the first 24 h of the experiment, when the respiration increased 10-fold in the two high-rDOM treatments. After 72 h, only a doubling of values due to the temperature effect was recorded, corresponding to Q10 values as high as 2370 and 12 for the elevated and in situ salinity, respectively (Fig. 1). The median Q10 value for the respiration measurements taken in all treatments conducted at the elevated temperature was 26.

Table 2. Direction of significant effects obtained from rmanova analysis, with the probability of a type 1 error and partial eta-squared values shown in parentheses (P-value, partial η2)
Variable/parameterRespirationBacterial growthBacterial abundanceFlagellate abundance
  1. The values for bacterial growth associated with the rDOM and salinity factors are similar at the given precision. Partial eta-squared values, showing the amount of variation accounted for by a given factor (0 < partial η2 < 1) that is not explained by other variables. The natural logarithm-transformed microbial respiration was used in the statistical analysis.

  2. ↑, Significant positive effect; ↓, significant negative effect; n.s., no significant effect.

Temperature (0.000, 0.610)n.s. (0.001, 0.255)↑ (0.000, 0.993)
rDOM (0.028, 0.146) (0.000, 0.341) (0.000, 0.689)n.s.
Salinityn.s↑ (0.000, 0.341)n.sn.s.
Temperature × rDOMn.s.n.s.n.s.n.s.
Temperature × salinityn.s.n.s.n.s.↑ (0.001, 782)
rDOM × salinityn.s. (0.005, 0.179)n.s.n.s.
Temperature × rDOM × salinityn.s.n.s.n.s. (0.016, 0.538)
Figure 1.

Primary data on microbial respiration rates (mean ± 95% confidence interval). The x-axis is marked with temperature/salinity/rDOM concentration from bottom to top.

rDOM had also a positive effect on microbial respiration, although its influence was clearly weaker (P = 0.028; partial η2 = 0.15, Table 2, Fig. 1). The only interaction that was close to significant (P = 0.079) was the temperature × rDOM × salinity interaction, although it presented a low partial η2 value (0.096) (Table 2). The temperature increase of 3 °C did not have a significant effect on bacterial biomass growth, and the estimated marginal means indicated a negative, rather than a positive effect (Table 2, Fig. 2). The significant salinity × rDOM interaction (P = 0.005; partial η2 = 0.18) revealed that these factors may act synergistically to promote bacterial growth. The obtained profile plots suggested that increased growth occurred at a higher salinity and rDOM in all of the examined cases, except for the 300 μM treatment performed at the in situ temperature (Fig. 3). However, the salinity × rDOM interaction showed a rather low partial η2 value, suggesting moderate explanation of the variation in bacterial growth.

Figure 2.

Primary data on bacterial growth (mean ± 95% confidence interval). The x-axis is marked as in Fig. 1.

Figure 3.

Profile plots for bacterial growth (BG) based on estimated marginal means and the two rDOM levels, plotted against the salinity treatments. Marginal means are the mean response of bacterial growth adjusted for all other variables in the model. Categories show temperature (°C), salinity (g kg−1) and rDOM (μM) treatment level. Error bars show ± the 95% confidence interval.

rDOM influenced the abundance of bacteria with a marked positive effect, similar to the size of the effect of temperature on respiration, showing a partial η2 value of 69% (Table 2, Fig 4). However, the increased temperature resulted in a significantly lower bacterial abundance, although with a clearly lower effect size (partial η2 = 0.26). Salinity did not have a significant effect, although the P-value was relatively low (Table 2).

Figure 4.

Primary data on bacterial abundance (mean ± 95% confidence interval). The x-axis is marked as in Fig. 1.

The number of flagellates varied between 0.64 × 104 (±0) and 2.22 × 104 (±2295) cells dm−3 (average ± 95% confidence interval, Fig. 5). There was a significant increase in flagellate abundance associated with the increased temperature (P = 0.001; partial η2 = 0.782). Additionally, the effects of the temperature × salinity interaction (P = 0.014; partial η2 = 0.55) and the interaction of all three factors (P = 0.016, partial η2 = 0.54) were significant. The profile plot showed a complex pattern in which the flagellate abundance decreased with increasing salinity at the in situ temperature, while the flagellate abundance in the 300 μM rDOM treatment was increased under a higher salinity and higher temperature. Treatments with both high temperature and rDOM showed a high final flagellate abundance. The observed heterotrophic flagellates were generally small (2.5–3 μm ∅) and of the same size. Flagellate respiration was calculated as 4 ± 0.3% (average ± 95% confidence interval) of the measured microbial respiration.

Figure 5.

Primary data on flagellate abundance (mean ± 95% confidence interval). The x-axis is marked as in Fig. 1.

The phytoplankton biomass was 4.81 (±1.9) μg carbon dm−3 on average across all treatments, and no significant treatment effect was observed (data not shown). This value corresponded to 6.7% of the in situ level at the time of sampling. The phytoplankton community was dominated by chlorophytes (mainly Chlorococcales spp. and Monaoraphidium spp.) and different diatoms (Cheatoseros spp., Pennales spp. and Nitzschia spp.). Applicable functions for calculating phytoplankton respiration were not found in the literature.

The total organic carbon level at the beginning of the experiment was 305 (±10) μmol dm−3 in the low-rDOM treatments (Table 3). In the high-rDOM treatments, the DOC level varied from 541 (±0) to 613 (±5). The DOC levels increased slightly over time. The total phosphorous level showed no difference between the treatments at time zero and was 0.13 μmol dm−3 on average. The phosphorous levels decreased during the first 24 h by 50% to 30% (in a gradient from treatments 1–8) and then remained at that level for the rest of the experiment. The total nitrogen level differed between the low-rDOM and high-rDOM treatments and varied from 8.3 (±0.4) to 15.7 (±0.1) at the beginning of the experiment. The nitrogen level remained stable over time in the different treatments (Table 3). The C : N : P ratio was 4914 : 131 : 1 on average for all treatments.

Table 3. Nutrients and TOC measured in the different treatments and at different sampling times (average ± standard deviation)
TreatmentTOCTOT-NTOT-P
0 h24 h72 h0 h24 h72 h0 h24 h72 h
  1. TOC, total organic carbon (μM); TOT-N, total nitrogen (μM); TOT-P, total phosphorus (μM).

1305 ± 17316 ± 0325 ± 08.3 ± 0.49.1 ± 0.19.4 ± 0.10.12 ± 0.030.06 ± 0.30.07 ± 0.03
2305 ± 17314 ± 5316 ± 08.3 ± 0.49.0 ± 0.19.2 ± 0.20.12 ± 0.030.06 ± 0.30.07 ± 0.03
3305 ± 10314 ± 10327 ± 58.3 ± 0.19.0 ± 0.19.5 ± 0.10.13 ± 0.030.06 ± 0.30.07 ± 0.03
4305 ± 10336 ± 13327 ± 58.3 ± 0.19.1 ± 0.19.2 ± 0.10.13 ± 0.030.07 ± 0.030.07 ± 0.02
5541 ± 0566 ± 17572 ± 514.1 ± 0.115.3 ± 0.415.2 ± 0.30.13 ± 0.020.08 ± 0.010.09 ± 0.01
6541 ± 0633 ± 14575 ± 814.1 ± 0.115.4 ± 0.115.5 ± 0.70.13 ± 0.020.09 ± 0.010.08 ± 0.02
7613 ± 5580 ± 5630 ± 515.7 ± 0.116.3 ± 0.416.6 ± 0.30.15 ± 0.020.10 ± 0.010.10 ± 0.02
8613 ± 5647 ± 21641 ± 815.7 ± 0.116.3 ± 0.116.2 ± 0.20.15 ± 0.020.10 ± 0.010.10 ± 0.01

Discussion

Elevated temperature increased microbial respiration rates and flagellate numbers but did not have such an effect on the other measured variables (Table 2, Figs 1 and 5). Temperature explained the major part of the variation observed for both respiration (partial η2 = 0.61) and flagellate abundance (partial η2 = 0.78). The effect of temperature was therefore dominant for these variables when examined in concert with rDOM and salinity and at the levels of all three parameters chosen to be relevant for projected climate change. However, rDOM also had a significant impact on respiration, albeit with a lower effect size (partial η2 = 0.15). Some of the carbon substrates in the rDOM therefore appeared to be capable of promoting respiration, even without experimental pretreatment with light irradiance. It is possible that the demonstrated effect of rDOM suggests a shift towards substrate limitation of respiration at higher temperatures. This shift may have been partly associated with increased bacterial growth, as positive effects of rDOM on bacterial growth and biomass were observed (Table 2 and Figs 2-4). Taken together, our findings indicate that there was no positive effect of temperature on bacterial growth and biomass, while for respiration, no influence of salinity could be demonstrated (i.e. different controlling factors). We infer from this observation that the control of respiration was partly uncoupled from bacterial biomass growth and may be based on a distinct pool of carbon substrates (Russel & Cook, 1995).

The observed effect of temperature on microbial respiration was likely primarily associated with bacterial respiration. Flagellate respiration constituted a small portion of the total respiration (4%), and few phytoplankton occurred in the dark-incubated samples. The abundance of phytoplankton in the experimental bottles was 6.7% of the in situ value. The in situ CO2 fixation at the time and depth of sampling was 0.83 μg C dm−3 day−1. Calculating phytoplankton dark respiration by assuming it to be equal 20% of the day-light CO2 fixation (Flynn, 2005) indicated a negligible contribution to the estimated microbial dark respiration (0.02%). However, we cannot exclude the possibility that some of the measured dark respiration was due to the presence of mixotrophic protozoa and picocyanobacteria.

The positive effect of rDOM on bacterial growth and abundance may be explained by both the higher concentrations of carbon substrates and the concomitant supply of nitrogen and phosphorus. Phosphorus is the major limiting nutrient in the area (Zweifel et al., 1993; Table 3, Figs 2 and 4). The occurrence of phosphorus limitation in the experimental bottles was also supported by the clear excess of carbon and nitrogen relative to phosphorus (4914 : 131 : 1), compared with a typical bacterial C : N : P cellular composition of 45 : 10 : 1 (Neidhardt et al., 1990, table 3). The small fraction of added phosphorus may still have supported anabolic metabolism to allow increase in bacterial biomass. The significant positive effect of an elevated salinity on bacterial growth suggested the existence of some physiological effect facilitating the bioavailability and/or assimilation of refractile substrates in the rDOM. In fact, the simultaneous interactive increase in rDOM and salinity observed with an increase in bacterial growth was the only significant interaction detected (Tables 2 and 3, Fig. 3). This pattern was found at all parameter combinations, except for at 300 μM DOM concentration and the in situ temperature. One explanation for this finding may be that the process transforming refractile DOM to bioavailable DOM proceeded too slowly when the temperature and rDOM were at their lowest levels. Salinity-facilitated transformation may have been particularly important for promoting the bioavailability of the expected limiting nutrient (phosphorus), and thereby bacterial growth. Phosphorus is typically found in association with organic and colloidal matter, possibly unavailable to enzymes (Lin et al., 2012). The observation that approximately half of the total phosphorus had been lost at 24 h indicated that some sedimentation of P or attachment of this nutrient to the walls of the experimental bottles was occurring.

Field data from the literature were in agreement with the observed increase in bacterial growth and abundance associated with added rDOM and an elevated salinity in our experiment (Figs 2-4). Wikner et al. (1999) showed that there was increased bacterial growth based on allochthonous DOC under estuarine (i.e. more saline) compared with limnic conditions in the study area. Stepanauskas et al. (1999) also demonstrated that enhanced utilization of organically bound riverine nutrients stimulated bacterial growth at a higher salinity. Bacterial growth rates also increased upon the exposure of bacteria and substrate material from a riverine environment to a higher salinity (Langenheder et al., 2003). However, our results did not consistently show a clear interaction effect of temperature and rDOM, as proposed by Pomeroy & Wiebe (2001). A tendency towards increased bacterial growth under an increased rDOM level at the lower temperature was found at a 1 g kg−1 salinity, but not at 3 g kg−1. The difference in our results compared with those of Pomeroy & Wiebe (2001) may primarily be because our experiment was performed at summer temperatures, while their conclusions were primarily relevant to winter temperatures in temperate waters of the northern hemisphere. Their conclusion was that elevated DOM levels primarily maintained bacterial activity at a decreasing temperature close to the freezing point.

The effect of temperature on respiration was strongest during the first 24 h of the experiment and then declined (Fig. 1). This development may be explained by the adaptation of the microbial community to the elevated temperature, a gradual depletion of substrates or a combination of the two factors. The positive effect of temperature on respiration is rather well established and has been observed previously in both in situ and mesocosm experiments in estuarine, coastal and open waters (Hopkinson & Smith, 2005; Robinson & le B Williams, 2005; Hoppe et al., 2008; Wohlers et al., 2009; Kritzberg et al., 2010). The calculated Q10 values (median, 26) are considered high, but similar levels were obtained in a field study in this area (J. Wikner, unpublished data). Additionally, other researchers have reported high Q10 values above 20, especially for bacterial growth in cold waters (Pomeroy & Deibel, 1986; Pedros-Alio et al., 2002; Vaque et al., 2009). However, our findings are novel given the applied experimental design, which allowed simultaneous quantification of the relative importance of three co-occurring environmental drivers. Similarly high Q10 values for respiration and their potential relationship with the high level of organic carbon and low phosphorus found in rDOM have not been previously reported to our knowledge. Furthermore, the differential control of bacterial respiration and growth is not sufficiently understood (cf. Lopez-Urrutia & Moran, 2007), and few direct tests of the Pomeroy–Wiebe hypothesis have been reported (Autio, 1998; Kirchman et al., 2005). Moreover, as far as we are aware, the interactive effect of rDOM and salinity and negative effect of elevated temperature on bacterial growth have not been reported previously.

Under a climate-change scenario in which temperature increases over a long period of time, it could be anticipated that the temperature effect would have a smaller impact because organisms would have time to adapt and evolve in the gradually warming environment. However, we cannot exclude the possibility that a continuous supply of appropriate substrates for the respiratory enzymes under natural conditions would also maintain a higher respiration rate at an elevated temperature over a longer period of time. In support of this notion, a mesocosm study showed that the effects of temperature on respiration may last for at least 80 days (Hoppe et al., 2008).

The lower bacterial abundance observed in the high-temperature treatment (Fig. 4) may be partly explained by the more rapid growth of bacterivorous flagellates at higher temperatures, resulting in a higher predation on bacteria. The significant effect of temperature on flagellate abundance was in accord with this explanation (Fig. 5, Table 2). A positive effect of temperature on bacterial grazing has also been reported in field studies conducted in the north-east Atlantic and Antarctica (Marrase et al., 1992; Vaque et al., 2009). The net outcome of increased temperature in our study was therefore negative for bacterial abundance, possibly due to a stronger positive effect of temperature on the numerical response of bacterivores.

Increases in microbial respiration, bacterial growth and bacterial abundance caused by an elevated rDOM level were not obviously expected (Tables 2 and 3), as the rDOM in the study area consists of approximately 80% humic substances (i.e. it is not an ideal carbon source for bacteria) (Pettersson et al., 1997). However, 20% of the rDOM consists of other compounds that are bioavailable, and it includes nutrients other than carbon, which may have primarily supported the observed bacterial growth (Wikner et al., 1999, table 3; Stepanauskas et al., 2002). Using rDOM instead of defined compounds was relevant, as our aim was to study the effect of increased river discharge and the combined effect of carbon, nutrients and trace elements on microbial activity. However, the nitrogen or phosphorus levels alone showed no significant effect on any of the examined variables. This finding was in accordance with the results of Kragh et al. (2008), who found that increase in bacterial growth only occurred under a 100-fold increase of the phosphate concentration compared with typical summer levels.

We may have underestimated the importance of rDOM for microbial activity due to the exclusion of light. The light exclusion condition was chosen to limit the number of total sample bottles required, simplify handling and the statistical analysis and focus on effects based on already photolysed compounds in the rDOM pool. Exposure to light may release carbon, nitrogen and phosphorus from substrates, making them available for bacteria (Bertilsson et al., 1999; Smith & Benner, 2005; Kragh et al., 2008). This means that some carbon compounds made bioavailable by photolysis prior to sampling would have contributed to growth in our experiment but would be depleted over time. Additionally, substrates produced by phyto- and zooplankton were not present in the experimental system, which could cause the importance of the cometabolism of refractile compounds to be underestimated (Münster & De Haan, 1998).

Assuming that our findings are applicable under a future changed climate, both increased microbial respiration and bacterial growth are expected, although driven by different environmental factors. Regarding bacterial abundance, no change was obvious in this study because the rDOM stimulated abundance, while temperature reduced it, possibly due to increased grazing pressure from bacterivores. The net outcome of these conditions is markedly increased respiration in the food web, which may extend already existing hypoxic areas and create new oxygen minimum zones especially in estuarine areas. Verifying our assumption of applicability to future climate change, however, will require a better understanding of the adaptive and evolutionary ecology of microorganisms.

Acknowledgements

Experimental analysis of nutrients and dissolved organic carbon was performed at Umeå Marine Sciences Centre. B.Sc. Jonas Forsberg conducted the taxonomic analysis of phytoplankton. We are grateful for the constructive comments by several reviewers. Funding was provided by the Kempe foundation (SMK-2458) to JW.

Ancillary