Effects of elevated CO<sub>2</sub> and temperature on phytoplankton community biomass, species composition and photosynthesis during an autumn bloom in the Western English Channel

The combined effects of elevated pCO 2 and temperature were investigated during an autumn phytoplankton bloom in the Western English Channel (WEC). A full factorial 36-day microcosm experiment was conducted under year 2100 predicted temperature (+4.5 °C) and pCO 2 levels (800 μatm). The starting phytoplankton community biomass was 110.2 (±5.7 sd) mg carbon (C) m −3 and was dominated by dinoflagellates (~ 50 %) with smaller contributions from nanophytoplankton (~ 13 %), cryptophytes (~ 11 %)and diatoms (~ 9 %). Over the experimental period total biomass was significantly increased by elevated pCO 2 (20-fold increase) and elevated temperature (15-fold increase). In contrast, the combined influence of these two factors had little effect on biomass relative to the ambient control. The phytoplankton community structure shifted from dinoflagellates to nanophytoplankton at the end of the experiment in all treatments. Under elevated pCO 2 nanophytoplankton contributed 90% of community biomass and was dominated by Phaeocystis spp., while under elevated temperature nanophytoplankton contributed 85 % of the community biomass and was dominated by smaller nano-flagellates. Under ambient conditions larger nano-flagellates dominated while the smallest nanophytoplankton contribution was observed under combined elevated pCO 2 and temperature (~ 40 %). Dinoflagellate biomass declined significantly under the individual influences of elevated pCO 2 , temperature and ambient conditions. Under the combined effects of elevated pCO 2 and temperature, dinoflagellate biomass almost doubled from the starting biomass and there was a 30-fold increase in the harmful algal bloom (HAB) species, Prorocentrum cordatum . Chlorophyll a normalised maximum photosynthetic rates (P B m ) increased > 6-fold under elevated pCO 2 and > 3-fold under elevated temperature while no effect on P B m was observed when pCO 2 and temperature were elevated simultaneously. The results suggest that future increases in temperature and pCO 2 do not appear to influence coastal phytoplankton productivity during autumn in the WEC which would have a negative feedback on atmospheric CO 2 .

there was a 30-fold increase in the harmful algal bloom (HAB) species, Prorocentrum cordatum. 28 Chlorophyll a normalised maximum photosynthetic rates (P B m) increased > 6-fold under 29 elevated pCO2 and > 3-fold under elevated temperature while no effect on P B m was observed 30 when pCO2 and temperature were elevated simultaneously. The results suggest that future 31 increases in temperature and pCO2 do not appear to influence coastal phytoplankton 32 Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-510 Manuscript under review for journal Biogeosciences Discussion started: 4 December 2017 c Author(s) 2017. CC BY 4.0 License.

Introduction 35
Oceanic uptake of atmospheric CO2 has increased by ~42% over pre-industrial levels, with an 36 on-going annual increase of ~0.4%. Current CO2 level has reached ~400 µatm and has been 37 predicted to rise to >700 µatm by the end of this century (Alley et al., 2007), with estimates 38 exceeding 1000 µatm (Raupach et al., 2007;Raven et al., 2005). The oceans are absorbing CO2 39 from the atmosphere, which results in a shift in oceanic carbonate chemistry resulting in a 40 decrease in seawater pH or 'Ocean Acidification' (OA). The projected increase in atmospheric 41 CO2 and corresponding increase in ocean uptake, is predicted to result in a decrease in global 42 mean seawater pH of 0.3 units below the present value of 8.1 to 7.8 (Wolf-gladrow et al., 1999). 43 Under this scenario, the shift in dissolved inorganic carbon (DIC) equilibria has wide ranging 44 implications for phytoplankton photosynthetic carbon fixation rates and growth (Riebesell, 45 2004). 46 Concurrent with OA, elevated atmospheric CO2 and other climate active gases have warmed the 47 planet by ~0.6 °C over the past 100 years (IPCC, 2007). Atmospheric temperature has been 48 predicted to rise by a further 1.8 to 4 °C by the end of this century (Alley et al., 2007). 49 Phytoplankton metabolic activity may be accelerated by increased temperature (Eppley, 1972), 50 which can vary depending on the phytoplankton species and their physiological requirements 51 (Beardall and Stojkovic, 2006). Long-term data sets already suggest that ongoing changes in 52 coastal phytoplankton communities are likely due to climate shifts and other anthropogenic 53 influences (Edwards et al., 2006;Smetacek and Cloern, 2008;Widdicombe et al., 2010). The 54 response to OA and temperature can potentially alter the community composition, community 55 biomass and photo-physiology. Understanding how these two factors may interact 56 (synergistically or antagonistically) is critical to our understanding and for predicting future 57 primary productivity (Boyd and Doney, 2002). 58 Laboratory studies of phytoplankton species in culture and studies on natural populations in 59 the field have shown that most species exhibit sensitivity, in terms of growth and 60 photosynthetic rates, to elevated pCO2 and temperature individually. To date, only a few studies 61 have investigated the interactive effects of these two stressors on natural populations (e.g. samples. The JVPSII rates were converted to Chl specific carbon fixation rates (mg C (mg Chl a) -1 222 m -3 h -1 ), calculated as: 223 where φE:C is the electron requirement for carbon uptake (molecule CO2 (mol electrons) -1 ), MWC 225 is the molecular weight of carbon and Chl a is the Chl a measurement specific to each sample. 226 Chl specific JVPSII based photosynthesis-irradiance curves were conducted in replicate batches 227 between 10:00 -16:00 to account for variability over the photo-period at between 8 -14 228 irradiance intensities. The maximum intensity applied was adjusted according to ambient 229 natural irradiance on the day of sampling. Maximum photosynthetic rates of carbon fixation 230 (P B m), the light limited slope (α B ) and the light saturation point of photosynthesis (Ik) were 231 estimated by fitting the data to the model by Webb et al., (1974): 232 Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-510 Manuscript under review for journal Biogeosciences Discussion started: 4 December 2017 c Author(s) 2017. CC BY 4.0 License.
Samples for FRRf fluorescence-based light curves were taken at T36. 234

Statistical analysis 235
To test for effects of high pCO2, high temperature and high pCO2 x high temperature on the 236 measured response variables (Chl a, total community biomass, POC, PON, photosynthetic 237 parameters and biomass of individual species), generalised least squares models with the 238 factors pCO2, temperature and time (and all interactions) were applied to the data between T0 239 and T36 incorporating an auto-regressive correlation structure of the order (1) to account for 240 auto correlation. To test for significant differences between experimental treatments at T36 in 241 all measured parameters, generalized linear models were applied to the data. Where main 242 effects were established, pairwise comparisons were performed using the method of Herberich 243 et al., (2010) for data with non-normality and/or heteroscedasticity. Weekly biomass values 244 from the L4 time-series were averaged over years to elucidate the variability and seasonal 245 cycles of the dominant species observed in the experimental community at T36, relative to the 246 time-series observations. The distribution of these species biomass at station L4 was also 247 analysed relative to the in-situ gradients of temperature (1993-2014) and pCO2 (2008-2014) 248 using frequency histograms. Analyses were conducted using the R statistical package (R Core 249 Team (2014). R: A language and environment for statistical computing. R Foundation for 250 Statistical Computing, Vienna, Austria). 251

Results 252
Chl a concentration in the WEC ranged between 0.02-~5 mg m -3 from 30 September -6 th 253 October 2015, with a concentration of ~1.6 mg m -3 at station L4 ( Fig. 1 A). Over the period 254 leading up to phytoplankton community sampling, increasing nitrate and silicate concentrations 255 coincided with a Chl a peak on 23 rd September (Fig. 1 B). Routine net trawl (20 µm) sample 256 observations indicated a phytoplankton community dominated by the diatoms Leptocylindrus 257 danicus and L. minimus with a lower presence of the dinoflagellates Prorocentrum cordatum, 258 Heterocapsa spp. and Oxytoxum gracile. Following decreasing nitrate concentrations, this 259 community transitioned to a P. cordatum bloom on 29 th September, the week before 260 experimental community sampling (data not shown). 261

Experimental carbonate system 262
Equilibration to the target high pCO2 values (800 µatm) within the high pCO2 and combination 263 treatments was achieved at T10 (Fig. 2 A). These treatments were slowly acclimated to 264 increasing levels of pCO2 over 7 days (from the initial dilution at T3) while the control and high 265 Biogeosciences Discuss., https://doi.org/10.5194/bg-2017-510 Manuscript under review for journal Biogeosciences Discussion started: 4 December 2017 c Author(s) 2017. CC BY 4.0 License. temperature treatments were acclimated at the same ambient carbonate system values as that 266 from station L4 on the day of sampling. Following equilibration, the mean pCO2 values within 267 the control and high temperature treatments were 394.9 (± 4.3 sd) and 393.2 (± 4.8 sd) µatm 268 respectively, while in the high pCO2 and combination treatments mean pCO2 values were 822.6 269 (± 9.4) and 836.5 (± 15.6 sd) µatm, respectively. Carbonate system values remained stable 270 throughout the experiment (Fig. 2 B-D). 271

Experimental temperature treatments 272
Mean temperatures in the control and high pCO2 treatments were 14.1 (± 0.35 sd) °C and in the 273 high temperature and combination treatments the mean temperatures were 18.6 (± 0.42 sd) °C. 274 There was a mean temperature difference between the ambient and high temperature 275 treatments of 4.46 (± 0.42 sd) °C (Supporting information, Fig. S1 A & B). 276 277

Chlorophyll a 278
Mean Chl a in the experimental seawater at T0 was 1.64 (± 0.02 sd) mg m -3 (Fig. 3 A). This 279 decreased in all treatments between T0 to T7, to ~0.1 (± 0.09, 0.035 and 0.035 sd) mg m -3 in the 280 control, high pCO2 and combination treatments, while in the high temperature treatment at T7 281 Chl a was 0.46 mg m -3 (± 0.29 sd). From T7 to T12 there was an increase in Chl a in all 282 treatments which was highest in the combination (4.99 mg m -3 ± 0.69 sd) and high pCO2 283 treatments (3.83 mg m -3 ± 0.43 sd) ( Table 1). At T36 Chl a concentration in the combination 284 treatment was significantly higher than all other treatments at 6.87 (± 0.58 sd) mg m -3 (Table  285 2) while the high temperature treatment concentration was significantly higher than the control 286 and high pCO2 treatments at 4.77 (± 0.44 sd) mg m -3 ( Table 2). Mean concentrations for the 287 control and high pCO2 treatments at T36 were not significantly different at 3.30 (± 0.22 sd) and

Phytoplankton biomass 291
The starting biomass in all treatments was 110.2 (± 5.7 sd) mg C m -3 (Fig. 3 B) and the 292 community biomass was dominated by dinoflagellates (~50%) with smaller contributions from 293 nanophytoplankton (~13%), cryptophytes (~11%), diatoms (~9%), coccolithophores (~8%), 294 Synechococcus (~6%) and picophytoplankton (~3%). Total biomass increased significantly in 295 all treatments over time ( Table 1) and at T10, it was significantly higher in the high 296 temperature treatment when the biomass reached 752 (± 106 sd) mg C m -3 . At T36 however, 297 total biomass was significantly higher in the high pCO2 treatment ( Table 1)  (± 182.68 sd) mg C m -3 , which increased more than 20-fold from T0. Total biomass in the high 299 temperature treatment increased more than 15-fold to 1735 (± 169.24 sd) mg C m -3 at T36 and 300 was significantly higher than the combination treatment and ambient control, which were 525 301 (± 28.02 sd) mg C m -3 and 378 (± 33.95 sd) mg C m -3 , respectively ( Table 2). 302 Measured POC followed the same trends as estimated biomass in all treatments between T0 and 303 T36 (Fig. 3 C) and despite some variability between the two measures, POC was within the 304 range of estimates (R 2 = 0.914, Fig. 3 D). At T36, POC was significantly greater in the high pCO2 305 treatment (2086 ± 155.19 sd mg m -3 ) followed by the high temperature treatment (1594 ± 306 162.24 sd mg m -3 ), which were significantly greater than the control and combination treatment 307 (Table 1). PON followed the same trends as POC over the course of the experiment (Fig. 3 E, 308 Table 1): at T36 concentrations were 147 (± 12.99 sd) and 133 (± 15.59 sd) mg m -3 in the high 309 pCO2 and high temperature treatments respectively, while PON was 57.75 (± 13.07 sd) mg m -3 310 in the combination treatment and 47.18 (± 9.32 sd) mg m -3 in the control ( Table 1). POC:PON 311 ratios increased significantly over time in all treatments except for the control. The largest 312 increase of 33 %, from 10.72 to 14.26 mg m -3 (± 1.73 sd) was in the high pCO2 treatment, 313 followed by an increase of 32 % to 9.83 (± 1.82 sd) mg m -3 in the combination treatment, and an 314 increase of 17 % to 12.09 (± 2.14 sd) mg m -3 in the high temperature treatment. In contrast, the 315 POC:PON ratio in the control declined by 20 % from T0 to T36, from 10.33 to 8.26 (± 0.50 sd) 316 mg m -3 ( Fig. 3 F, Table 1). 317

Community composition 318
At T36 diatoms dominated the phytoplankton community biomass in the ambient control with a 319 substantial contribution from nanophytoplankton ( Fig. 4 A), while the high temperature and 320 high pCO2 treatments exhibited near mono-specific dominance of nanophytoplankton (Figs.

& C). The most diverse community was in the combination treatment where dinoflagellates and 322
Synechococcus became more prominent (Fig. 4 D). 323 Between T10 and T24 the community shifted to nanophytoplankton in all experimental 324 treatments. This dominance was maintained through to T36 in the high temperature and high 325 pCO2 treatments whereas in the ambient control and combination treatment, the community 326 shifted away from nanophytoplankton ( Fig. 5 A). At T36 nanophytoplankton biomass was 327 significantly higher in the high pCO2 treatment followed by the high temperature treatment 328 (Table 2) when biomass attained 2216 (± 189.67 sd) mg C m -3 and 1489 (± 170.32 sd) mg C m -3 , 329 respectively. In the combination treatment nanophytoplankton biomass was 238 (± 14.16 sd) 330 mg C m -3 at T36 which was significantly higher compared to the ambient control (162 ± 20.02 sd 331 mg C m -3 ; Table 2). In addition to significant differences in nanophytoplankton biomass 332  Table 2) followed by the high temperature treatment (57 ± 6.87 sd mg C m -3 , 359 Table 2). There was no significant difference in dinoflagellate biomass between the high pCO2 360 treatment and ambient control at T36 when biomass was low. In the combination treatment, 361 dinoflagellate biomass shifted away from the larger G. spirale and was dominated by P. 362 cordatum which contributed 59 ± 12.95 sd mg C m -3 (66 % of biomass in this group). 363 Synechococcus biomass was significantly higher at T36 in the combination treatment (59.9 ± 364 4.30 sd mg C m -3 , Fig. 5 D, Table 2) followed by the high temperature treatment (30 ± 5.98 sd 365 mg C m -3 , Table 2). In both the high pCO2 treatment and ambient control at T36 Synechococcus biomass of picophytoplankton (Fig. 5 E), cryptophytes (Fig. 5 F) and coccolithophores (Fig. 5  368 G) remained low in all treatments throughout the experiment. Though picophytoplankton 369 responded positively to the high pCO2 and combination treatments at T36 (high pCO2: 6.93 ± 370 0.63 sd mg C m -3 ; combination: 11.26 ± 0.79 sd mg C m -3 ; Table 2). 371

Chl a 455
Chl a concentration was significantly higher in the combination treatment at T36 when total 456 biomass was lower, but Chl a was significantly lower in the high pCO2 treatment when biomass

Biomass 463
This study shows that the phytoplankton community response to elevated temperature and 464 pCO2 is highly variable. pCO2 elevated to ~800 µatm induced higher community biomass in

Photosynthetic carbon fixation rates 488
At T36, under elevated pCO2 P B m was > 6 times higher than the ambient control, which has also Photosynthetic rates have been demonstrated to decrease beyond a temperature of 20 °C 501 (Raven and Geider, 1988) which can be modified through photoprotective rather than 502 photosynthetic pigments (Kiefer and Mitchell, 1983). This may explain the difference in P B m 503 between the high pCO2 and high temperature treatments (in addition to differences in 504 nanophytoplankton community composition in relation to Phaeocystis spp. discussed above), as 505 the experimental high temperature treatment in the present study was ~4.5 ° C higher than 506

ambient. 507
There was no significant effect of combined elevated pCO2 and temperature on P B m, which was 508 strongly influenced by taxonomic differences between the experimental treatments. Warming  anoxia through the production of mucus foam which can clog the feeding apparatus of 545 zooplankton and fish (Eilertsen & Raa, 1995). 546 The response of diatoms to elevated pCO2 and temperature has been variable. For example, A 547 study by Taucher et al., (2015) showed that Thalassiosira weissflogii incubated at 1000 µatm 548 pCO2 increased growth by 8 % while for Dactyliosolen fragilissimus, growth increased by 39 %; 549 temperature elevated by + 5°C also had a stimulating effect on T. weissflogii but inhibited the 550 growth rate of D. fragilissimus; and when the treatments were combined growth was enhanced 551 in T. weissflogii but reduced in D. fragilissimus. In partial agreement, the results of the present 552 experiment show that elevated pCO2 increased biomass in diatoms but elevated temperature 553 and the combination of these factors reduced biomass. A distinct size-shift in diatom species 554 was observed in all treatments, from the larger Coscinodiscus spp., Pleurosigma and 555 Thalassiosira subtilis to the smaller Navicula distans. This was most pronounced in the 556 combination treatment where N. distans contributed 89 % of diatom biomass. Navicula spp. 557 previously exhibited a differential response to both elevated temperature and pCO2. At stimulated along a CO2 gradient at a shallow cold-water vent system (Baragi et al., 2015). 563 Synechococcus grown under pCO2 elevated to 750 ppm and temperature elevated by 4 °C 564 resulted in increased growth and a 4-fold increase in P B m (Fu et al., 2007) which is similar to the 565 results of the present study. 566 The combination of elevated temperature and pCO2 significantly increased dinoflagellate 567 biomass which almost doubled, accounting for 17 % of total biomass. This was due to P. 568 In laboratory studies at 1000 ppm CO2, growth rates of the HAB species Karenia brevis increased 572 by 46 %, at 1000 ppm CO2 and + 5 °C temperature it's growth increased by 30 % but was 573 reduced under elevated temperature alone (Errera et al., 2014). A combined increase in pCO2 574 and temperature enhanced both the growth and P B m in the dinoflagellate Heterosigma akashiwo, 575 whereas in contrast to the present findings, only pCO2 alone enhanced these parameters in P. 576 cordatum (Fu et al., 2008). In addition to strong links to toxic algal events, mixotrophy has also been reported in P. 604 cordatum. In a study by Stoecker et al., (1997) up to 50% of P. cordatum sampled from 605 Chesapeake Bay in the summer contained cryptophyte material. The authors concluded that P. 606 cordatum feeding is a mechanism for supplementing carbon nutrition and this may explain why 607 the ratio of nanophytoplankton:dinoflagellates was significantly lower in our combination 608 experimental community compared to the other treatments. ocean-atmosphere equilibrium (Riebesell, 2004).  cordatum biomass values relative to total phytoplankton biomass. Black line is smoothed running average of total phytoplankton biomass, grey area is standard deviation, dotted line is mean P. cordatum biomass and symbols are maximal P. cordatum biomass from weekly observations by year (as per figure legend for B & C.).