Thermal discharge-created increasing temperatures alter the bacterioplankton composition and functional redundancy

Climate warming has led to an accelerating pace of elevated seawater temperature (EST) (Brown et al. 2004), which is evidenced by long-term observations and prediction model (Pachauri et al. 2014; Vezzulli et al. 2012). The EST has imposed negative effects on ocean ecosystems (Hoegh-Guldberg and Bruno 2010), such as expanded range of red tides (McLeod et al. 2012), enhanced mass mortality of benthic organisms (Coma et al. 2009), and increased virulence of bacterial pathogens (Bally and Garrabou 2007). The indoor mesocosms studies have found that warming tightens the coupling between phytoplankton production and bacterial respiration, thus leading to an increased recycling of organic matter (and CO₂) (Hoppe et al. 2008; Scheibner et al. 2014; Wohlers et al. 2009). It is now recognized that the biological carbon pool of the oceans largely relies on the balance between primary production and microbial respiration (Calleja et al. 2005), which is one of the major uncertainties in predicting future climate warming. However, despite its importance, the knowledge on how bacterioplankton community responds to ocean warming is scarce.

Long-term observational data have shown that ocean warming significantly alters microbial community composition (Hoegh-Guldberg and Bruno 2010; Vezzulli et al. 2012). However, this approach is restricted because there are few long-term data sets available. Alternatively, laboratory manipulation of temperature has been adopted to address warming effects on microbial activities, while the conclusions are contradictory (Hoppe et al. 2008; Sarmento et al. 2010; Wohlers et al. 2009). For example, bacterial respiration is consistently accelerated (Hoppe et al. 2008), or unaffected (Wohlers et al. 2009) along seawater temperature gradient. Interestingly, there are similarly confusing results regarding the structures of bacterioplankton community, with some cases showing thermal sensitivity (Dziallas and Grossart 2011; Scheibner et al. 2014), and others not (Shade et al. 2012b). It is likely that the direct warming effects are overridden by other abiotic and/or biotic factors, e.g., warming-induced changes in nutrient levels and grazing rate. Consistently, a few studies have shown that predation pressure is the major driver of differences in bacterial assemblages (Scheibner et al. 2014; Zöllner et al. 2009). However, the stochastic nature of hydrology, i.e., daily tide and water flow, could create pronounced homogeneity of microbial communities (Finlay 2002), consequently modify or even obscure the warming effects. Nevertheless, to predict the consequences of ocean warming on microbial assemblages, analyses of natural communities in situ environmental conditions might yield different, and perhaps more reliable, inferences of underlying temperature relationships than the responses derived from simulation experiments. Indeed, it is likely that the microbial communities respond differentially to natural and experimental seawater warming (Sarmento et al. 2010). Furthermore, available microcosm and/or mesocosm experiments to date only design one or two elevated temperature levels (Dziallas and Grossart 2011; Lindh et al. 2013; Scheibner et al. 2014). For these reasons, it is unclear whether the responses of the natural bacterioplankton community are temperature dependent, mirroring what has been observed in artificial experiments (Hoppe et al. 2008). Given the functional importance of microbial community in biogeochemical cycling (Azam and Malfatti 2007), the linkage between microbial composition and function have received considerable attention. However, no consensus has been emerged, with some studies show a close interplay between microbial community composition and ecosystem function (Logue et al. 2016; Wu et al. 2015; Xiong et al. 2015b), while others demonstrate weak (Lindström et al. 2010), or inconsistent (Freimann et al. 2013; Langenheder et al. 2005) associations. A plausible explanation for these discrepant patterns is that microbial communities exhibit functional redundancy (different microbial communities execute a functional process at the same rate) (Allison and Martiny 2008). Findings from both field studies (Vezzulli et al. 2012) and laboratory experiments (Dziallas and Grossart 2011; Scheibner et al. 2014) have shown, for example, that warming exerts a strong filtering effect on bacterioplankton assemblages (Adams et al. 2010). Changes in bacterial composition, in turn, could alter functional redundancy (Allison and Martiny 2008). Recently, functional redundancy has been applied as an index to evaluate the buffering capacity of a microbial community to changing environments (Miki et al. 2014; Wilhelm et al. 2015). Therefore, to better understand the functional responses of ecosystem to EST, it is required to quantify to what extent that the functional redundancy of bacterial communities is altered by ocean warming.

Here we explore bacterioplankton communities within the temperature flume of a coal power plant in Xiangshan Bay, China. Thermal discharge has resulted in elevated temperatures from 15.0 to 18.6 °C in a local scale (Additional file 1: Table S1), thereby facilitating the predicted range of future temperature predictions for the next century (Pachauri et al. 2014). Thus, the temperature gradient design provides an ideal model to delineate real responses of bacterioplankton community to the ongoing ocean warming. Although this system is not exact proxy for global ocean changes, it can offer some in situ insights into bacterioplankton responses. A comprehensive study has shown that the successions of diatomaceous and macrophytic populations are linearly associated with the temperature gradient created by the thermal flume of a nuclear power plant (Hillebrand et al. 2010). In addition, mesocosms studies consistently show that elevated temperatures promote bacterial respiration (Hoppe et al. 2008; Scheibner et al. 2014; Wohlers et al. 2009), suggesting that bacterial functional redundancy is low in response to warming. For these reasons, we hypothesize that increasing temperatures would select specialist species (either increase/decrease or replace one another), resulting in an incremental dissimilarity and decreased functional redundancy of bacterioplankton community. If this is the case, specialist assemblages can be identified to indicate this responsive pattern.

It is commonly accepted that marine biota is under the threat of ocean warming (Coma et al. 2009; Pachauri et al. 2014). However, studies of climate warming in oceans have lagged behind that of terrestrial ecosystems (Hoegh-Guldberg and Bruno 2010). Here, to mimic ocean warming effects on bacterioplankton community, we collected a unique dataset from a temperature gradient created by thermal discharge. The bacterial diversity, composition and functional redundancy were substantially altered by increasing temperatures, while the variation was primarily controlled by the indirectly warming effects. Our findings were congruent with ecological theory positing relationships between disturbance intensity and microbial response (Allison and Martiny 2008; Shade et al. 2012a), mirroring what has been observed for macroorganisms (Laliberte et al. 2010).

Thermal discharge substantially stimulated bacterial density and grazing rate (Additional file 1: Figure S1), which is in accordance with the notion that the growth rates of protists increase to a similar degree as those of bacteria with increasing temperatures (Rose and Caron 2007). However, both long-term observations and theoretical predictions point towards a decrease in bacterial abundance with increasing temperature (Brown et al. 2004; Sarmento et al. 2010). This discrepancy could be due to the ecosystem complexity. In oligotrophic conditions, warming-stimulated bacterial growth could be constrained by resource limitation (López-Urrutia and Morán 2007), whereas our study bay is highly eutrophic as a result of a typical aquaculture area for three decades (Xiong et al. 2015a). Warming-stimulated nutrient and Chl a levels (Additional file 1: Table S2, as a proxy of phytoplankton primary production) provide available substrates to heterotrophic bacteria, which is supported by the report that elevated temperature diminish the time lag between the peaks of phytoplankton primary production and bacterial degradation (Hoppe et al. 2008). Our findings are in accordance with previous studies wherein an increase in temperature with sufficient substrate supply enable bacterial abundance to temporarily exceed grazing losses (Sarmento et al. 2010; Scheibner et al. 2014). Thus, warming stimulated rate of bacterial proliferation is faster than that of grazing, resulting in an increased biomass flux from bacteria to grazer (Additional file 1: Figure S1).

Given the huge population sizes and rapid dispersal abilities of microbiota (Finlay 2002), the true effects of temperature on BCCs could be masked or modified by immigrants from nearby localities. However, we found temperature was an important factor in constraining bacterial diversity and composition (Fig. 2b; Additional file 1: Figure S1), though only a minor variation was directly explained by temperature (Table 2; Fig. 4). One possible explanation for this pattern is that water flow and tide lead to a stochastic colonization by dispersed bacteria (Shade et al. 2012a). Consistently, about half of the variation was unconstrained by the selected variables (Fig. 4). Alternatively, the dimensionality reduction procedure loses some information, leading to a relatively low explained variation (depend on the complexity of the original data) (Digby and Kempton 2012). Nevertheless, a minor direct warming effect has been repeatedly detected on soil bacterial composition, whereas the predominant variation is controlled via indirect warming effects on biotic and geochemical variables (Xiong et al. 2014a; Zhou et al. 2012). Consistently, we found that grazing rate alone constrained a high percentage (12.8 %) of the community variation (Fig. 4). In addition, a significant distance-decay for similarity relationship was detected (Fig. 3), which is in concert with the result that geographic distance contributed the highest variation of BCCs (Fig. 4). Taken together, the spatial distribution of bacterioplankton community is not random, which is subjected to temperature departures by thermal discharge.

Subsequently, we asked whether sensitive bacterial assemblages are indicative for temperature gradient. As a result, 24 bacterial families were screened, whose relative abundances were significantly associated with increasing temperature (Fig. 5). Importantly, for a given bacterial family, the change pattern (linearly increased or decreased along the temperature gradient) of the relative abundance is supported by its known biology and ecology. For example, bacterial families affiliated with Bacteroidetes monotonically increased along the temperature gradient (Fig. 5), indicating that they share the same life strategy, i.e., prefer warming. This assertion is evidenced by the high similarities in the core genomes among many Bacteroidetes species (Fernández-Gómez et al. 2013). Members of this group are able to degrade complex organic substrates, while the remineralization rate is temperature dependent (John et al. 2014). The recruitment of Bacteroidetes group in a warmer ocean may reduce burial rates of organic carbon, thereby contributing a positive feedback to warming. Gammaproteobacteria species are vulnerable to grazing (Beardsley et al. 2003). Consistently, grazing rate was promoted with increasing warming (Additional file 1: Figure S1A), thereby reducing the relative abundances of Gammaproteobacteria assemblages, with the exception of Alteromonadaceae (Fig. 5). Indeed, some families of Gammaproteobacteria, such as Psychromonadaceae and Shewanellaceae, are known to thrive in cold habitats (Hau and Gralnick 2007), evidenced by decreased relative abundances of those families along increasing temperatures (Fig. 5). In contrast, Alteromonadaceae species have relatively high optimal growth temperatures, and are motile by using one polar flagellum (López-Pérez and Rodriguez-Valera 2014), thereby facilitating them escape from prey. It should be noted that some Alteromonadaceae species are fish pathogen (Gudmundsdóttir and Björnsdóttir 2007), thus its enrichment may impose a serious threat to adjacent aquaculture. This assertion is supported by the finding that the abundance of predicted bacterial invasion factors positively correlated (Pearson coefficient, r = 0.347, P = 0.02) with increasing temperature (Data not shown). Warming can selectively promote cyanobacterial growth due to their high optimal growth temperatures (Jöhnk et al. 2008), which is in accordance with our result that the relative abundance of Synechococcaceae was positively associated with increasing temperature (Fig. 5). Similarly, the increased abundance of Synechococcaceae has been detected in warming mesocosms (Scheibner et al. 2014) and long-term observation data (Vezzulli et al. 2012). Overall, changes in the relative abundances of these sensitive bacterial families are temperature dependent, which could indicate temperature departures.

The ultimate goal of ecological studies is to predict how compositional changes of microbial community would alter ecosystem functions (Shade et al. 2012a). However, this interplay is generally confounded by community functional redundancy (Allison and Martiny 2008). To our knowledge, no studies have directly tested how shifts in community assembly affect the functional redundancy. Here, we showed that increasing temperature linearly decreased functional redundancy (Fig. 6), which is in parallel with pronounced changes in BCCs. Thus, increasing temperature has enhanced niche selection, thereby reducing the stochastic processes of birth and colonization of new species (Wu et al. 2015). Consistently, a relevant work shows that disturbances select more phylogenetically convergent bacterioplankton communities, and increase deterministic processes in shaping BCCs (Xiong et al. 2015a). It is likely that less genetically diverse communities lose taxa with complementary response traits after a disturbance (Shade et al. 2012a). If this is the case, increasing temperatures in the thermal discharge would reduce resilience and decrease functional stability of bacterioplankton communities. Currently, our capability to predict the direction of ocean ecosystem to climate change is hampered by the uncertainty on how microbial response, e.g., functional adaptation or change, to warming (Shade et al. 2012a). Our results confirmed the latter case, suggesting that the rates of community adaptation do not keep up with these fast environmental changes. However, the activity of bacteria determines the air-sea exchange of CO₂ via heterotrophic respiration, or carbon sequestration through biological carbon pump and carbon flux to fisheries (Azam and Malfatti 2007). The decreased functional redundancy, in turn, may disrupt the balance of microbial food loop as it was. For this reason, it is mandatory to investigate the ecological consequences of such declines in functional redundancy in further works.

In conclusion, global warming has imposed a major threat to marine ecosystem. Here, we applied thermal discharge-created increasing temperatures to mimic the climate warming effects on the bacterioplankton community. We found that increasing temperatures, combined with substrate supply and trophic interactions significantly altered the BCCs, while the variation was primarily shaped by the indirectly warming effects. In addition, we identified 24 bacterial families that closely associated with increasing temperatures, thus holding the potential to indicate temperature departures. Notably, increasing temperatures reduce the community functional redundancy, which, in turn, may affect the stability of marine ecosystem. For this reason, further studies are needed to evaluate the ecological consequences of increasing temperatures on a time scale.