The Arctic region is experiencing the steepest warming rate on Earth, warming three times faster than the global mean (ACIA, 2004; Trenberth et al., 2007), and resulting in an abrupt reduction in ice cover, that exceeds the range of natural variability over the past millennia (Walsh, 2008). This rapid warming is expected to continue in the future, with more than 6°C warming expected throughout the twenty-first century under some scenarios (ACIA, 2004).

Climate warming is largely derived from anthropogenic CO₂ emissions, which have raised atmospheric CO₂ from pre-industrial levels of 280 ppm (Caldeira and Wickett, 2003) to close to 400 ppm at present (Forster et al., 2007), and with expected atmospheric CO₂ values between 700 and 1000 ppm by the end of this century (Meehl et al., 2007). The anthropogenic perturbation of CO₂ propagates to the ocean through air-sea exchange leading to increased CO₂ in the oceans and to an associated decline in pH (e.g., Caldeira and Wickett, 2003), with an overall decrease of 0.1 pH units since the pre-industrial period (Doney et al., 2009). Further emissions of fossil-fuel CO₂ into the atmosphere could lead to a reduction of 0.4 pH units in the ocean by the end of this century (Caldeira and Wickett, 2003). The largest pH decrease projected worldwide is expected in the Arctic Ocean (−0.45 pH units), where sea ice retreat would amplify air to sea carbon flux. In fact, the dilution of DIC by freshwater lowers the CO₂ partial pressure, that, in combination with an increase in ice-free area, enhances uptake of anthropogenic carbon through gas exchange (Steinacher et al., 2009).

Phytoplankton communities, which represent the basis of the Arctic food web, can be affected by both temperature increase and increased CO₂. Hence, predicting the responses to these concurrent changes is required to understand the vulnerability of the Arctic ecosystem to global change. Phytoplankton species growing in polar oceans are adapted to the low temperatures characteristic of these regions, and have been shown to respond to even small increases in temperature (Neori and Holm-Hansen, 1982; Li et al., 1984; Holding et al., 2013; Coello-Camba et al., 2014). Temperature plays a fundamental role in the regulation of metabolic processes (Iriberri et al., 1985; White et al., 1991), causing an increase in algal net growth rates with increasing temperatures (Eppley, 1972; Goldman and Carpenter, 1974; Goldman, 1977; Redalje and Laws, 1983; Montagnes and Franklin, 2001). Phytoplankton responses to warming depend on species characteristics, such as cell size, which also constrain key phytoplankton functions (Malone, 1980).

Increased CO₂ stimulates phytoplankton growth rates (Kim et al., 2006) and primary production (Hein and Sand-Jensen, 1997), and affects the pigment content in the cells (Xia and Gao, 2005) and the species composition of the community (Tortell et al., 2002). Whereas the effects of increased CO₂ levels on phytoplankton have been examined in several oceanic regimes (i.e., Hein and Sand-Jensen, 1997; Tortell et al., 2008; Beardall et al., 2009), results for polar communities, where low temperatures significantly increase CO₂ solubility and lead to equilibrium CO₂ concentrations more than two times higher than those of tropical waters (Tortell et al., 2008), are still few and inconclusive (Silyakova et al., 2012; Engel et al., 2013; Tanaka et al., 2013). Moreover, warming and increased CO₂ co-occur and, thus, their effects should be studied in concert. No study examining the response of Arctic phytoplankton communities to warming and increased CO₂ has been conducted to-date.

Here we examine the response of Arctic phytoplankton to an experimental increase in seawater temperature together with CO₂. We did so through a factorial experiment, allowing the elucidation of interactions, synergies or compensatory effects, between these drivers. The response of the phytoplankton community was characterized by their community structure, abundance, growth, primary production, and cell viability.

A large volume (about 840 L) of Arctic open water was sampled just outside Isfjorden (78°N 13°E) using a water pump on board the vessel “Viking Explorer” on 23 June 2010 with an in situ temperature of 1.4°C and a salinity of 33.5. The water was distributed in 60 L polypropylene containers previously treated with HCl 0.1 N for at least 48 h and thoroughly rinsed with seawater from the sampling station. Seawater containers were immediately transported to the controlled temperature chambers of UNIS (The University Centre in Svalbard) in Longyearbyen (Svalbard, Norway), where the experiment took place, from 24 June to 8 July 2010.

The incubation temperatures set for each treatment remained very stable along the experiment (Table 1). However, pCO₂ values experienced important variations during the first week of incubation, particularly in the experiments subject to warming, with a simultaneous time lag observed until the first clear responses, i.e., in chlorophyll a concentration or cell abundances. As a result, the pCO₂ values for “present pCO₂” and “increased pCO₂” treatments conformed closely to the target values at the 1°C treatment, but warming of the water mass to 6°C and 10°C, together with increased microbial activity in these warmer treatments, led to median pCO₂ values in those treatments well beyond the target values. These difficulties, largely derived from biological feedbacks of the plankton community, are consistent with those experienced in past experiments regulating CO₂ levels in Arctic plankton communities (e.g., Torstensson et al., 2012; Engel et al., 2013), as the metabolism of these communities interfere with the treatment variable (Duarte et al., 1997).

The initial average chlorophyll a concentration was 0.64 μg chl a L⁻¹. The Arctic fjord phytoplankton community present was classified at genus level or defined according to cell sizes for those phytoplankton groups that could not be identified. At the beginning of the experiment the Arctic fjord phytoplankton community included centric diatoms (4 and 7 μm average diameter), and photosynthetic flagellates of three size classes (average diameter 5.9, 12.5, and 20 μm). The most abundant groups were flagellates 5.9 μm and centric diatoms 7 μm in diameter, comprising up to 95% of the cells present. Along the incubation two more groups appeared in the cell counts, a group of centric diatoms (11.1 μm in diameter) and the flagellate Pyramimonas sp. (6 μm).

After an initial decrease in all treatments, the chl a concentration increased along the second week of the experiment (Figure 1, Table 2), reaching higher values at the lower (1.82°C) treatments than those at higher temperature (6.31°C and 10.32°C, p < 0.01). These differences were particularly clear at the end of the experiment (Figure 2), when chlorophyll a concentrations were significantly higher (p < 0.05) at the 1.82°C treatment, specially at the low pCO₂ treatment, than at the other two temperatures. In fact, this change in chl a concentration at the end of the experiment showed a three-fold increase in the community growing at the lowest temperature, with smaller values under high pCO₂ (Figure 2). Growth rates estimated from chlorophyll a concentrations (Figure 3), showed a negative and significant response (p < 0.05) to increased temperature (Table 2). When referred to pCO₂ increase, no significant differences were found between treatments, despite the slightly lower values found under high pCO₂ conditions (Table 2), and there was no significant interaction between growth rate and temperature and pCO₂ (t-test, P > 0.05, Table 2). As observed for chl a concentration, community biovolume declined with increasing temperature (Table 2, Figure 4) and also decreased, although not significantly, with increasing pCO₂ (Table 2, Figure 4). A compensatory effect was found, as reflected in a significant positive interaction between warming and increased CO₂ on community biovolume (Table 2). The community also showed important changes, with large centric diatoms (11.1 μm in diameter) becoming dominant at low temperature at the end of the experiment, along with large photosynthetic flagellates (20 μm) (Figure 4). The community under the warmest temperature was dominated by the smallest photosynthetic taxa (Figure 4). At the end of incubation the biovolumes of flagellates 20 μm, centric diatoms 11.1 μm, and total community biovolume were lower at higher temperature treatments (p < 0.01) (Figure 4). For centric diatoms 4 μm, biovolume at the end of the experiment was higher at the 6.71°C treatment (p < 0.05) than at 1.82 and 10.32°C (Figure 4).

Primary production increased in all treatments over the course of the experiment from 23.68 ± 9.25 mg C L⁻¹ d⁻¹ on day 3 to 91.80 ± 11.53 mg C L⁻¹ d⁻¹ on day 14 (values averaged across all treatments ±SE). Primary production decreased significantly with increasing temperature (p < 0.05, Table 2) with this decline being steeper under high pCO₂ (Figure 5). At 10.31°C primary production rates were higher at low pCO₂ than at higher pCO₂ (p < 0.05), without significant differences between pCO₂ treatments at lower temperatures. At the end of the experiment, when responses to experimental treatments were strongest, primary production declined significantly with increasing temperature (p < 0.01), with this decline being steeper at high pCO₂ (p < 0.05).

The growth rates estimated from cell biovolume of centric diatoms 11.1 μm in diameter declined strongly with increasing temperature (p < 0.01), while the growth rate of centric diatoms 7 μm in diameter increased with warming. Increased pCO₂ caused different responses in phytoplankton, with a significant growth increase of centric diatoms 7 μm and Pyramimonas sp. with increasing pCO₂ (p < 0.05), and a decline in growth rate of centric diatoms 11.1 μm with increasing pCO_2. A positive and significant interaction between increased pCO₂ and temperature was found for the changes experienced in total biovolume (p < 0.05), and more specifically in the biovolume of centric diatoms 11.1 μm (p < 0.05) (Table 2).

The average initial percentage of living cells in the community was estimated at 75.3% for centric diatoms 7 μm and 43.1% for flagellates 5.9 μm, the dominant groups at the beginning of the experiment. The living cell abundance increased with temperature and under higher pCO₂ levels for centric diatoms 7 μm (p < 0.05), but the presence of both factors resulted in a slight antagonistic interaction (p < 0.05) (Table 3). We also observed that the abundance of living flagellates 5.9 μm showed a negative but not significant response with increased temperature (Table 3). The effect of increased pCO₂ was opposite but also not significant (p > 0.05) (Table 3).

The results obtained here show that warming should have an overriding effect on Arctic phytoplankton communities relative to that of the increase in pCO₂ expected by the year 2100. The decline in chlorophyll a concentration, biovolume and primary production of the Arctic phytoplankton community with the experimental warming tested is consistent with results from previous experiments with Arctic communities (Coello-Camba et al., 2014). Moreover, where increased CO₂ had significant effects these involved a steeper decline in primary production with warming than observed at lower pCO₂, although phytoplankton biovolume increased with CO₂ at warmer temperatures.

The response of phytoplankton to the experimental conditions was slow as observed before in arctic communities (Silyakova et al., 2012; Holding et al., 2013; Coello-Camba et al., 2014). The kind of experiment performed here, using microcosms, cannot be run for a long time, but the results indicated that the stationary phase was not reached. This suggests that phytoplankton biomass may still increase, especially in response to the pCO₂ conditions. In any case, what was clearly shown in the results was the shift in phytoplankton communities.

Community structure shifted from a dominance of large species to communities dominated by phytoplankton species with smaller cell sizes at warmer temperatures. This is consistent with the observation that warming is associated with a decline in phytoplankton cell size (Agawin et al., 2000; Hare et al., 2007; Li et al., 2009; Morán et al., 2010). This decrease in cell size with warming is also consistent with the responses reported for previous temperature manipulations of Arctic plankton communities, where an increased growth of small-sized species, as Micromonas sp., was observed with experimental warming (Coello-Camba et al., 2014).

The physiological adaptation of phytoplankton species to cold polar waters may have narrowed their thermal windows (Smetacek and Nicol, 2005), leading to the negative effects of warming observed here. Previous experiments with Arctic communities showed an optimum temperature for cell growth around 5–6°C (Coello-Camba et al., 2014) suggesting that the warmer temperatures tested here may be beyond their thermal optimum. This temperature optimum may differ between species (Jacques, 1983), which are therefore expected to show a wide range of responses to warming, as shown by the diversity of responses of the different functional groups present in the community tested here.

In contrast to the general decline in chl a concentration and growth with warming, responses to increased pCO₂ were modest, with relatively small effects on chlorophyll a concentration, total community biovolume and net growth rates, and varied among functional groups. Previous experiments have also documented small effects of increased CO₂ on marine phytoplankton growth rate and/or community structure (Hare et al., 2007; Riebesell et al., 2007; Engel et al., 2008). Also, previous studies testing the effect of increased CO₂ on primary production have shown contrasting results, ranging from increased primary production (Hein and Sand-Jensen, 1997; Tortell et al., 2002; Engel et al., 2013) to little or no effect of CO₂ increase, particularly for studies examining the response of Arctic phytoplankton to increased CO₂ (Thingstad et al., 2008; Engel et al., 2013). The studies showing a positive response of phytoplankton to elevated pCO₂ explain this response as a consequence of the lower energetic cost of C assimilation (Tortell et al., 2002, 2008). In contrast, however, the responses to increased CO₂ observed here were non-intuitive, involving a decline in growth and biomass, in contrast to the positive effects expected. Such non-intuitive responses to increased CO₂ have been shown for Arctic phytoplankton in the past, interpreted as the results of enhanced competition for inorganic nutrients between phyto- and bacterioplankton with a consequent decrease in net community production at high pCO₂ (Thingstad et al., 2008; Engel et al., 2013). Indeed, the effects of increased pCO₂ in the community vary depending on other factors, such as nutrient availability (Tortell, 2000; Raven et al., 2005), and other factors that could affect the operation of carbon concentration mechanisms (Wolf-Gladrow et al., 1999). However, the EPOCA experiment involved nutrient additions and yet, phytoplankton did not show a positive response to increased CO₂ there (Engel et al., 2013). Those effects may also be species-specific, so would vary depending on the community composition (Tortell et al., 2002; Riebesell, 2004; Kim et al., 2006; Blackford and Gilbert, 2007; Fu et al., 2007). However, most phytoplankton species have active uptake systems for inorganic carbon that renders photosynthetic carbon assimilation essentially saturated under current CO₂ levels, so it is possible that elevated CO₂ conditions could not result in an increase on their assimilation rates (Giordano et al., 2005).

As for temperature increase, increased pCO₂ has also been observed to cause an increase in the abundance of small sized cells (Brussaard et al., 2013). In our experiment, although no pico-sized cells were observed, Pyramimonas sp. and centric diatoms 7 μm experienced a significant increase in their biovolume under increased pCO₂ while large centric diatoms showed a negative response to elevated CO₂, supporting a size-dependent response of phytoplankton species to elevated CO₂.

Whereas, past experiments have addressed the responses to warming (Holding et al., 2013; Coello-Camba et al., 2014) and to elevated CO₂ (Thingstad et al., 2008; Engel et al., 2013), our results provide the first assessment of interactive effects of warming and elevated CO₂. However, interactions were generally weak. Elevated CO₂ was observed to lead to steeper decline in primary production with warming. In contrast, the interaction between warming and elevated CO₂ had a small, but significant compensatory effect on the decline in community biomass (as biovolume) observed with warming and elevated CO₂. This compensatory effect offset, and slightly reverted, the effects of elevated CO₂ under warm conditions.

Our analysis also provided insight onto the effects of warming and increased CO₂ on cell status, assessed as cell viability. Previous papers have shown that the cell viability measured using this kit reflects the actual preferences of the cells on a specific environment, and that these preferences are not affected by other external factors that could be acting upon them (Alonso-Laita and Agustí, 2006; Lasternas and Agustí, 2010; Lasternas et al., 2010). Our results showed contrasting effects of temperature and pCO₂ on the presence of living cells in the species studied. This supported the results obtained for the changes in net growth rates, reflecting the specificity of the responses that can be found on a natural community composed of different cell types with diverse ecologic and metabolic characteristics. Both for centric diatoms 7 μm and flagellates 5.9 μm the changes in the percentage of living cells paralleled the changes in cell growth under increased temperature and CO₂, indicating that these responses reflect cellular-level responses to these factors, rather than indirect responses (i.e., due to changes in cell competition or predation).

The combined effects of different environmental factors on phytoplankton communities has been studied in the past (i.e., Hare et al., 2007; Echeveste et al., 2011), but despite the future increasing trends in temperature and pCO₂ predicted, there have been few explicit tests of the combined effect of both factors on phytoplankton communities. Previous studies from other oceanographic provinces considering both factors reported contrasting responses. Hare et al. (2007) reported that changes in photosynthetic rates in the Bering Sea were mainly driven by elevated temperature with little effect of increased pCO₂, while both factors were shown to have additive effects on the responses of the phytoplankton community in the North Atlantic (Feng et al., 2009). Moreover, the responses to warming and CO₂ are often species-specific, as observed by Fu et al. (2007), consistent with the contrasting responses between large and small components of the community examined here.

In summary, our results suggest that future increases in water temperature and pCO₂ would lead to a decrease in the community chl a concentration and biomass in the present Arctic phytoplankton communities examined, also leading to communities dominated by smaller nano-phytoplankton groups, with important consequences for the flow of carbon and food web dynamics. Whereas the effect of warming is stronger than those of pCO₂, increased pCO₂ does not offset the effects of warming. Our results point to a less productive Arctic Ocean in the future.