Phenotypic Plasticity and Genetic Variation of Leaf Pigment Content in a Southern Beech

12 Physiological, morphological and phenological attributes are potentially adaptive traits 13 that determine functional responses to certain environmental conditions. They are crucial for 14 understanding adaptations to environmental variation along a species natural range. In 15 particular, leaf pigment content can be a good proxy to the physiological and phenological tree 16 state. Our goal was to evaluate the variation in pigment traits among Nothofagus alpina 17 populations in a common garden trail during two years to infer local adaptation and/or 18 phenotypic plasticity. We also aim to analyse the correlation between pigment traits and 19 phenological traits and climatic data from the geographic location of the populations. To 20 comprise the entire range of the species in Argentina, we analysed 400 individuals from eight 21 natural populations coming from four lake watersheds. Pigment traits were estimated using a 22 spectrophotometer and analysed with linear mixed model (LMM). Significant differences 23 among fixed factors (populations - years and watershed – years) were found in chlorophyl a, 24 b, total, carotenoids and anthocyanins concentrations. Higher concentrations were found for 25 2018, year with the highest number of rainy days and accumulated precipitation. Two 26 populations (Boquete and Tren Tren) were always the most contrasting ones. was significant in almost all cases. Conversely, the correlation between the means of 29 population pigment concentrations and the main geographic, climatic and bioclimatic variables 30 of the home range were not significant . The significance of the environmental factor (year) in 31 the linear mixed models tested is evidence of phenotypic plasticity of pigment content, 32 suggesting flexibility for acclimatization to moderate inter-annual changes in climatic 33 conditions. The significance of population and watershed and the influence of the family factor 34 on the variance of the pigment traits are evidence of the genetic control as well as the 35 potential adaptive value of leaf pigment content in N. alpina, giving a base for adaptation to a 36 long-lasting change in climate. High correlations between phenological and pigment traits 37 indicate that, in N. alpina , the determination of only one pigment concentration could be used 38 as a proxy of bud burst, senescence and growing degree days. 39

was significant in almost all cases. Conversely, the correlation between the means of 29 population pigment concentrations and the main geographic, climatic and bioclimatic variables 30 of the home range were not significant. The significance of the environmental factor (year) in 31 the linear mixed models tested is evidence of phenotypic plasticity of pigment content, 32 suggesting flexibility for acclimatization to moderate inter-annual changes in climatic 33 conditions. The significance of population and watershed and the influence of the family factor 34 on the variance of the pigment traits are evidence of the genetic control as well as the 35 potential adaptive value of leaf pigment content in N. alpina, giving a base for adaptation to a 36 long-lasting change in climate. High correlations between phenological and pigment traits 37 indicate that, in N. alpina, the determination of only one pigment concentration could be used 38 as a proxy of bud burst, senescence and growing degree days. 39 40

INTRODUCTION 41
Climate has long been identified as a key factor shaping the geographic distribution of plants 42 (Major 1951). In present times, however, climate is changing rapidly (IPCC 2014) shifting 43 species distribution (Lloyd et al. 2011;Fisichelli et al. 2014; Parmesan and Hanley 2015). Tree 44 species are particularly affected by these changes, that occur faster than their long life-cycle, 45 and could respond through migration, phenotypic plasticity and local adaptation (Parmesan 46 and Hanley 2015). However, uncertainty remains about dispersal rates in pace with rapidly 47 Potentially adaptive traits are physiological, morphological and phenological attributes 56 that determine functional responses to certain environmental conditions (Violle et al. 2007). 57 Variation in leaf pigment content can be a good proxy to the physiological tree state (Sims and 58   Pigments were extracted following the protocol reported by Gould et al. (2000). 163 Chlorophylls and carotenoids were extracted using a solution of acetone: H2O (4:1 v/v) while 164 anthocyanins were extracted using a solution of 3 M HCl: H2O: MeOH (1:3:16 v/v/v). For each 165 tree, three foliar discs, 1 cm 2 total area, were cut and submerged in each solution during 24 166 hours in dark at 4 °C. After this period, centrifugation (1 min at 1000 rpm, 4°C) was used to 167 separate solid residues. Pigment concentrations were estimated using a Pharmacia Biotech 168 Ultrospec 1000 spectrophotometer. For the acetone extracts, absorbance was measured at Evaluation of environmental variation between both years at the common garden trial was 186 based on temperature, precipitation and aridity index data. First, we tested the difference 187 between years for monthly mean, minimum and maximum air temperature (Tmean, Tmin, Tmax 188 respectively), and aridity index (AI) by t-tests. For monthly total precipitation (precip.) and rainy days (rd) we used the Mann-Whitney test. Normality of the data distribution was assessed 190 using a histogram and a Shapiro test for each variable. Homoscedasticity was checked with a 191 Bartlett test. Second, we compared the distribution pattern of monthly mean air temperature, 192 total precipitation and aridity index for both years and mean historical values for the period

211
For both LMMs, we used the "lme4" package (Bates et al., 2016) in R 3.3.0 software 212 (RStudio Team 2020). For each pigment trait model, we used a "residuals versus fitted plot" to 213 detect non-linearity, unequal error variances, and outliers. A likelihood ratio test (LRT) was 214 used to analyse the significance of random effects, considering the complete model and a Since the measurements of pigment content were made in consecutive years, the 217 same development stage (juvenile trees) was evaluated. So, the traits are not expected to be 218 influenced by the age of the trees, and the possible ontogenetic effect on the inter-annual 219 variation was consequently dismissed. 220 Pearson correlation tests were performed between pigment traits and the main 221 geographic and environmental variables of each natural population (Table 1). In addition, we 222  (Table 2). 244 However, no significant differences were found, by t-test or Mann-Whitney test, between years 245 for the analysed variables (i.e. monthly rainy days, total precipitation, mean, minimum and 246 maximum air temperatures, and aridity index, Table 2). 247 Table 2. Environmental conditions for the common garden trial located in Las Golondrinas for 249 the evaluated years, including rainy days (rd), total precipitation (precip.), and mean values of 250 monthly total precipitation (precip.), total rainy days (rd), mean, minimum and maximum air 251 temperature (Tmean, Tmin,Tmax respectively) and aridity index (AI

255
Considering variation over the year, monthly, mean air temperature, total precipitation 256 and aridity index were not equally distributed for both years (Fig. 2). Total precipitation for 257 2018 was globally higher than the historical reference, except during late autumn and early 258 winter; while total precipitation for 2019 was globally lower than the historical reference ( Fig.  259 2). Mean temperatures were higher than those of the historical periods for almost all months, 260 and for 2019 it was globally higher than for 2018. Changes on total precipitation showed a 261 higher impact on aridity index (Fig. 2).

Leaf pigment variation 275
Fixed factors were significant for both LMMs, while interaction between both fixed factors (i.e. 276 population and year, or watershed and year) was not detected for any pigment trait (Table 3). 277 Both for populations and for watersheds, the concentration of Chla, Chlb, Chla+b, Car and Ant 278 all presented significant differences between years (p < 0.05), with higher concentrations in 279 2018 than in 2019 ( Fig. 3 and Fig. 4). The difference was stronger for anthocyanin   showing the highest and lowest concentration respectively; while the other watersheds were in 304 the middle of these extremes. The block factor had a significant effect (pLRT<0.001), although it 305 only explained around 5% in Chla and Ant and around 10 % in Chla+b, Chlb and Car of the 306 total variance. The population nested into watershed factor did not have a significant effect in 307 the model. 308 For both populations and watersheds, anthocyanins showed the highest level of intra-309 specific variation. For example, for populations, pigment content was 35%, 20%,17.5%, 17%, 310 and 16% higher for Boquete than for Tren Tren, for Ant, Chlb, Car, Chla and Chla+b 311 respectively. Considering watersheds, pigment content was 17%, 14%, 13%, 13% and 12% 312 higher for Lolog than for Lacar, for Ant, Chlb, Car, Chla+b and Chla respectively. 313 On the contrary, Chla / Chlb ratio and Chla+b / Car did not present significant 314 differences among populations, watersheds or years, with the exception of Chla+b / Car for 315 which the difference between 2018 and 2019 was significant (Table 3). 316

Fig. 3. Year by Population interactions for leaf pigment content in Nothofagus alpina. 318
Represented values are the mean effects on the LMM, with standard errors represented by 319 whiskers. Icon shape indicates watersheds: circles for Tromen, triangles for Curruhue-320 Huechulafquen, squares for Lolog, diamond for Lácar. double column fitting image. 321

Correlation and principal component analyses 327
At population level, the correlation between the means of pigment concentrations and 328 phenological traits was significant in almost all cases, being high and negative for day of the 329 year to bud burst (DOY BB) and growing degree days (basal temperature = 5 °C; GDD Tb 5° 330 concentrations and the main geographic and environmental variables of the home range 333 (Table 1) were not significant. Moreover, significant correlations were not detected between 334 pigment content and climatic and bioclimatic variables. The first two dimensions of PCA analysis express 93.55% of the total variation (Fig. 6). (correlation = 0.87, 084 and -0.72 respectively) (Fig. 6b)

. Regarding watersheds, Lolog and 355
Lácar showed a high positive and negative correlation with this dimension (cos 2 = 0.99 and 356 0.98 respectively). 357 The second dimension separate Tromen Alto (to the bottom of the graph, 358 characterized by a strongly negative coordinate on the axis) and Paimún (to the top of the 359 graph, characterized by a strongly positive coordinate on the axis) populations (Fig. 6a). variables with good representation in the plane (i.e. cos 2 greater than 0.6).

Phenotypic plasticity of N. alpina pigment content 375
The two analysed years varied particularly in the distribution of total precipitation and, 376 consequently, aridity index over months. This inter-annual variation constitutes a good 377 scenario to test temporal phenotypic plasticity. The significance of the environmental factor 378 (year) in the linear mixed models tested is evidence of phenotypic plasticity of pigment 379 content, since their concentration varied when the same genotypes were exposed to different

Ethics approval and consent to participate 498
Not applicable 499

Consent for publication 500
Not applicable 501

Availability of data and material 502
The data will be available upon acceptance in the public institutional repository INTA Digital 503 (www.repositorio.inta.gob.ar) 504

Competing interest 505
The authors declare that they have no known competing financial interests or personal 506 relationships that could have appeared to influence the work reported in this paper. 507

Author contributions 512
PM and VEM conceptualized the study. JAR, PM, VEM performed sample collections. JAR 513 performed pigment quantification and statistical analyses and wrote the manuscript with the 514 contributions of PM and VE. MP installed and maintained the common garden trial. All authors 515 were involved in the final revision and editing process of the manuscript. 516

Acknowledgements 517
We would like to thank Fernando Barbero for his collaboration in seed collection and seedling