Impacts of intraspecific variability and local adaptation on the ecophysiology of mosses: an example with Sphagnum magellanicum

Background Bryophytes are a diverse plant group and are functionally different from vascular plants. Yet, plant ecology theories and hypotheses are often presented in an inclusive term. The trait-based approach to ecology is no exception; largely focusing on vascular plant traits and almost exclusively on interspecific traits. Currently, we lack information about the magnitude and the importance of intraspecific variability to the ecophysiology of bryophytes and how these might translate to local adaptation—a prerequisite for adaptive evolution. Method We used transplant and factorial experiments involving moisture and light to ask whether variability in traits between morphologically distinct individuals of Sphagnum magellanicum from habitat extremes was due to phenotypic plasticity or local adaptation and the implications for the ecophysiology of the species. Key Results We found that the factors that discriminated between the plant origins in the field did not translate to their ecophysiological functioning and the pattern of variability changed with the treatments, which suggests that the trait responses were due largely to phenotypic plasticity. The trait responses suggest that the need for mosses to grow in clumps where they maintain a uniform growth rate may have an overriding effect on responses to environmental heterogeneity, and therefore a constraint for local adaptation. Conclusion The circumstances under which local adaptation would be beneficial in this plant group is not clear. We conclude that extending the trait-based framework to mosses or making comparisons between mosses and vascular plants under any theoretical framework would only be meaningful to the extent that growth form and dispersal strategies are considered.

• Key Results We found that the factors that discriminated between the plant origins in the 36 field did not translate to their ecophysiological functioning and the pattern of variability 37 changed with the treatments, which suggests that the trait responses were due largely to 38 phenotypic plasticity. The trait responses suggest that the need for mosses to grow in 39 clumps where they maintain a uniform growth rate may have an overriding effect on 40 responses to environmental heterogeneity, and therefore a constraint for local adaptation. 41 Introduction group may not necessarily have the same ecological meaning as it is understood for vascular availability and high irradiance that are prevalent in hummocks, we therefore predict that 145 morphological and physiological responses of hummock-originated plants would be less 146 sensitive to light and drought treatments compared with hollow-originated plants. We extracted four hummock monoliths, which comprised a continuous carpet of S. 164 fuscum into surface peat to a depth of about 20 cm. The monoliths allowed us to incorporate the 165 ecophysiological peculiarities (e.g. neighbourhood effect and vertical movement of moisture 166 through litter matrices) of our study system into the experiment. Each monolith was gently 167 placed in an 8.83-litre cylindrical pot. Each monolith was partitioned into equal halves with a 168 stick, which was inserted horizontally into the surface of the moss carpet in each pot. Individuals 169 of S. magellanicum from the two home environments (hummock versus hollow) were randomly 170 assigned to a monolith and were inserted into the carpet of S. fuscum. Specifically, we inserted 171 fifteen S. magellanicum hummock-originated individuals into one half of each monolith and breadth of "home" environment for individuals that were collected on hummocks in terms of in dimension and are built from PVC pipes. The shade boxes were covered with breathable 50% 179 neutral density shade cloth. The 50% shade approximates the proportion of the photosynthetic 180 active radiation (PAR) admitted into the Sphagnum carpet by the dominant vascular plant species 181 (Myrica gale) at our site. This was obtained by measuring PAR below and above the canopy 182 using the point sensor of a LI-250 light meter (LI-COR, Lincoln, Nebraska). These 183 measurements were used to compute percentage of light admitted into the moss surface. The 184 above canopy PAR ranged from 1206 -2035 µmol m -2 s -1 whereas below canopy values ranged 185 from 224 -1714 µmol m -2 s -1 . We did not find a difference in the moisture profiles of hummocks 186 sampled along moisture gradient in our site, therefore, we did not vary moisture for this 187 experiment. 188

189
Factorial light x moisture experiment 190 Our second experiment involved a 3 x 2 factorial experiment with two plant origins (hummock 191 versus hollow), two light treatments (full light; 50% light) and two water treatments (saturated; 192 low water). This experiment represents the breadth of "home" environment for hollow-originated 193 individuals in terms of substrate conditions, while hummock-originated plants in this case were 194 transplanted onto "away" substrates. The shade treatment was imposed as described above. The 195 drought treatment was created by maintaining treatment pots at an average volumetric water 196 content of about 12%, which is the mean summer volumetric water content at the top 1 cm of 197 moss in the field site. The saturated water treatment was maintained by monitoring and topping 198 up the experimental pots with water, and volumetric water content consistently exceeded 21%. 199 The water contents across all experimental pots were monitored with a portable Hydrosense soil 200 moisture meter (Campbell Scientific, Inc., USA). 201 The experimental pots were filled with 3 cm of deep peat moss underneath a 1 cm layer 202 of surface peat. The deep peat was from a commercial source while the surface peat was 203 extracted from the field in an area near the Sphagnum collections in hollow. The pots were 227.4 204 cm 3 in size, with holes at the base through which water was fed into the pots. There were 9 plants moisture treatments), which we replicated twice. Thus, a total of 144 plants were used in the 207 experiment. Because bogs are nutrient-poor and typically fed by rainwater, the plants were not 208 fertilized and were watered exclusively with rainwater that was harvested in Guelph. 209 In the context of local adaptation, each experiment contains aspects of a "home" versus 210 an "away" treatment (Kawecki and Ebert, 2004; Blanquart et al., 2013). In the first transplant 211 experiment, hummock individuals transplanted onto the hummock mesocosms represent a 212 "home" treatment while hollow individuals represent an "away" treatment. However, this 213 transplant experiment is an incomplete design but it was not possible for us to maintain hollow 214 mesocosms due to the extremely unconsolidated (low bulk density) nature of hollow surface 215 soils and species homogeneity. The combination of the experiments nonetheless represents a 216 range of environment that the species is typically exposed to and allows us to at least reduce the 217 confounding effects of unmeasured environments (Kawecki and Ebert, 2004). 218 219

Quantification of traits 220
The two experiments ran fully from July 2016 to January 2017. At the end of the experiments, 221 we measured a suite of traits on individuals from each treatment. We quantified two traits related 222 to growth, including vertical growth rate (growth per time) and biomass. We also measured 223 allocation of biomass into capitulum, branch, and stem. The capitulum is taken as the top 1 cm of 224 the plant (Clymo 1970). Branch mass was determined by removing, drying and weighing the 225 stem, leaves and branches (fascicles), which were collectively measured as branch mass. The 226 exposed stem after removal of capitula and branches was dried and weighed to obtain stem mass. 227 We also quantified the dark respiration as a measure of metabolic activity. Respiration 228 rates were measured on six individuals per treatment, which were selected at the end of the 229 experiment. For these individuals, we placed the entire plant in a dark glass jar. The jars were 230 sealed with stopcocks and placed under their respective treatment environment. The CO 2 in the 231 jar headspace was drawn three times at 3 hr intervals with gas-tight syringes. The CO 2 232 concentration was analyzed with an EGM-4 infrared gas analyzer (PP Systems, Hitchin, 233 Hertfordshire, UK). We performed linear regressions of CO 2 concentration against time, using 234 the slopes of these relationships as our measurement of respiration rate. We then used the dry photosynthetic efficiency. The dark-adapted F v /F m measurements were taken at the end of the 238 experiment. Individuals from each treatment were placed in the dark for at least 6 hours to ensure 239 that Q A electron acceptors are fully reduced and that reaction centers are in the 'open' state. We 240 then quantified dark-adapted F v /F m on each plant using a pulse-modulated fluorometer (OS1p,  241 Opti-Sciences, Hudson, NH). 242 243

Statistical analyses 244
The data were explored for normality and where there was a departure from normality (vertical 245 growth rate and branch mass), they were transformed using a logarithm transformation. Because 246 the plants in the hummock transplant experiment were grown in only four pots, we tested for 247 differences in trait values using mixed effect models, where we analyzed pot ID as a random 248 effect to account for lack of independence. Multiple mean comparisons were obtained for models 249 with interaction effects using "lsmeans" package in R. We tested for mean trait values in the 250 factorial experiment using 3-way ANOVA and obtained multiple mean comparisons for 251 interaction effects using Tukey HSD. We explored patterns of trait variability across 252 experimental treatments by partitioning the variance in the data using the varpart function in R 253 package "Vegan". We used this approach combined with redundancy analysis to examine how 254 the experimental treatments influenced within-trait variability and total trait variability. All 255 analyses were performed in R 3.2 (R core Development Team 2015) and all statistical tests were 256 conducted at α = 0.05. 257

Hummock-transplant experiment 261
Hummock-originated plants had lower F v /F m than hollow plants ( Fig. 1a) with no other 262 significant main effects or interactions (Table 1). Vertical growth rate, capitulum mass, and 263 respiration were consistently higher under the shade than the high light treatment (Fig. 1b & c). 264 Total biomass and stem biomass was influenced by a plant origin x light interaction (Fig 1d). 265 Hummock plants tended to have lower total and stem biomass than hollow plants but only in the 266 shade treatment. 267 example between vertical growth rate and respiration rate and between respiration rate and 270 biomass for both hummock and hollow plants (r 2 = 0.24, p < 0.05 and r 2 = 0.56, p < 0.001) and 271 hollow plants (r 2 = 0.30, p < 0.05 and r 2 = 0.73, p < 0.001) ( Fig. 2a & b). However, the effect of 272 plant origin on these relationships was not statistically significant (p > 0.05). 273 For most traits, plant origin did not explain a significant amount of variation in individual 274 traits (0 -10%), while light explained between 0 -46% ( In the factorial experiment, traits were more generally influenced by the main effects of origin 283 and moisture than their interaction effects or the main effect of light (Table 3). The post-hoc tests 284 showed that capitulum mass was greater in hummock plants than in hollow plants under the high 285 moisture treatment (p < 0.05) but did not significantly differ between the plant origins under the 286 low moisture treatment. The opposite trend was true for branch mass as hollow plants had a 287 greater branch mass than hummock plants under the high moisture treatment (p < 0.05) but there 288 was no difference in branch mass between the origins under the low moisture treatment. 289 Interestingly, the stem mass of hollow plants subjected to low moisture was greater than stem 290 mass of hummock plants subjected to high moisture (p < 0.001). Vertical growth rate was fastest 291 under the high moisture treatments regardless of light (Fig. 3a) and was lower under the low 292 moisture treatments. Biomass was greatest at the high light and high moisture treatment and 293 tended to be lowest under the low moisture treatments across both light treatments (Fig. 3b). 294 F v /F m was higher in hollow individuals than in hummock individuals, especially under high 295 moisture (Fig. 3c). Respiration was higher under high moisture than the low moisture treatment 296 and did not vary with light (Fig. 3d).
correlations between respiration and biomass and between respiration and vertical growth for 299 both hummock (r 2 = 0.25, p < 0.001 and r 2 = 0.57, p < 0.001) and hollow (r 2 = 0.53, p < 0.001 300 and r 2 = 0.65, p < 0.001) plants ( Fig. 4a & b). However, unlike in the hummock transplant 301 experiment, the pair-wise comparison of the relationships between respiration and biomass was 302 influenced by plant origin (z = 2.65, p = 0.01). 303 Plant origin explained the most variation in stem mass (44%) relative to moisture and 304 light. In general, the influence of the light treatments explained little or no variation among traits 305 in this experiment. Moisture explained a significant amount of variation in all traits except for 306 capitulum mass and was particularly important for respiration and branch mass variation. We 307 found that plant origin and moisture explained similar levels of total variation across traits (Table  308 4). The data were also split into two independent datasets based on plant origin and were 309 accordingly explored for variability due to light and moisture effects. colour and some with a tint of green, which is consistent with the findings that pigmentation of S. 372 magellanicum is plastic (Yousefi et al., 2017). This means that the pigmentation is induced by 373 the environment. The generally low F v /F m in hummock plants suggests that the reddish 374 pigmentation might prevent an optimal photosynthetic response, which would mean that there is 375 a cost to achieving photoprotection. However, we did not find any relationship between F v /F m 376 and total biomass, which is often used as a proxy for fitness in plants (Younginger et al., 2017).  (Bolnick et al., 2011). For instance, high trait variability could help a population 394 to transition to a new trait optima and therefore to a new adaptive peak through natural selection 395 (Bürger, 1999). In this study, most of the variability remained unexplained by our treatments. 396 However, it is important to note that most traits measured in this study exhibited low levels of 397 variation. It is also important to note that clonality is common in Sphagnum, especially at fine 398 scales, which may lead to low phenotypic variation. Low phenotypic variation may be 399 advantageous for morphological integration. Although our sampling design was intended to 400 avoid repeatedly sampling clones, it is not uncommon for a Sphagnum population to be 401 responses to a single environment. Additionally, because phenotypic differentiation may not 422 necessarily have a genetic basis, it is possible in a common garden experiment to confuse or 423 conflate adaptive differentiation arising from phenotypic plasticity with that arising from local 424 adaptation (Gienapp et al., 2008). 425 Finally, there is an on-going taxonomic revision to S. magellanicum. The species is 426 considered a complex, comprising at least three species-S. divinum and S. medium in eastern 427 North America, and S. magellanicum sensu stricto in South America (Hassel et al., 2018). These 428 species have distinct morphological, molecular and distributional characters. The preliminary 429 study suggests that S. medium has an amphi-Atlantic distribution while S. divinum is circumpolar 430 in its distribution. Since the pigmentation of "S. magellanicum" (as we currently know it) lacks 431 genetic basis (Yousefi et al., 2017) and considering the pattern of distribution of these species 432 relative to our field site in Southern Ontario, it is unlikely that we sampled across a mix of S. 433 medium and S. divinum in a way that would bias our findings. Also, considering that origin had 434 little effect on trait variability, a more likely scenario is that we sampled one species or the other. 435 However, because further study is required on the distribution and identification of these 436 subspecies (Hassel et al., 2018), we are unable to accordingly characterize our species and 437 therefore maintain the name S. magellanicum for the purpose of this study. 438 439

Conclusion 440
In summary, we explored the magnitude and pattern of trait variability in S. 441 magellanicum from contrasting habitats in the context of phenotypic plasticity and local 442 adaptation. We found that the trait responses were due largely to phenotypical plasticity with 443 little influence on whether plants originated from hummocks or hollows. We also found that trait 444 variability depends on the prevailing light or moisture environment. However, most trait 445 variation remained unexplained by our experimental treatments. Collectively, our results suggest 446 that using traits to draw inferences about the ecology of Sphagnum would require an 447 understanding of the mechanisms driving traits and the pattern of trait variability. Lastly, because 448 morphological integration may have an overriding influence on growth traits, we are unclear 449 about the circumstances under which local adaptation might occur or benefit this plant group. 450 We hope that future studies will further explore this area of inquiry in mosses, with consideration