Responses of Two Understory Herbs, Maianthemum

1 Jacques et al. Response of two understory herbs to warming.

3 Responses of two understory herbs, 4 Maianthemum canadense (Liliaceae) and Eurybia macrophylla 5 (Asteraceae), to experimental forest warming: early emergence is the 6 key to enhanced reproductive output.1

8 Marie-Hélène Jacques,1 Line Lapointe,1,4 Karen Rice,2 Rebecca A. Montgomery2, Artur 9 Stefanski2, Peter B. Reich2,3

10 1 Département de biologie and Centre d’étude de la forêt, Université Laval, Québec City, 11 QC, Canada, G1V 0A6 12 2 Department of Forest Resources, University of Minnesota, St-Paul, Minnesota, 55108, USA.

13 3 Hawkesbury Institute for the Environment, University of Western Sydney, Penrith, New South 14 Wales, Australia.

15 4 Author for correspondence

17 FOOTNOTE 18 1manuscript received May 7 2014; revision accepted ______.

19 4 [email protected]

21 Acknowledgement

22 The authors thank Stefan Hupperts, Colton Miller, Hannah Tuntland, Camdilla Wirth, 23 Michelle Cummings, Kaleb Remski, Philip Green, Ryan Steenson, Byron Meeks, Jon 24 Shepard, Kaleb Welch, Lindsay Wallis and Elliot Vaughan for field assistance and 25 Dominique Manny, Antoine Daignault, and Timothée Bitsch for laboratory assistance. 26 We would like to thank Roy Rich and Mikhail Titov for providing data, Gilbert Ethier 27 and Steeve Pepin for their scientific advice, and the anonymous reviewers for helpful 28 comments and suggestions during the preparation of this manuscript. This research was 29 funded by the Office of Science (BER), US Department of Energy, Grant No. DE-FG02- 30 07ER64456, the Minnesota Department of Natural Resources, the University of 31 Minnesota College of Food, Agricultural, and Natural Resources Sciences, the UM 32 Wilderness Research Foundation, and NSERC and FRQNT postgraduate fellowship to 33 MHJ.

35 ABSTRACT

36 37 Premise of the study: Understory herbaceous species might be the most sensitive plant 38 form to global warming in deciduous forests, yet they have been little studied in the 39 context of climate change.

40 Methods: A field experiment set up in Minnesota, USA simulated global warming in a 41 forest setting and provided the opportunity to study the responses of Maianthemum 42 canadense and Eurybia macrophylla in their natural environment in interaction with 43 other components of the ecosystem. Effects of +1.7 and +3.4°C treatments on growth, 44 reproduction, phenology, and gas exchange were evaluated along with treatment effects 45 on light, water and nutrient availability, potential drivers of herb responses.

46 Key results: Overall, growth and gas exchanges of these two species were modestly 47 affected by warming. Their emergence occurred up to 16 (E. macrophylla) to 17 days (M. 48 canadense) earlier in the heated plots compared to control plots, supporting early season 49 carbon gain under high light conditions prior to canopy closure. This additional carbon 50 gain in spring likely supported reproduction. Eurybia macrophylla only flowered in the 51 heated plots; and both species had some aspect of reproduction that was highest in 52 +1.7°C treatment. The reduced reproductive effort in the +3.4°C plots was likely due to 53 reduced soil water availability, counteracting positive effects of warming.

54 Conclusions: Global warming might improve fitness of herbaceous species in deciduous 55 forests, mainly by advancing their spring emergence. However, other impacts of global 56 warming such as drier soils in the summer might partly reduce the carbon gain associated 57 with early emergence.

58 59 Key words: Foliar nutrient concentrations, global warming, hardwood forest, phenology, 60 photosynthetic rates, respiratory rates, understory light environment. 61

62 INTRODUCTION 63 64 Warming of the climate system is unequivocal (IPCC, 2007). The herbaceous layer, with 65 its high species richness, plays an important role in maintaining the functional integrity of 66 temperate forest ecosystems through its interactions with the woody seedlings that 67 determine, in part, the future canopy and through its role in carbon, energy and nutrient 68 cycling (Gilliam, 2007). Much less is known about responses to warming of forest herbs 69 compared to the larger trees that co-occur with. Climate change will require the 70 migration, acclimation or adaptation of plant species as temperature is a crucial factor in 71 determining the distribution of species (Woodward and Williams, 1987). Since several 72 forest understory species are very slow colonizers (Verheyen et al., 2003) it is estimated 73 that their rate of migration might not match the rate imposed by changes in climate. For 74 these reasons, it is urgent to assess the effect of temperature on the reproductive output of 75 herbaceous plants (Hovenden et al., 2008). So far, both positive and negative effects have 76 been reported on the reproductive output for the few herbaceous species that have been 77 studied (Pearson and Shah, 1981; Hovenden et al., 2008; De Frenne et al., 2009; De 78 Frenne et al., 2010; De Frenne et al., 2011); early spring flowering species and more 79 northern populations appear to respond more positively to warming than summer 80 flowering and more southern populations (De Frenne et al. 2009).

81 Global warming will also likely modify precipitation regimes although projection 82 models do not agree, even at the regional scale. For Minnesota, the PCM B1 model 83 predicts an increased annual rainfall with limited changes in summer precipitations, 84 whereas GFDL A1F1 scenario predicts a decrease in annual and in summer precipitations 85 (Handler et al. 2014). As both models predict increases in temperature during the summer 86 and more frequent extreme events, it seems reasonable to suggest that understory species 87 will experience more frequent and/or more severe drought in the future. This can have 88 direct adverse effects, and as well might dampen any positive impact of warming on 89 nitrogen supply through increased soil net nitrogen mineralization (Griffin, 1981; Rustad 90 et al., 2001; Schimel et al., 2007; Manzoni et al., 2012).

91 Whether growth responses to warming of understory herbs differs from that of 92 sympatric trees remains unclear. So far, the general response of plant productivity in 93 climate warming experiments has been positive, but as many of the studies used in meta- 94 analyses were conducted in Arctic ecosystems these results are not necessarily indicative 95 of temperate forest herbs (Arft et al., 1999; Rustad et al., 2001; Dormann and Woodin, 96 2002; Walker et al., 2006). A recent meta-analysis used the plant functional group 97 approach and included 127 studies, many at mid-latitudes (Lin et al., 2010). That study 98 found that the increase in biomass of herbaceous species (+5.2%) was smaller than the 99 increase of woody species (+26.7%) although much of this difference was explained by 100 the fact that the effect on herbaceous species from studies of lower latitudes was 101 negative. Milder effects or negative effects of warming on herbaceous plants, where they 102 cohabit with woody species, might be explained through competitive interactions with 103 plants of higher strata. The smaller size of herbaceous plants makes them more 104 vulnerable to shading from taller plants (Castro and Freitas, 2009) and their shallow roots 105 make them more vulnerable to drier soil conditions (Jackson et al., 1996).

106 Climate change is expected to lengthen the growing season in temperate climates. 107 Earlier emergence of spring herbs following an increase in temperature has been reported 108 (Farnsworth et al., 1995; De Frenne et al., 2011) which can be a great advantage since 109 many understory herbs produce most of their photosynthates before the closure of the 110 canopy (Bazzaz and Bliss, 1971; Brown et al., 1985; Lapointe, 2001). However, this 111 advantage can be reduced if trees leaf out earlier, thus reducing the duration of the high 112 light period for herbaceous growth (Routhier and Lapointe, 2002). Herbaceous species 113 that emerge during or after canopy closure exhibit limited changes in phenology (Ishioka 114 et al., 2013). The response of the phenology of the overstory to warming is more complex 115 than that of the herbaceous layer since trees and shrubs are also influenced by 116 photoperiod or require a certain degree of winter chilling in order to become temperature 117 sensitive in spring (Körner and Basler, 2010). Temperature sensitivity of tree leaf 118 emergence also decreases with an increase in latitude (Menzel et al., 2006; Wang et al., 119 2015). This could partly explain why spring emergence in time series show stronger 120 advancement trends in herbaceous than in tree species (Ge et al. 2015). Yet, variability

121 among species response within a same latitude or community precludes predictions of 122 changes in spring phenology at the species level (Parmesan, 2007).

123 Early senescence of herbaceous species subjected to higher temperature regime 124 was also observed (Farnsworth et al., 1995; Jochum et al., 2007), possibly indicating the 125 development of a sink limitation (Lapointe, 2001). Another important component of plant 126 response to warming is physiological acclimation, which is the adjustment of 127 photosynthetic and respiratory rates to improve the plant performance at the new growth 128 temperature (Lambers et al., 1998). Although photosynthesis and respiration are both 129 known to be temperature sensitive, there is growing evidence that both processes 130 acclimate, sometimes even to the extent of homeostasis, in plants exposed to a moderate 131 range of temperature (Campbell et al., 2007, Sendall et al., 2015). We conducted a study 132 on the effect of experimental warming on two understory species widespread in the 133 southern boreal forests of North America: Maianthemum canadense Desfontaines and 134 Eurybia macrophylla (Linnaeus) Cassini. The study system, the B4WarmED experiment, 135 is a free-air warming facility used to study the effect of a warming climate on the 136 regeneration of woody species (Reich et al., 2015; Rich et al., 2015). As the warming 137 system is installed in a forest, it allowed us to study the effect of warming on the two 138 species of interest, but also its interaction with the soil system and other herbaceous and 139 woody species. We tested the following series of hypotheses (Fig. 1).

140 Direct and indirect effects of warming on the ecosystem components will likely 141 translate into changes in growth and reproduction of the studied species. Some of these 142 effects are expected to improve growth and reproductive output of the herbaceous 143 species, e.g. increased nutrient availability and earlier emergence, whereas factors such as 144 lower water availability or lower light due to shading by taller plants are expected to 145 negatively affect the two studied species. Furthermore, some factors might have no net 146 effect on carbon gain, for instance, acclimation of the gas exchange rates. Specifically, 147 the objectives of the present study were to quantify the impact of three years of warming 148 on M. canadense and E. macrophylla growth and reproduction. Furthermore, we tested if 149 light availability in the understory both in the spring and in the summer, plant phenology, 150 leaf photosynthetic and respiratory rates and/or plant nutrient concentration were

151 modulated by the warming treatments. This study quantifies the growth response of two 152 herbaceous species to warming and helps to unravel some of the factors and mechanisms 153 that contribute to this response.

154

155

156 MATERIAL AND METHODS

157 Biology of the studied species─ Maianthemum canadense (False Lily of the Valley) is a 158 rhizomatous clonal plant 5 to 20 cm high. The species is found in moist woods and 159 thickets where it sometimes forms continuous mats (Gleason and Cronquist, 1963). 160 Maianthemum canadense is an early bloomer, and thus emerges and flowers early in the 161 growth season. The fruits mature slowly throughout the summer to become, in late 162 summer, bright red berries sometimes borne by a plant with totally senesced leaves 163 (Jacques, MH. pers. obs). The species is common in deciduous and boreal forests from 164 northern British Columbia and Alberta to southeastern Montana and Wyoming (Flora of 165 North America, 2015). Its range continues southward along the Blue Ridges to 166 Tennessee.

167 Eurybia macrophylla (Large leaf aster) is a rhizomatous clonal plant larger than 168 M. canadense. It often forms dense mats excluding other species (Schulz and Adams, 169 1995). It is found in mixed temperate forests and it tolerates a wide range of light 170 regimes, from shady undergrowth to wide sunny gaps (Schulz and Adams, 1995). It 171 flowers from late summer to late autumn (Marie-Victorin, 1935) and produces winged 172 achenes. The species is common in northeastern deciduous or mixed forests. Its range 173 extends from the southern boreal forest of Eastern Canada to northern Georgia and along 174 the Blue Ridges (Flora of North America, 2015). It becomes rare at the western edge of 175 its range (Manitoba, Illinois, Iowa, and Missouri). 176

177 Study site─The study was conducted within the B4WarmED experiment (Boreal Forest 178 Warming at an Ecotone in Danger). This experiment was established to evaluate, through

179 experimental warming of forest plots, the extent to which, and the mechanisms whereby, 180 projected global warming could alter the composition of tree species of the southern 181 boreal forest (Reich et al. 2015; Rich et al. 2015).

182 The B4WarmED experiment has two locations. The site located at the Cloquet 183 Forestry Center (CFC) is at a slightly lower latitude (Cloquet, MN, 46 ° 31' N, 92 ° 30' 184 W, 386 m asl, 4.5°C MAT, 807 mm MAP) than the site located at the Hubachek 185 Wilderness Research Center (HWRC) (Winton, MN, 47 ° 55' N, 92 ° 30' W, 453 m asl, 186 3.0°C MAT, 722 mm MAP). Although mean annual and seasonal temperatures were 187 historically higher at the southern site, the growing season temperatures during the 188 experiment have been similar at the two sites (Rich et al. 2015; Table 1). Nevertheless, 189 growing season was longer (226 vs. 211 days) at the more southern site.

190 Within each site, half of the plots are located in a mixed forest stand about 40-60 191 years old, mainly composed of aspen, birch and fir trees, and the other half are located in 192 a section of the forest that has recently been clearcut. Clearcut areas are invaded by 193 herbaceous species commonly associated with fields or disturbed areas. For this reason, 194 the present study was restricted to plots located in the intact forest stands, to focus on the 195 impact of global warming on the natural understory layer.

196

197 The warming technique─The B4WarmED experiment combines two techniques: 198 infrared lamps and buried heating cables (Rich et al., 2015). This combination overcomes 199 some of the main disadvantages of these techniques taken individually. It prevents the 200 soil from being not properly warmed by the lamps by compensating with heating cables 201 and avoids the decoupling of belowground and aboveground parts by heating them 202 simultaneously. One can thus achieve a more realistic simulation than if only the 203 belowground or the aboveground was heated (Rich et al., 2015).

204 Infrared lamps of 1000 W are located around the perimeter of each plot at an 205 angle of 45°. Dummy lamps are used for the control plots. The heating cables, buried to a 206 depth of 10 cm, are spread 20 cm apart. The plots are connected to a monitoring system 207 that maintains a constant temperature differential relative to control plots by turning on

208 and off lamps and cables as needed (Table 1, see Rich et al 2015 for more information). 209 The heating system is in operation about eight months a year; it is turned on when the 210 temperature is ≥ 1°C for 5 consecutive days in spring and is turned off when the 211 temperature is ≤ 1°C for 5 consecutive days in autumn.

212

213 Experimental design─The experimental design is a generalized split-plot with the site as 214 the main plot and the treatment as the subplot. There are two warming treatments, +1.7 215 °C and +3.4 °C, the ambient temperature control plots, plus one treatment at ambient 216 temperature with buried but inactive cables to control for the effect of soil disturbance. 217 Data collected in the plots of both treatments at ambient temperature show no significant 218 differences indicating that the disturbance of the soil had very little impact on the 219 processes under study (unpublished data, B4WarmED). Therefore, this paper will only 220 present results for three treatments: control (either control plots depending on the 221 presence of the studied species within each plot), +1.7 °C and +3.4 °C. Mean 222 aboveground and soil temperature along with soil moisture at both sites for 2011 are 223 presented in Table 1. For photosynthesis, respiration and plant nutrient concentration, 224 data were only collected in the control and +3.4°C plots, due to the time required to 225 complete photosynthesis and respiration measurements. Each treatment is repeated twice 226 within a block, which is why the design is said to be «generalized», and there are three 227 blocks per site for a total of 24 or 36 plots depending on the data type (2 sites × 2 or 3 228 treatments × 3 blocks × 2 within-block repetitions).

229 The plots are circular with an area of 7 m2. In 2008, 11 bare-root seedlings of 11 230 woody species (121 plants per plot) were planted and 60 seeds per species per plot were 231 seeded in the first year and 40 in the second. The warming treatment was turned on in 232 spring 2009; the growing seasons 2011 and 2012, during which data for this project were 233 collected, were the third and fourth year of the warming treatment. The two studied 234 species were naturally present in the plots and were chosen because they were present at 235 both study sites and in the vast majority of plots. The mature trees in the stands are not 236 experimentally heated except for perhaps a small fraction of roots extending into warmed 237 plots.

238 The two herbaceous species studied are clonal, which means that in theory the 239 area occupied by clones may exceed the surface of the plot. In such a case, some ramets 240 of a clone may be subjected to the treatment while others are not because they grow 241 outside the plot. However, all rhizomes were cut when the cables were installed, thus 242 isolating ramets inside the plot from those outside. Three years later, a number of ramets 243 have emerged nearby the plots, but the slow growth of these two species (Meier et al., 244 1995) allows us to assume that this number is limited compared to the number of ramets 245 within the plots. In addition, the warming effect extends to a 30-cm strip around the plots 246 (unpublished data, B4WarmED). Thus, the warming treatment may be mitigated to some 247 extent by the presence of ramets outside the plots and the exchanges that may occur 248 between these and the ramets within the plot, but we expect this impact to be limited.

249

250 Light and water availability─Each site is equipped with photometers that record data at 251 regular intervals. To assess the degree of competition for light, measurements of 252 photosynthetically active radiation were taken directly above the leaves of E. 253 macrophylla and M. canadense with a portable photometer (LI-189 with LI-190 sensor, 254 Li-Cor Biosciences, Lincoln, Nebraska) in 2011. Measurements were taken on five 255 ramets per species per plot under diffuse light conditions, that is to say, a cloudy day. The 256 light readings were taken twice a week before and during canopy closure, and about once 257 every two weeks afterwards. The measurements taken at the plant level were then 258 compared to the value recorded by the photometer of the open habitat to calculate the 259 percentage of available light that reached the plants. A mathematical model, the inverted 260 Gompertz curve (Seber and Wild, 1989), was then applied on the data to establish a 261 relationship between the percentage of available light and date throughout the season. It 262 was then possible to calculate the proportion of available light at the plot level at any time 263 during the season. Soil water content was measured continuously in each plot using 30 264 cm long TDR probes (MiniTrase 6050X3, Soil Moisture Equipment Corporation, Goleta, 265 California) placed at 45° in the soil at a depth of 21 cm.

266

267 Total leaf area, reproductive output and phenology─Various sizes of leaves of M. 268 canadense and E. macrophylla were collected outside the plots. These leaves were 269 immediately scanned and their length, width and area were estimated with the ImageJ 270 software (free ware, developed by the National Institute of Health, USA). We then 271 determined that the best equation to estimate the area of these leaves is a second order 272 polynomial regression (Sigma Plot for Windows version 11.0, Systems Software, Inc.). 273 For M. canadense, this relationship has been established from 69 leaves and corresponds 274 to the following equation: Area = -1.002 + 1.613×width + 0.722×width2 (R2 = 0.982), 275 whereas for E. macrophylla, the equation was obtained from 53 leaves: Area = -2.395 + 276 8.040×width + 0.859×width2 (R2 = 0.987). Total leaf area per plant was estimated from 277 measurements of individual leaf width and used to compare the size of plants among 278 treatments. These measures were taken once leaf unfolding was completed in 2011, on 20 279 ramets per species per plot, or for all individuals in the plot if there were fewer than 20. 280 The percentage of M. canadense ramets that had flowers was evaluated on one quarter of 281 the plot, whereas the entire area of the plot was used for E. macrophylla. Flowers, or 282 inflorescences in the case of aster, were counted on each flowering individual, within 283 both species. The fruits were harvested, counted, dried and weighed. Given the sequential 284 flowering and the trade-off between protecting the seeds against the wind and allowing 285 open access to pollinating insects, not all of the seeds of E. macrophylla could be 286 collected. Plant cover per species has been estimated once a year since 2008, on two 287 subplots of 1 m2 per plot. Annual changes in plant cover within each plot (2011/2009), 288 for the two species under study was used to estimate the impact of global warming on 289 clonal growth after taking into account warming impact of individual plant growth. The 290 phenological stage was evaluated every week during the growing season in 2011 and 291 2012. Stages used for analysis are leaf unfolding (date at which the first leaves started to 292 unfold) and leaf withering (date at which the leaves started to turn yellow) to assess the 293 duration of the photosynthetically active season.

294

295 Photosynthesis and respiration─Photosynthetic and respiratory rates were measured 296 using the portable gas exchange system LI-6400 (Li-Cor Biosciences, Lincoln, Nebraska) 297 during the 2011 growing season. Photosynthetic rates under saturating conditions (Asat) 298 and night respiratory rates (Rn) were recorded six times during the season, once plants 299 had reached their maximum size, i.e. from mid-June to late August. A light saturation 300 curve allowed us to determine that an irradiance of 800 µmol m-2 s-1 of photons was

301 sufficient to saturate photosynthesis. All measurements were performed at a CO2 302 concentration of 400 µmol mol-1. Asat and Rn were measured on one different individual 303 per plot each time, within control and +3.4°C treatment plots. During measurements, the 304 LI-6400 console was located outside the plots, thus the air pumped by the instrument was 305 at the control temperature, that is to say the ambient temperature that was prevailing at 306 that moment. Thus, leaf temperatures during gas exchange measurements fluctuated 307 according to ambient temperature (ranged from 15.5 to 31.8°C, Appendix S1; see 308 Supplementary Data with the online version of this article) but were not different 309 between control and heated plots. Thus the rates should be considered as having been 310 measured at a standard temperature, rather than at contrasting treatment temperatures. 311 The measured leaves were then collected, scanned, dried and weighed to calculate their 312 specific leaf area (cm2 g-1). There were no M. canadense leaves collected at the HWRC 313 site because they were too scarce.

314

315 Leaf analysis─Leaves collected during the 1st, 4th and 6th Asat measuring dates were 316 sent to the Laboratoire Daishowa, Pavillon de l’Envirotron, Université Laval to assay for 317 nutrient concentrations. These leaves were dried at 60°C for a minimum of 72 hours. Leaf

318 material was then digested in H2SO4-H2O2-H2SeO3. N and P concentrations were 319 determined by colorimetry (Nkonge and Ballance, 1982) whereas K, Ca and Mg 320 concentrations were determined by atomic absorption spectroscopy. At mid-July, two 321 leaves per plot per species were collected for an estimate of chlorophyll concentration. 322 Chlorophyll was extracted in acetone and quantify by spectrophotometry with a Spectro 323 Cary 300 Bio (Agilent, Mississauga, Canada) and calculations were made according to 324 (Porra et al., 1989).

325

326 Statistical Analyses─Analysis of variance with mixed effects on a generalized split plot 327 design was performed for each variable recorded once during the growth season: total 328 leaf area, data on reproductive effort, chlorophyll concentrations, percent cover, and light 329 (the parameters of the Gompertz curves were compared between treatments: they are 330 unique values obtained from repeated measurements). The fixed effect was the site in the 331 main plot and the treatment in the subplot. The random effects were block nested in site 332 and plot nested in (treatment × block(site)), in accordance with the experimental design. 333 When the Anova indicated that the treatment was significant, a protected LSD test was 334 performed to determine which treatments differed. When the interaction treatment × site 335 was significant, contrasts within each site were performed.

336 For measurements that were taken several times during the summer, such as Asat, 337 Rn and leaf data (SLA and nutrients), or in two years as for the phenology and climate 338 data (stages or parameters were analyzed separately), we performed a repeated split-plot 339 Anova with mixed effects in which time was included as a categorical fixed effect. The 340 Akaike information criterion was used to determine the structure of dependence that best 341 fit the data. When the Anova indicated that the treatment was significant, a protected 342 LSD test was performed to determine which treatments differed. When the interaction 343 treatment × time (or year) was significant, contrasts within each time (or year) were 344 performed. Data were transformed when necessary. All analyses were performed using 345 the mixed procedure in SAS version 9.2 TS Level 1M0 (SAS institute Inc. Cary, North 346 Carolina). The only exception is the percent flowering which was analysed with the 347 glimmix procedure in order to take into account the binary nature of the response 348 variable. To fit a curve to the data for light availability, the nlin procedure was used. 349 When data were collected at a single site only (e.g. M. canadense flowering), a 350 generalized random block design was used.

351 352

353 RESULTS

354 Light and water availability─ The light transmitted to the understory was maximal 355 early in the season, before leaf emergence of trees and shrubs. During this period, the 356 CFC site received about twice the light recorded at the HWRC site (Fig. 2). In the 357 summer, when the total canopy had reached its seasonal maximum, E. macrophylla 358 received less light under the two heated treatments than in the control plots, at both sites 359 (P = 0.03; Appendix, S2; see Supplementary Data with the online version of this article). 360 The trend was the same for M. canadense (P = 0.06). Both air and soil temperatures 361 differed among treatments as expected (Table 1 and 2). Mean air temperature during the 362 2011 growing season was 0.1°C lower and 0.4°C higher than the mean of the previous 10 363 years at CFC and HWRC, whereas in 2012, air temperature in the two locations differed 364 from the mean by 0.2°C and -0.6°C, respectively. Therefore, the +1.7°C treatment 365 induced conditions that were warmer than the typical conditions for the area, and even 366 slightly warmer than the warmest year of the previous 10 years. The water content of the 367 soil was always lower in the heated plots than in the control plots, and the +3.4°C plots 368 were drier than the +1.7°C (Table 1 and 2).

369

370 Plant phenology─For M. canadense, a higher growth temperature advanced the onset of 371 leaf unfolding in both years (Table 2, Fig. 3). Senescence started at the same time for all 372 treatments in 2011, but was earlier in the +3.4°C compared to the +1.7°C and the control 373 plots in 2012. The early emergence in the heated plots allowed a longer growing season 374 for M. canadense in 2011 compared to the control plots. In 2012, the +1.7°C plots 375 experienced a longer growing season than the +3.4°C, but both heated treatments did not 376 differ in terms of length of the growing season from the control plots. In the case of E. 377 macrophylla, the time of leaf unfolding was advanced and the leaf senescence was 378 delayed by a higher growth temperature in both 2011 and 2012, resulting in a longer 379 season for the plants of the heated plots in both years. A much earlier warm-up in spring 380 2012 than 2011 resulted in earlier leaf-out for both species in all treatments.

381

382 Total leaf area─The total leaf area of M. canadense was not affected by growth 383 temperature and individuals were smaller at the HWRC site than at the CFC site (Table 3, 384 Fig. 4). Eurybia macrophylla total leaf area was affected by the growth temperature, but 385 in one site only (CFC), where control plants were smaller than plants in heated plots. 386 There was a non-linear response in which growth was greatest at the +1.7°C treatment in 387 both species, except for M. canadense at HWRC.

388

389 Reproductive output and clonal growth─Maianthemum canadense only flowered at 390 the CFC site in accordance with their larger size than at HWRC (Fig. 4). Warming had no 391 significant effect on the percentage of ramets that flowered, the number of flowers per 392 ramet, nor the number of fruits per ramet (Table 3 and Fig. 5). However, the dry mass per 393 fruit was higher in plants from the +1.7°C plots than in both control and +3.4°C plots.

394 Eurybia macrophylla only flowered in the heated plots, and did so at both sites. 395 There were 3 times more plants in flower in the +1.7°C plots than in the +3.4°C in plots 396 (Table 5 and Fig. 6). The flowering ramets in the +1.7°C plots had almost twice as many 397 inflorescences as those in the +3.4°C plots. These two variables presented a nonlinear 398 response in both species. Although we could not collect all seeds produced, we counted 399 on average 355 ± 23 seeds per flowering ramet in the +1.7°C plots, and 332 ± 31 seeds 400 per flowering ramet in the +3.4°C in plots. The dry mass of E. macrophylla seeds did not 401 vary with growth temperature.

402 Change in percent cover across years can be used as a proxy for clonal growth 403 over time in both species, considering that plant size was not affected by warming, except 404 for E. macrophylla in one site. Cover remained stable in the control plots over the three- 405 year period, whereas in many heated plots, the cover of both species increased 406 dramatically (Appendix S3; see Supplementary Data with the online version of this 407 article). However, to warming were variable across plots, such that there was not a 408 statistical difference among treatments in change in cover from 2009 to 2011 (Table 3; 409 Appendix S3; see Supplementary Data with the online version of this article).

410

411 Specific leaf area─Specific leaf area (SLA) of M. canadense increased during the 412 growing season while it decreased for E. macrophylla (Fig. 7A). Growth temperature did 413 not significantly affect the SLA of either species (Table 4). For M. canadense there is a 414 non-significant trend for plants of the +3.4°C plots to have a higher SLA. For E. 415 macrophylla, the effect of growth temperature on the SLA was almost significant (P = 416 0.065); leaves in the control plots tended to have a higher SLA than the ones in the 417 +3.4°C plots, at the CFC site (Fig. 7B).

418

419 Photosynthesis and respiration─Asat and Rn were expressed on an area basis (µmol -2 -1 -1 -1 420 CO2 m s ) and on a mass basis (nmol CO2 g s ) to take into account the trends 421 observed for the specific leaf area. Asat of M. canadense decreased with time (Table 4). 422 When expressed on an area-based unit, Asat tended to be higher at the ambient 423 temperature than at +3.4°C: the differences were significant for a few dates only, mostly 424 at the CFC site (Figs. 8A and B). The differences between treatments were less obvious 425 when Asat was expressed on a mass basis (only one date for which control plants had 426 higher Asat than the +3.4°C; data available only for the CFC site, Fig. 8E). Leaves from 427 control plots tended to have a lower SLA (Fig. 7A), which could explain the larger 428 differences between treatments observed for area-based than for mass-based Asat.

429 For E. macrophylla, Asat per leaf area were different among treatments at the

430 CFC site (Asat,+3.4°C > Asat,control), but not at HWRC (Table 4; Figs. 8C and D). The 431 difference between treatments disappeared when Asat was expressed on a per mass basis 432 (Figs. 8F and G). The difference caused by the change in units might be due to the 433 specific leaf area that tended to be higher in plants growing at the ambient temperature at 434 the CFC site. Asat did not differ between treatments either on an area-based or on a mass- 435 based unit for the plants of the HWRC sites, in accordance with the very similar SLA 436 reported for the two treatments at that site (Fig. 7C).

437 For both species, Rn was not strongly affected by the warming treatment (Table 4; 438 Figs. 8H to K). Rn tended to be higher in control than in +3.4°C plots but differences 439 were significant for only a few dates in both species. Rn expressed on per leaf area and

440 on a per mass unit exhibited very similar changes over time and between treatments 441 (Figs. 8L to N).

442

443 Leaf nutrients and chlorophyll─ Time affected most of the leaf nutrient concentrations 444 in both species, whereas growth temperature had limited effect (Table 5). The nutrient 445 data were analyzed both on a mass and on a leaf area based unit. In E. macrophylla, some 446 nutrient concentrations (mg g-1) increased with time (Ca and Mg) while others decreased 447 (N, P and K). Ca and Mg concentrations were higher in leaves of plants grown in the 448 heated plots early in the season, but this difference was no longer present at the second 449 and third harvests (Appendix S4; see Supplementary Data with the online version of this 450 article). In M. canadense, K, Ca and Mg concentrations increased with time on a mass 451 basis, whereas the pattern was not clear for N and P (Appendix S5; see Supplementary 452 Data with the online version of this article). P concentration on an area basis was higher 453 in plants grown in the heated plots at first harvest, whereas K concentrations on a mass 454 basis was higher in plants of the control plots but only at second harvest (Table 5; 455 Appendix S5; see Supplementary Data with the online version of this article). There was 456 no effect of the treatments on the total chlorophyll concentration (chl a +b) or on the ratio 457 of chlorophyll a to b (Table 3) in both species.

458 459

460 DISCUSSION

461 462 In general, experimental warming influenced some but not all measures of phenology, 463 growth, metabolism, and chemistry for the two forest herbaceous species studied, M. 464 canadense and E. macrophylla, yet, after three growing seasons, these translated into an 465 increased reproductive output. The main impact of the warming treatments appears to be 466 the earlier emergence of the understory herbs which most likely increased their C gain 467 during the high light period in spring before canopy closure. Many variables responded to 468 warming in a non-linear fashion, for instance leaf area and some of the reproductive

469 output variables. Positive warming effects might thus be limited to a narrow window of 470 temperature increases. These non-linear trends also remind us that predicting the response 471 of plants is complex: multiple factors are influenced by warming and are thus involved in 472 the response of plants in natural habitats (Fig. 1); this limits our capacity to predict how 473 specific species will respond to warming without empirical data.

474 M. canadense and E. macrophylla emerged earlier in the heated plots and it is 475 likely that they fixed a large amount of their annual C during the open canopy period. 476 Spring photosynthesis is a key element for the productivity of many understory species of 477 hardwood forests (Bazzaz and Bliss, 1971; Brown et al., 1985; Lapointe, 2001; Neufelf 478 and Young, 2014), although many understory species do emerge during or after canopy 479 closure. Photosynthetic rates of spring emerging species are at the highest while the light 480 is at its maximum availability and rates drop after canopy closure. For some of these 481 species, photosynthetic rates drop after only a few weeks, sometimes even days (Larcher, 482 2003). This strategy is called the «phenological escape» because these species advance 483 their emergence to avoid the shade of taller species that develop later. However, early 484 emergence also increases the risk of spring frost which could strongly reduce annual C 485 gain.

486 After tree leaf emergence, plants in the heated plots received less light due to the 487 improved growth of the transplanted tree seedlings in these plots than in the controls. 488 However, reduced light during the summer did not affect SLA or chlorophyll a/b ratio, 489 two variables known to be sensitive to light availability. These results indicate that the 490 lowest light levels that prevail in the heated plots in summer are not dramatic enough to 491 induce strong morphological or physiological acclimation, compared to the change in 492 light from spring to summer that induce a reduction in Asat in many species (Taylor and 493 Pearcy, 1976). It is also possible that these plants were already as much acclimated to 494 shade as their plasticity allowed them. In this experimental setup, the overstory canopy is 495 not exposed to the treatment. However, the saplings of woody species planted in the plots 496 were heated. These saplings also advanced their phenology in response to warming 497 (Schwartzberg et al 2014; and data not published, B4WarmED) as reported in other 498 studies (Fu et al. 2014; Kaye & Wagner, 2014). While herbaceous species rely more

499 strictly on the temperature and snow melt as a signal for emerging (Farnsworth et al., 500 1995; Iversen et al., 2009; De Frenne et al., 2011; Cornelius et al., 2013), the response of 501 trees to temperature is more complex and varies among species, because of different 502 sensitivities to temperature, photoperiod and winter chilling (Murray et al. 1989; Borchert 503 et al., 2005; Vitasse et al. 2009; Körner and Basler, 2010). Although most warming 504 studies have used seedlings or saplings rather than mature trees for practical reasons, 505 ontological differences cause young trees to leaf out earlier than their mature 506 conspecifics, even after taking into account vertical differences in temperature within the 507 forest (Vitasse, 2013). It is therefore important to quantify the warming impact on the 508 different functional groups within a forest, including the overstory trees, in order to 509 determine which understory plants if any will be able to take advantage of an earlier 510 leafing out under a warmer climate. It is expected that the effect of warming on the 511 phenological escape strategy used by many understory species will most likely vary with 512 composition and successional stage of the forest stand. Furthermore, results from 513 experimental warming studies will need to be confirmed by observational data recorded 514 along time series as it appears that the former may underpredict the observed advances in 515 spring phenology (Wolkovich et al., 2012).

516 Not only emergence, but also senescence of the two herbaceous species was 517 affected by the warming treatment. Maianthemum canadense senesced earlier in warmed 518 plots in 2012, similarly to what has been reported for another spring flowering species, 519 Smilacina japonica (Ishioka et al. 2013) or for heated plots containing mostly M. 520 canadense and Uvularia sessilifolia (Farnsworth et al.1995). In contrast, E. macrophylla 521 plants lasted longer in the heated plots in both years, whereas other studies reported either 522 no change in three forest summer herbs (Ishioka et al., 2013) or even an earlier 523 senescence in Panax quinquefolius (Jochum et al., 2007). There are numerous factors 524 controlling the initiation of senescence, among them environmental stresses, photoperiod, 525 temperature (Munne-Bosch and Alegre, 2004), or source-sink relationships (Lapointe, 526 2001). The different timing in the senescence of the two species can be partially 527 explained by their reproductive strategies: early bloomer for M. canadense and late 528 summer bloomer for E. macrophylla. The fruits of M. canadense are ripe in September 529 and senescence seems to be induced once seeds complete their development. There is no

530 other major sink within the plant at seed ripening as carbohydrate storage in rhizomes 531 have already taken place (Lapointe, unpubl. data). Eurybia macrophylla seeds mature 532 until the very end of the season; some seeds were actually not ripe when the plant finally 533 senesced. This species appears to lengthen its growth season as much as possible, and 534 night frost most likely induces senescence. Delayed senescence under a warmer climate 535 would allow the plant to mature more seeds. The effect of warming on shoot senescence 536 might therefore depend on the reproductive strategy of the species which may partly 537 explain the diversity of responses reported in the literature (Menzel et al., 2006).

538 The effect of the heating treatment on Asat and Rn was not pronounced in either 539 species, as reported for other understory herbaceous species in soil-warming alone 540 experiments (Ishioka et al., 2013). The results support the hypothesis that there was a 541 shift of the thermal optimum of Asat and Rn, with no change in the elevation of the 542 curve, as the rates in the heated plots were lower when measured at the common ambient 543 temperature. For the planted deciduous tree species in B4WarmED, acclimation of Asat 544 and Rn was a general response (Sendall et al., 2015; also unpublished data). Although it 545 was not possible to establish the temperature response curves of Asat in the field, the

546 literature indicates that most C3 plants have a broad optimum (Sage and Kubien, 2007), 547 located between 15° and 30°C (Larcher, 2003), as was the case for the deciduous trees in 548 B4WarmED (Sendall et al., 2015). Moreover, Asat of the juvenile trees were as well 549 matched to their thermal regimes in the warmed as in the ambient plots in 2009 to 2011 550 (Sendall et al., 2015). If true for the two herb species, this would also contribute to 551 minimal differences in Asat across treatments.

552 Additionally, other aspects of plant function can alter realized carbon exchange 553 rates, beyond the shape of the response curve. For example, if changes in leaf N occur 554 under warming, it could translate into changes in elevation of Asat and Rn anywhere 555 along a temperature response curve (Lambers et al., 1998). However, warming studies in 556 general indicate limited impacts on leaf N, about 3.3% increase according to a meta- 557 analysis (Bai et al. 2013). In the present study the effect of warming on the nutrient 558 content of the leaves was also very small. Therefore, the main impact of the warming

559 treatment appears to be on the duration of the photosynthetic period rather than on 560 instantaneous gas exchange rates.

561 The modest response of Asat to warming of E. macrophylla and M. canadense is 562 also consistent with the location of the experimental site relative to their north-south 563 range distributions. For the 11 B4WarmED tree species, those with northern ranges close 564 to the study site (e.g. maples, oaks) had increased Asat in response to warming, whereas 565 those whose southern range was close to the study sites (fir, spruce) had decreased Asat 566 (Reich et al., 2015). The ranges of both E. macrophylla and M. canadense extend far to 567 the north and the south of northern Minnesota; thus perhaps both the ambient and 568 warmed thermal regimes fall well within the general range of variability to which the 569 species are adapted.

570 The increased reproductive output reported in both species supports the 571 assessment that plants in heated plots fixed more C during the season, most likely due to 572 their extended growth season. Indeed, it is remarkable that E. macrophylla only flowered 573 and produced seeds in the heated plots despite no significant change in Asat or total leaf 574 area. Small but cumulative differences in total C gain over the season could translate, 575 three years later, into a significant investment into sexual reproduction in heated plots. In 576 another spring flowering understory species, Anemone nemorosa, an early emergence 577 was correlated with an increased percentage of flowering and increased seed mass (De 578 Frenne et al., 2011). However, in summer flowering species, warming did not improve 579 total seed production (De Frenne et al. 2009), contrarily to what we reported here for E. 580 macrophylla. The response of reproductive output to warming appears to be partly 581 dependent upon their phenology (De Frenne et al., 2009; De Frenne et al., 2011) but the 582 impact of warming on plant phenology needs to be characterized for many species before 583 we can generalize the response of understory species to climate warming.

584 The non-linear response of the reproductive output of both species, which was 585 also observed for the total leaf area and the percent cover (although both not significant), 586 might be related to potential water stress at the higher temperature regime. Soil water 587 content decreased with the increase in temperature, and this effect was present throughout

588 the season (Rich et al 2015). However, the summer 2011 was not dry enough to reduce 589 stomatal conductance in the +3.4°C treatment compared to the control plots (data not 590 shown). The reduced growth recorded in 2011at the highest warming treatment compared 591 to the milder warming treatment, could be the result of cumulative effects of water stress 592 from the two previous seasons (Inghe and Carl, 1985; Inghe and Tamm, 1988), which 593 were both relatively dry in this area (Reich et al. 2015).

594 We noted the presence of a pathogen on E. macrophylla leaves at the CFC site. 595 This pathogen is the pine needle rust, Coleosporium asterum Dietel) Syd. & P. Syd, of 596 which E. macrophylla is an alternate host. Only E. macrophylla located in control plots 597 were attacked by the pine needle rust, which could explain why these plants were smaller 598 and tended to have a higher SLA than plants in heated plots. The urediospores of C. 599 asterum were certainly on E. macrophylla leaves in each plot, but conditions suitable for 600 germination were only found in the control plots. The optimal temperature for 601 urediospores germination is 20°C while only 1% of the urediospores germinate at 30°C 602 and none at 35°C (Fergus, 1959; Nicholls et al., 1967). The heated plots were most likely 603 too warm for germination, but perhaps also too dry. We observed on a regular basis 604 throughout the season that the dew had already evaporated in the heated plots as we 605 arrived at the study site in the morning while it was still present on the plants of control 606 plots. In a warmer future, changes in the phenology, severity and distribution of biotic 607 agents are to be expected, with potentially complex impacts on growth and reproduction 608 (e.g. Liu et al 2011).

609 610

611 Conclusion─ Warming induced an earlier emergence in both M. canadense and E. 612 macrophylla, two early emerging species, whereas the timing of senescence responded 613 inconsistently, being advanced, unaffected or retarded in heated plots. The growing 614 season was nevertheless longer and although it did not translate into consistently larger 615 plants, it did allow some individual plants to grow larger and produce more flowers and 616 seeds. Mid-summer conditions in warmed plots (lower light and soil water availability)

617 might have counteracted the positive effects of the earlier emergence and thus explain the 618 limited impact of the lengthened growing season on M. canadense and E. macrophylla 619 growth, as well as only modest and inconsistent impacts on leaf gas exchange rates. 620 Furthermore, leaf nutrient concentration was not generally increased by the warming 621 treatment, which could partly explain the lack of positive response of photosynthesis and 622 respiration to the warming treatment. The earlier emergence and heightened reproduction 623 would be advantageous under a warmer climate, especially if shrub and tree strata do not 624 shift their emergence as much as the herbs. We thus need to quantify the warming impact 625 on the phenology of the different forest plant groups but also of specific species to predict 626 how different forest ecosystems will respond to global warming.

627

LITERATURE CITED

Arft, A. M. et al. 1999. Responses of tundra plants to experimental warming: Meta- analysis of the international tundra experiment. Ecological Monographs 69: 491-511.

Bazzaz, F. A. and L. C. Bliss. 1971. Net primary production of herbs in a central Illinois deciduous forest. Bulletin of the Torrey Botanical Club 98: 90-94.

Bai, E., L. Shanlong, W. Xu, W. Li, W. Dai, and P. Jiang. 2013. A meta-analysis of experimental warming effects on terrestrial nitrogen pools and dynamics. New Phytologist 199: 441-451.

Bellemare, J., and Moeller, D.A. 2014. Climate change and forest herbs of temperate deciduous forests. In: F.S. Gilliam [ed], The herbaceous layer in forests of Eastern North America, 2nd ed, 460-495. Oxford University Press, Oxford, UK.

Borchert, R., K. Robertson, M. D. Schwartz, and G. Williams-Linera. 2005. Phenology of temperate trees in tropical climates. International Journal of Biometeorology 50: 57-65.

Brown, R. L., J. W. Ashmun, and L. F. Pitelka. 1985. Within-species and between-species variation in vegetative phenology in two forest herbs. Ecology 66: 251-258.

Campbell, C., L. Atkinson, J. Zaragoza-Castells, M. Lundmark, O. Atkin, and V. Hurry. 2007. Acclimation of photosynthesis and respiration is asynchronous in response to changes in temperature regardless of plant functional group. New Phytologist 176: 375- 389.

Castro, H. and H. Freitas. 2009. Above-ground biomass and productivity in the Montado: From herbaceous to shrub dominated communities. Journal of Arid Environments 73: 506- 511.

Cornelius, C., N. Estrella, H. Franz, and A. Menzel. 2013. Linking altitudinal gradients and temperature responses of plant phenology in the Bavarian Alps. Plant Biology 15 (Suppl. 1): 57-69.

De Frenne, P. et al. 2011. Temperature effects on forest herbs assessed by warming and transplant experiments along a latitudinal gradient. Global Change Biology 17: 3240-3253.

De Frenne, P. et al. 2010. Significant effects of temperature on the reproductive output of the forest herb Anemone nemorosa L. Forest Ecology and Management 259: 809-817.

De Frenne, P. et al. 2009. Unravelling the effects of temperature, latitude and local environment on the reproduction of forest herbs. Global Ecology and Biogeography 18: 641-651.

Dormann, C. F. and S. J. Woodin. 2002. Climate change in the Arctic: using plant functional types in a meta-analysis of field experiments. Functional Ecology 16: 4-17.

Farnsworth, E. J., J. NunezFarfan, S. A. Careaga, and F. A. Bazzaz. 1995. Phenology and growth of three temperate forest life forms in response to artificial soil warming. Journal of Ecology 83: 967-977.

Fergus, C. L. 1959. The influence of environment upon germination and longevity of aeciospores and urediospores of Coleosporium solidaginis. Mycologia 51: 44-48.

Flora of North America. 2015. Website http://www.efloras. org / [accessed 08 January 2015].

Fu, Y.S.H., M. Campioli, Y. Vitasse, H.J. De Boeck, J. Van den Berge, H. AbdElgawad, H. Asard, S. Piao, G. Deckmyn and I.A. Janssens. 2014. Variation in leaf flushing date influences autumnal senescence and next year's flushing date in two temperate tree speices. Proceedings of the National Academy of Sciences of the United States of America 111: 7355-7360.

Ge, Q., H. Wang and J. Dai. 2015. Phenological response to climate change in China: a meta‐ analysis. Global Change Biology 21: 265-274.

Gilliam, F. S. 2007. The ecological significance of the herbaceous layer in temperate forest ecosystems. Bioscience 57: 845-858.

Gleason, H. A. and A. R. T. H. Cronquist. 1963. Manual of vascular plants of northeastern United States and adjacent Canada. Van Nostrand, Princeton, USA.

Griffin, D. M. 1981. Water and microbial stress. Advances in Microbial Ecology 5: 91-136.

Handler S, Duveneck MJ, Iverson L et al. (2014) Minnesota forest ecosystem vulnerability assessment and synthesis: A report from the northwoods climate change response framework project. Gen. Tech. Rep. NRS-XX. Newtown Square, PA; U.S., Department of Agriculture, Forest Service, Northern Research Station. http://www.fs.fed.us/nrs/pubs/gtr/gtr_nrs133.pdf [accessed 14 July 2015].

Hovenden, M. J., K. E. Wills, R. E. Chaplin, J. K. V. Schoor, A. L. Williams, Y. Osanai, and P. C. D. Newton. 2008. Warming and elevated CO2 affect the relationship between seed mass, germinability and seedling growth in Austrodanthonia caespitosa, a dominant Australian grass. Global Change Biology 14: 1633-1641.

Inghe, O. and O. T. Carl. 1985. Survival and flowering of perennial herbs. IV. The behaviour of Hepatica nobilis and Sanicula europaea on permanent plots during 1943- 1981. Oikos 45: 400-420.

Inghe, O. and C. O. Tamm. 1988. Survival and flowering of perennial herbs. V. Patterns of flowering. Oikos 51: 203-219.

IPCC. 2007. Contribution of working groups I, II and III to the fourth assessment report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland.

Ishioka, R., O. Muller, T. Hiura and G. Kudo. 2013. Responses of leafing phenology and photosynthesis to soil warming in forest-floor plants. Acta Oecologica 51: 34-41.

Iversen, M., K. A. Brathen, N. G. Yoccoz, and R. A. Ims. 2009. Predictors of plant phenology in a diverse high-latitude alpine landscape: growth forms and topography. Journal of Vegetation Science 20: 903-915.

Jackson, R. B., J. Canadell, J. R. Ehleringer, H. A. Mooney, O. E. Sala, and E. D. Schulze. 1996. A global analysis of root distributions for terrestrial biomes. Oecologia 108: 389- 411.

Jochum, G. M., K. W. Mudge, and R. B. Thomas. 2007. Elevated temperatures increase leaf senescence and root secondary metabolite concentrations in the understory herb Panax quinquefolius (Araliaceae). American Journal of Botany 94: 819-826.

Kaye, M.W. and R.J. Wagner. 2014. Eastern deciduous tree seedlings advance spring phenology in response to experimental warming, but not wetting, treatments. Plant Ecology 215: 543-554.

Körner, C. and D. Basler. 2010. Phenology under global warming. Science 327: 1461- 1462.

Lambers, H., F. S. I. Chapin, and T. L. Pons. 1998. Plant physiological ecology. Springer, New York, USA.

Lapointe, L. 2001. How phenology influences physiology in deciduous forest spring ephemerals. Physiologia Plantarum 113: 151-157.

Larcher, W. 2003. Physiological Plant Ecology. Springer, Berlin, Germany.

Laube, J., T. H. Sparks, N. Estrella, J. Höfler, P. A. Donna, and A. Menzel. 2014. Chilling outweighs photoperiod in preventing precocious spring development. Global Change Biology 20: 170-182.

Lin, D. L., J. Y. Xia, and S. Q. Wan. 2010. Climate warming and biomass accumulation of terrestrial plants: a meta-analysis. New Phytologist 188: 187-198.

Liu, Y., P. B. Reich, G. Li and S. Sun. 2011. Shifting phenology and abundance under experimental warming alters trophic relationships and plant reproductive capacity. Ecology 92:1201–1207.

Manzoni, S., J. P. Schimel, and A. Porporato. 2012. Responses of soil microbial communities to water stress: results from a meta-analysis. Ecology 93: 930-938.

Marie-Victorin, f. F. E. C. 1935. Flore laurentienne. Imprimerie De-la-Salle, Montréal, Canada.

Meier, A. J., S. P. Bratton, and D. C. Duffy. 1995. Possible ecological mechanisms for loss of vernal-herb diversity in logged eastern deciduous forests. Ecological Applications 5: 935-946.

Menzel, A et al. 2006. European phenological response to climate change matches the warming pattern. Global Change Biology 12: 1969–1976.

Munne-Bosch, S. and L. Alegre. 2004. Die and let live: leaf senescence contributes to plant survival under drought stress. Functional Plant Biology 31: 203-216.

Murray, M. B., M. G. R. Cannell, and R. I. Smith. 1989. Date of budburst of 15 tree species in Britain following climatic warming. Journal of Applied Ecology 26: 693-700.

Neufeld, H.S., and Young, R.D. 2014. Ecopysiology of the herbaceous layer in temperate deciduous forests. In: F.S. Gilliam [ed], The herbaceous layer in forests of Eastern North America, 2nd ed, 35-95. Oxford University Press, Oxford, UK.

Nicholls, T. H., R. F. Patton, and E. P. Van Arsdel. 1967. Life cycle and seasonal development of Coleosporium pine needle rust in Wisconsin. Phytopathology 58: 822-829.

Nkonge, C. and G. M. Ballance. 1982. A Sensitive Colorimetric Procedure for Nitrogen Determination in Micro-Kjeldahl Digests. Journal of Agricultural and Food Chemistry 30: 416-420.

Parmesan, C. 2007. Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biology 13: 1860-1872.

Pearson, C. J. and S. G. Shah. 1981. Effects of temperature on seed production, seed quality and growth of Paspalum dilatatum. Journal of Applied Ecology 18: 897-905.

Porra, R. J., W. A. Thompson, and P. E. Kriedemann. 1989. Determination of accurate extinction coefficients and simultaneous equations for assaying chlorophyll a and chlorophyll b extracted with four different solvents - Verification of the concentration of chlorophyll standards by atomic absorption spectroscopy. Biochimica et Biophysica Acta 975: 384-394.

Reich, P. B., K. M. Sendall, K. Rice, R. L. Rich, A. Stefanski, S. E. Hobbie and R. A. Montgomery. 2015. Geographic range predicts photosynthetic and growth response to

warming in co-occurring tree species. Nature Climate Change 5: http://www.nature.com/doifinder/10.1038/nclimate2497

Rich, R.L., A. Stefanski, R. A. Montgomery, S. E. Hobbie, B. A. Kimball and P. B. Reich. 2015. Design and performance of combined infrared canopy and belowground warming in the B4WarmED (Boreal Forest Warming at an Ecotone in Danger) experiment. Global Change Biology doi 10.1111/gcb.12855

Routhier, M.-C. and L. Lapointe. 2002. Impact of tree leaf phenology on growth rates and reproduction in the spring flowering species, Trillium erectum (Liliaceae). American Journal of Botany 89: 499-504.

Rustad, L. E., J. L. Campbell, G. M. Marion, R. J. Norby, M. J. Mitchell, A. E. Hartley, J. H. C. Cornelissen, and J. Gurevitch. 2001. A meta-analysis of the response of soil respiration, net nitrogen mineralization, and aboveground plant growth to experimental ecosystems warming. Oecologia 126: 543-562.

Sage, R. F. and D. S. Kubien. 2007. The temperature response of C3 and C4 photosynthesis. Plant, Cell & Environment 30: 1086-1106.

Schwartzberg, E. G., M. A. Jamieson, K. F. Raffa, P. B. Reich, R. A. Montgomery and R. L. Lindroth. 2014. Simulated climate warming alters phenological synchrony between an outbreak insect herbivore and host trees. Oecologia 175: 1041-1049.

Schimel, J., T. C. Balser, and M. Wallenstein. 2007. Microbial stress-response physiology and its implications for ecosystem function. Ecology 88: 1386-1394.

Schulz, K. E. and M. S. Adams. 1995. Effect of canopy gap light environment on evaporative load and stomatal conductance in the temperate forest understory herb Aster macrophyllus (Asteraceae). American Journal of Botany 82: 630-637.

Seber, G. A. F. and C. J. Wild. 1989. Nonlinear Regression. Wiley, New York, USA.

Sendall, K. M., P. B. Reich, C. Zhao, H. Jihua, X. Wei, A. Stefanski, K. Rice, R. L. Rich and R. A. Montgomery. 2015. Acclimation of photosynthetic temperature optima of temperate and boreal tree species in response to experimental warming. Global Change Biology doi: 10.1111/gcb.12781.

Taylor, R. J. and R. W. Pearcy. 1976. Seasonal Patterns of Co-2 Exchange Characteristics of Understory Plants from A Deciduous Forest. Canadian Journal of Botany 54: 1094- 1103.

Verheyen, K., O. Honnay, G. Motzkin, M. Hermy, and D. R. Foster. 2003. Response of forest plant species to land-use change: a life-history trait-based approach. Journal of Ecology 91: 563-577.

Vitasse, Y., A.J. Porté, A. Kremer, R. Michalet and S. Delzon. 2009. Responses of canopy duration to temperature changes in four temperate tree species: relative contributions of spring and autumn leaf phenology. Global Change Biology 161: 187-198.

Vitasse, Y. 2013. Ontogenic changes rather than difference in temperature cause understorey trees to leaf out earlier. New Phytologist 198: 149-155.

Walker, M. D. et al. 2006. Plant community responses to experimental warming across the tundra biome. Proceedings of the National Academy of Sciences of the United States of America 103: 1342-1346.

Wang, H., J. Dai, J. Zheng and Q. Ge. 2015. Temperature sensitivity of plant phenology in temperate and subtropical regions of China from 1850 to 2009. International Journal of Climatology 35: 913-922.

Woodward, F. I. and B. G. Williams. 1987. Climate and plant distribution at global and local scales. Vegetatio 69: 189-197.

Wolkovich, E. M., B. I. Cook, J. M Allen, T. M Crimmins, J. L.Betancourt, S Travers et al. 2012. Warming experiments underpredict plant phenological responses to climate change. Nature, 485: 494–497.

TABLE 1. Average (± standard error) aboveground temperature (°C), soil temperature (°C) and soil moisture (%) for the different treatments at each site for the growing season 2011, i.e. from 9 April to 21 November at CFC and from 21 April to 18 November at HWRC and for the growing season 2012, i.e. from 21 March to 24 November at CFC and from 27 March to 16 November at HWRC. Results of Anova and of a posteriori contrasts for these data are presented in Table 2.

Average aboveground Average soil Average Site Treatment Year temperature (°C)1 temperature (°C)2 soil moisture (%)3 2011 CFC Ambient 11.0 ± 0.08 11.5 ± 0.06 22.0 ± 0.94 CFC +1.7 °C 12.6 ± 0.04 13.3 ± 0.05 19.2 ± 1.09 CFC +3.4 °C 13.9 ± 0.04 14.9 ± 0.04 18.4 ± 0.71

HWRC Ambient 12.7 ± 0.10 12.6 ± 0.10 24.9 ± 3.28 HWRC +1.7 °C 14.5 ± 0.08 14.4 ± 0.05 21.9 ± 2.16 HWRC +3.4 °C 16.0 ± 0.11 16.1 ± 0.05 20.8 ± 1.91 2012 CFC Ambient 9.8 ± 0.09 10.8 ± 0.11 17.3 ± 0.23 CFC +1.7 °C 11.5 ± 0.04 12.4 ± 0.11 15.8 ± 0.23 CFC +3.4 °C 13.1 ± 0.05 13.7 ± 0.08 16.0 ± 0.15

HWRC Ambient 10.5 ± 0.08 11.0 ± 0.07 24.0 ± 0.58 HWRC +1.7 °C 12.1 ± 0.05 12.9 ± 0.10 21.5 ± 0.63 HWRC +3.4 °C 13.7 ± 0.06 14.3 ± 0.07 20.1 ± 0.51

Notes: The differences between the percentages of soil moisture are not part of the treatment but a result of it.

TABLE 2. Repeated measure analysis of variance with mixed effects testing the warming treatments and site effects on the date of leaf unfolding, leaf withering and length of the growth season of Mainthemum canadense and Eurybia macrophylla as well as on mean air and soil temperature and soil moisture content.

Site × Site × treat Site Treatment Year Site × Year Treat × Year Treatment × Year M. canadense F P F P F P F P F P F P F P

Leaf unfolding1 0.96 0.38 22.4 <.001 0.07 0.93 59.1 <.001 0.01 0.92 3.95 0.03 1.32 0.29

Leaf wilting2 0.17 0.70 7.83 0.002 0.94 0.41 23.2 <.001 0.70 0.41 5.10 0.01 1.32 0.29 Length of season3 0.42 0.55 4.88 0.02 1.59 0.22 0.08 0.79 0.25 0.62 3.46 0.05 7.24 0.005

E. macrophylla

Leaf unfolding4 55.8 0.002 40.7 <.001 1.00 0.38 159 <.001 22.6 <.001 1.44 0.25 2.64 0.09

Leaf wilting5 0.54 0.5 9.85 <.001 2.69 0.09 14.5 <.001 7.39 0.01 1.04 0.37 0.55 0.58 Length of season6 9.28 0.04 31.9 <.001 4.60 0.02 2.84 0.10 28.3 <.001 2.19 0.13 2.75 0.08

Climate conditions Air temperature7 149 <.001 2694 <.001 1.86 0.17 2656 <.001 286 <.001 30.6 <.001 1.05 0.36 Soil temperature8 77.9 <.001 1799 <.001 0.10 0.90 1492 <.001 110 <.001 2.44 0.10 6.76 0.002 Soil moisture content9 13.9 0.02 7.39 <.001 0.41 0.67 9.68 0.003 5.79 0.02 0.26 0.78 0.30 0.75 Notes: Results of contrasts comparing treatments within year, the two sites pooled.

TABLE 2 (cont.)

1 2011: Control a; +1.8°C b; +3.6°C b; 2012: Control a; +1.8°C b; +3.6°C c. 2 2011: Control a; +1.8°C a; +3.6°C a; 2012: Control a; +1.8°C a; +3.6°C b. 3 2011: Control b; +1.8°C a; +3.6°C a; 2012: Control ab; +1.8°C a; +3.6°C b. 4 2011: Control a; +1.8°C b; +3.6°C b; 2012: Control a; +1.8°C b; +3.6°C c. 5 2011: Control b; +1.8°C a; +3.6°C a; 2012: Control b; +1.8°C a; +3.6°C a. 6 2011: Control b; +1.8°C a; +3.6°C a; 2012: Control b; +1.8°C a; +3.6°C a. 7 2011: Control a; +1.8°C b; +3.6°C c; 2012: Control a; +1.8°C b; +3.6°C c. 8 2011: Control a; +1.8°C b; +3.6°C c; 2012: Control a; +1.8°C b; +3.6°C c. 9 2011: Control a; +1.8°C ab; +3.6°C b; 2012: Control a; +1.8°C b; +3.6°C c.

TABLE 3. Analysis of variance with mixed effects testing warming treatments and site effects on the total leaf area, flower and fruit production, percent plant cover and leaf chlorophyll concentration of Mainthemum canadense and Eurybia macrophylla.

Site Treatment Site × treatment F P F P F P M. canadense Total leaf area 29.14 0.01 2.44 0.11 1.39 0.27 Flowering (%) ------1.38 0.29 ------Number of flowers / ramet ------0.07 0.94 ------Number of fruits / ramet ------2.37 0.13 ------Dry mass per fruit (mg) ------6.50 0.01 ------Percent cover 2011/2009 2.02 0.23 2.61 0.09 1.05 0.37 Total chlorophyll (mg g-1) ------0.07 0.94 ------Chlorophyll a/ b ------0.38 0.69 ------

E. macrophylla Total leaf area 13.98 0.02 6.27 0.01 4.89 0.02 Flowering (%) 0.81 0.42 27.59 <0.001 2.7 0.06 Number of flowers / ramet 5.38 0.08 8.06 0.01 0.82 0.37 Number of fruits / ramet ------Dry mass per fruit (mg) 0.19 0.68 0.99 0.37 2.32 0.19 Precent cover 2011/2009 3.99 0.12 1.77 0.19 2.11 0.14 Total chlorophyll (mg g-1) 0.65 0.47 1.18 0.32 2.45 0.11 Chlorophyll a/ b 0.06 0.82 2.45 0.11 0.26 0.77

TABLE 4. Analysis of variance with mixed effects testing the warming treatments, site and time effects on the specific leaf area, the Asat and Rn both on an area-based and on a mass-based units for Maianthemum canadense and Eurybia macrophylla.

Site × Treatment × Site × Treat Site Treatment Time Site × Time Treatment Time ×Time F P F P F P F P F P F P F P M. canadense SLA (cm2 g-1) ------3.11 0.22 ------5.37 <.001 ------0.95 0.45 ------Asat (µmol m-2 s-1) 2.05 0.23 10.20 0.01 1.22 0.29 18.67 <.001 1.81 0.12 2.42 0.04 2.77 0.02 Asat (nmol g-1 s-1) ------6.27 0.03 ------2.72 0.03 ------1.64 0.17 ------Rn (µmol m-2 s-1) 1.37 0.31 3.54 0.08 4.63 0.05 9.53 <.001 15.42 0.001 1.15 0.34 1.88 0.11 Rn (nmol g-1 s-1) ------18.84 0.003 ------9.09 <.001 ------2.54 0.04 ------

E. macrophylla SLA (cm2 g-1) 0.16 0.71 6.39 0.06 3.81 0.12 7.46 <.001 0.69 0.64 1.67 0.15 0.38 0.86 Asat (µmol m-2 s-1) 0.84 0.41 5.51 0.03 7.36 0.01 1.16 0.33 1.08 0.38 0.96 0.45 0.27 0.93 Asat (nmol g-1 s-1) 0.46 0.54 0.01 0.93 0.75 0.40 6.17 <.001 1.43 0.22 1.62 0.16 0.18 0.97 Rn (µmol m-2 s-1) 3.62 0.13 5.14 0.04 0.46 0.51 35.32 <.001 11.37 <.001 0.42 0.84 1.23 0.30 Rn (nmol g-1 s-1) 1.40 0.30 9.16 0.01 0.34 0.57 35.70 <.001 7.80 <.001 1.21 0.31 1.02 0.41

TABLE 5. Analysis of variance with mixed effects testing the warming treatments, site and time effects on the N, P, K, Ca and Mg concentration of Mainthemum canadense and Eurybia macrophylla leaves.

Site × Site × Treat Site Treatment Time Site × Time Treat × Time Treatment × Time M. canadense F P F P F P F P F P F P F P N (g m-2) ------3.52 0.09 ------13.81 <.001 ------0.11 0.89 ------N (mg g-1) ------1.75 0.22 ------1.29 0.30 ------3.21 0.06 ------P (g m-2) ------8.11 0.02 ------2.51 0.11 ------0.77 0.48 ------P (mg g-1) ------1.82 0.21 ------2.54 0.11 ------0.24 0.79 ------K (g m-2) ------0.01 0.91 ------5.36 0.01 ------1.29 0.30 ------K (mg g-1) ------11.12 0.01 ------4.81 0.02 ------3.83 0.04 ------Ca (g m-2) ------0.12 0.74 ------2.23 0.14 ------2.90 0.08 ------Ca (mg g-1) ------3.81 0.08 ------11.69 <.001 ------1.08 0.36 ------Mg (g m-2) ------1.29 0.28 ------5.05 0.02 ------0.44 0.65 ------Mg (mg g-1) ------1.38 0.27 ------6.87 0.01 ------0.13 0.88 ------

E. macrophylla N (g m-2) 1.91 0.24 0.66 0.43 0.71 0.41 2.47 0.10 0.75 0.48 0.85 0.43 2.72 0.08 N (mg g-1) 0.38 0.57 2.79 0.11 0.25 0.63 33.23 <.001 3.74 0.03 6.06 0.01 0.21 0.81 P (g m-2) 24.07 0.01 1.14 0.30 0.38 0.55 0.27 0.77 0.56 0.58 1.13 0.33 0.53 0.59 P (mg g-1) 16.51 0.02 2.58 0.12 1.00 0.33 9.60 <.001 0.35 0.71 2.39 0.11 0.18 0.84 K (g m-2) 30.50 0.01 0.00 0.97 0.01 0.94 1.37 0.27 1.69 0.20 0.27 0.77 0.39 0.68 K (mg g-1) 20.24 0.01 0.35 0.71 1.61 0.23 12.87 <.001 1.14 0.33 0.36 0.84 0.25 0.86 Ca (g m-2) 0.24 0.65 21.49 0.00 0.55 0.47 48.02 <.001 1.18 0.32 2.04 0.15 0.11 0.90 Ca (mg g-1) 7.34 0.05 6.80 0.02 0.69 0.42 107.03 <.001 1.37 0.27 6.10 0.01 0.58 0.57 Mg (g m-2) 2.23 0.21 7.48 0.01 2.72 0.11 28.27 <.001 0.46 0.64 5.57 0.01 0.64 0.54 Mg (mg g-1) 18.16 0.01 8.41 0.01 1.83 0.19 39.06 <.001 0.56 0.58 12.26 <.0001 0.17 0.84

Figure Legends

Fig. 1. The network of hypotheses being tested in the present study.

Fig. 2. Percentage of the total available light that reached M. canadense (A, C) and E. macrophylla (B, D) plants throughout the season for three warming treatments at the two study sites (CFC and HWRC). The curves are produced by fitting an inverted Gompertz curve on the mean value for each treatment, at each date. The arrows represent the average date of leaf unfolding (not emergence) of the 3 treatments.

Fig. 3. Effects of the warming treatments on the mean (± s.e.) dates of leaf unfolding, leaf withering and length of the growing season for A) M. canadense and B) E. macrophylla at each experimental site.

Fig. 4. Effect of the warming treatments on the mean (±s.e.) total leaf area of A) Mainthemum canadense and B) Eurybia macrophylla at both study sites, CFC and HWRC. Letters represent significant differences according to protected LSD tests.

Fig. 5. Effect of the warming treatments on different aspects of the reproductive output of Maianthemum canadense at the CFC site. A. The mean (±s.e.) percentage of ramets that flowered per treatment. B. The mean (±s.e.) number of flowers per ramet. C. The mean (±s.e.) number of fruits per ramet. D. The mean (±s.e.) dry mass of individual fruits. Letters represent differences according to a protected LSD test.

Fig. 6. Effect of the warming treatments on different aspects of the reproductive output of Eurybia macrophylla; with results shown pooled across both sites. A. The mean (±s.e.) percentage of ramets that flowered per treatment. B. The mean (±s.e.) number of flower per ramet. C. The mean (±s.e.) dry mass of individual seeds (akenes). Letters represent differences according to protected LSD tests.

Fig. 7. Effect of the warming treatments on the mean (±s.e.) specific leaf area of Maianthemum canadense at the CFC site (A) and of Eurybia macrophylla at CFC (B) and

HWRC site (C) throughout the season. There was no M. canadense collected at HWRC site because there were too few plants in the plots.

Fig. 8. Effect of the warming treatments on the mean (±s.e.) photosynthetic rates under saturating light conditions (Asat) for Maianthemum canadense and Eurybia macrophylla through the season on an area basis (A-B-C-D) and on a mass basis (E-F-G). Effect of the warming treatments on the mean (±s.e.) respiratory rates measured at night (Rn) for M. canadense and E. macrophylla through the season on an area basis (H-I-J-K) and on a mass basis (L-M-N). Asterisks represent significant differences per date based on contrasts analyses done a posteriori when the interaction treatment × time or treatment × site × time was significant.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8