High plasticity in germination and establishment success in the dominant forest tree Fagus sylvatica across Europe

Aim: Distribution ranges of temperate tree species are shifting poleward and upslope into cooler environments due to global warming. Successful regeneration is crucial for population persistence and range expansion. Thus, we aimed to identify environmental variables that affect germination and seedling establishment of Europe's dominant forest tree, to compare the importance of plasticity and genetic variation for regeneration, and to evaluate the regeneration potential at and beyond the southern and northern distribution margins. Europe. Time period: 2016– 2018. Major taxa studied: European beech ( Fagus sylvatica (L.)). Methods: We investigated how germination, establishment and juvenile survival change across a reciprocal transplantation experiment using over 9,000 seeds of beech from 7 populations from its southern to its northern distribution range margins. Results: Germination and establishment at the seedling stage were highly plastic in response to environmental conditions. Germination success increased with warmer


| INTRODUC TI ON
Distribution ranges of species are projected to shift polewards or upwards in elevation (Lenoir et al., 2008;Parmesan, 2006;Sykes et al., 1996), whereas populations at the rear edge are threatened due to global warming (Thomas et al., 2004;Thuiller et al., 2005).
However, these predictions from species distribution models generally neglect the potential for in situ responses of organisms to changing environmental conditions through phenotypic plasticity, local genetic adaptation and demographic processes (Aitken et al., 2008;Savolainen et al., 2007;Scherrer et al., 2020;Valladares et al., 2014).
Local genetic adaptation results in highest fitness of populations at their origin compared to non-local genotypes of the same species at the same location (Kawecki & Ebert, 2004). Thus, genetic differences can explain variation in traits between populations. However, certain traits might also be strongly affected by the environment and thus respond plastically to changing environmental conditions. Local adaptation is expected to be particularly important in marginal populations due to stronger selective pressure of the environment at range margins than in the distribution centre and due to a probably limited genetic mixing caused by geographic isolation (Choler et al., 2004;Kawecki, 2008;Paul et al., 2011). In the case of strong local adaptation, assisted gene flow within a species range, that is the transplantation of genotypes from origins climatically similar to the expected future conditions of a given site, or assisted migration, that is the transplantation of genotypes outside of a species range, might be considered as an adaptation strategy by forest managers to ensure sustainable forest ecosystem services under rapid change in climatic conditions (Aitken & Bemmels, 2016;Williams & Dumroese, 2013). In contrast, plastic responses to changing environmental conditions could allow for short-term acclimatization and thus might buffer the impact of climate change. These in situ plastic responses play an important role especially for plant species with limited migration potential due to low seed dispersal capacity and a long life span . This is especially true for trees, as they are long-living organisms and can hardly compensate the consequences of ongoing rapid climate change through migration (Jump & Peñuelas, 2005).
Successful regeneration plays an essential role in the persistence of populations, in particular for climax forest tree species. Therefore, germination and establishment of young trees are critical bottlenecks for the persistence of the population, especially as young trees respond more sensitively to changing climatic conditions than adult trees (Jackson et al., 2009). Moreover, germination and establishment are under strong natural selection and might favour rapid local adaptation (Donohue et al., 2010;Petit & Hampe, 2006). However, knowledge about plasticity and local adaptation in the early lifehistory traits is still scarce, though these processes are critical to better predict the potential of natural regeneration, for projecting distribution range shifts, and for conservation strategies in times of climate change.
The few studies on germination and establishment of forest trees in the face of climate change so far had a local to regional focus (e.g. Arana et al., 2016;Dulamsuren et al., 2013). Successful tree regeneration is obviously important for the expansion of the distribution range at the leading edge. Limited, but still existing success of regeneration at the rear edge, however, is potentially as important because the strong selection during the earliest life stages may determine surprisingly quick genetic adaptation (Jump, Hunt, Martínez-Izquierdo, et al., 2006) and thereby mitigate range and declined with colder air temperature, whereas establishment and survival were hampered under warmer and drier conditions. Germination differed among populations and was positively influenced by seed weight. However, there was no evidence of local adaptation in any trait.

Main conclusions:
The high plasticity in the early life-history traits found irrespective of seed origin may allow for short-term acclimatization. However, our results also indicate that this plasticity might not be sufficient to ensure the regeneration of beech in the future due to the low survival found under dry and hot conditions. The future cli- contraction. Studies accounting for the whole distribution range from the rear to the leading edge are therefore necessary to better understand population dynamics under climate change (Rumpf et al., 2018).
European beech (Fagus sylvatica L., hereafter called beech), the dominant native forest tree in Central Europe, grows under a wide range of climatic and environmental conditions (Bolte et al., 2007;Fang & Lechowicz, 2006;Leuschner et al., 2006). The potential distribution range of beech is limited by late frost events as well as severe winter frost towards the north in Sweden and north-east in Poland, and by heat and drought towards the rear edge in Greece, Italy and Spain (Bolte et al., 2007;Giesecke et al., 2007;Jump, Hunt, & Peñuelas, et al., 2006;van der Maaten et al., 2017). Beech is absent or hardly grows in regions with an Ellenberg's climate quotient [average July temperature (°C)*1000/average annual precipitation (mm)] above 30°C/mm (Leuschner & Ellenberg, 2017).
Under rapid global warming, the high plasticity found in phenology and fitness-related traits like growth in beech could help in mitigating range contractions (Gárate-Escamilla et al., 2019;Kramer et al., 2017;Vitasse et al., 2010). Genetic diversity of beech populations was found to be lower in the centre and towards the leading edge of its distribution, whereas it becomes higher toward the rear edge of its distribution due to the postglacial recolonizations from glacial refugia (Magri et al., 2006). This might favour local adaptations at the rear edge, which is supported by findings of local adaptation of dry-marginal populations to drought (Bolte et al., 2016;Thiel et al., 2014). However, the reported lower genetic diversity towards the north is valid at population level, whereas the genetic variation within populations may remain quite high across the whole range of the species (Vornam et al., 2004). Adaptive traits and performance of beech under drought are genetically determined (Cuervo-Alarcon et al., 2021), arise from both local genetic adaptation and phenotypic plasticity (Bolte et al., 2016;Stojnić et al., 2018), and are also moderated by other ecological factors like light availability (Wang et al., 2021). To better understand the population dynamics of this species across its whole distribution range, it is therefore crucial to consider the increasing likelihood of drought events in large parts of the distribution of beech with climate change  and its projected distribution range shift towards the north (Kramer et al., 2010;Saltre et al., 2015). For this, studies that focus on the potential and limits of local adaptation and plasticity at the recruitment stage are urgently needed, as germination and juvenile establishment are crucial life-history traits for successful natural regeneration.
Here, we used a unique, fully reciprocal transplantation experiment of European beech along a large climatic gradient from its southern distribution margin in Spain to its northern margin in Sweden and Poland as well as beyond its current distribution range, in order to assess the potential for plasticity and local adaptation to the environment during its early life stages. We transplanted more than 9,000 seeds, harvested and sowed in the same year at all sites, and determined the germination success as well as the success of establishment after the first year and survival after the second year as fitness traits for the earliest life stages. We aimed: (a) to identify which environmental variables of the transplant site or the population origin affect these earliest life stages; (b) to directly compare the importance of plasticity and genetic variation for germination, seedling establishment and survival of juvenile trees. Here, we expected local environmental conditions to dominate the outcome, that is a prevalence of plasticity over genetic adaptation; and (c) to explicitly evaluate the potential for regeneration at and beyond the southern and northern distribution margins.

| Experimental design and trait measurements
We conducted a fully reciprocal transplantation experiment across and beyond the distribution range to evaluate the importance of plasticity and local adaptation on germination and establishment success of beech ( Figure 1, Table 1). The sites have diverse winter cold and summer drought conditions (Table 1, Figure 1a, see Supporting Information Table S1 for the climatic characteristics of the population origins) as these are important variables that limit the distribution range of beech (Bolte et al., 2007;Giesecke et al., 2007;Jump et al., 2006;van der Maaten et al., 2017). All sites within the range were beech dominated. The external sites beyond the distribution range were surrounded by deciduous trees and they were dominated by oak in Poland (Quercus robur L.) and Spain (Quercus pyrenaica Willd.) and by mountain ash (Sorbus aucuparia L.) in Sweden. The canopy cover at the experimental sites was comparable between all sites (approx. 60%-70% canopy closure). The soil texture of all sites varied little from poor silty sand to sandy silt (Supporting Information Table S2). The litter layer was removed before planting the seeds.
In autumn 2016, open-pollinated seeds of all populations were collected with a slingshot (Notch BIG SHOT Standard Kit, SHERRILLtree) from four mother trees in close proximity (in a range of not more than 250 m) to the experiment unit at each population site. Beech regeneration is strongly driven by masting events, during which enormous amounts of seeds are produced that can exceed 7 million seeds per hectare (Övergaard et al., 2007). In conjunction with the widespread pollination of many trees during those mast years, a high number of seeds with a high genetic diversity are available (Tachiki & Iwasa, 2010). Thus, our seed collection during a continental-wide masting event (Ascoli et al., 2017) ensured high quantity and quality of seeds from all sites except for the Swedish sites, where poor fructification was observed. The swedish sites would have represented cold populations, however, the Polish population GD remained representative of cold populations in our data set. We checked the initial seed quality by discarding empty and infested beech nuts after examining each seed carefully and verifying with multiple destructively opened samples per site. A good initial viability of the seeds is supported by a relatively high germination success of more than 50% at some of our sites (Supporting Information Table S3, compare Bartsch et al., 2020;Barna, 2011). At each site, the experiment was replicated in 10 blocks into which seeds of each population and mother tree were systematically sown in autumn 2016 (n = 11 sites × 10 replicate blocks × 7 populations × 4 mother trees × 3 seeds = 9,240 seeds in total, i.e. 120 seeds per population at each site, Figure 1). Seeds to be sown in each replication block were drawn at random for each population. We planted the seeds maximally 3 weeks after harvest to allow for the important natural cold stratification of beech seeds (Barna, 2011;Suszka et al., 1996).
The seeds were covered by standardized seedless beech litter to simulate natural and comparable seed bed conditions at each site.
The seeds in each block were protected against predation by rodents and against potential incoming falling seeds using cages with small-meshed (5 mm) wire ( Figure 1c) and larger fences protected against deer and wild boar intrusion. Still, three blocks in Poland (site GD) were affected by rodents and had to be excluded from the analysis (252 seeds).

F I G U R E 1 (a) Site locations within (black dots) and beyond (red dots) the distribution range of
Fagus sylvatica (blue area), source: http:// www.eufor gen.org. The climatic gradient is based on differences between the warmest and coldest site in winter and between the driest and wettest site in summer; water balance is calculated by precipitation -potential evapotranspiration, the climatic parameters were averaged over the 30-year period 1986-2015 (Harris et al., 2014, CRU TS v. 4.03), (b) climatic distribution range of F. sylvatica (extracted and calculated from CRU TS v. 4.03, Harris et al., 2014) and (c) experimental unit (19 m 2 ) of 10 replicate blocks as it was set up identically at each study site (picture: site GD, Poland). See Table 1 for site ID codes used in (a)   Note: Geographic position [longitude degree (°E), latitude degree (°N)], elevation (above sea level), mean annual air temperature, mean coldest month air temperature, mean warmest month air temperature, and the volumetric soil water content measured at each site and averaged over the observation period (winter 2016/2017 to autumn 2018).
In autumn 2017, we measured the germination success after 1 year by counting all seeds that ever had germinated (i.e. also dead seedlings) by checking them for signs of germination (emergence of cotyledons). We measured establishment by counting all seedlings that were still alive when visually inspected in autumn (Supporting   Information Tables S3 and S4). The survival of seedlings was assessed again in the second year (autumn 2018). As response variables, we calculated the germination success (the ratio of germinated seeds to planted seeds after 1 year) as well as the establishment in the first year ( Table S1).

| Environmental data and statistical analyses
We analysed climatic effects (long-term average climate variables over 30 years, see Supporting Information Table S5)  For cases in which both the population origin (representing genetic differences between populations) and transplant site   Table S6]. In detail, populations originating from regions with wetter winters (Figure 3a,d) had a higher germination success.

| Germination increased under warmer and drier conditions -establishment and survival positively affected by cooler and moister conditions
Looking at the importance of the meteorological conditions at the transplant sites, germination increased when spring was warmer (more degree hours above 5°C) and with increasing soil temperature in winter (Figure 3b,c). Our analyses revealed temperature thresholds above which a sharp increase in germination was observed, that is, at soil temperature of 1°C in winter and more than 10,000 degree hours above 5°C in spring.

| Plasticity as well as genetic differences determine germination, whereas plasticity alone drives establishment and survival
Plasticity as well as genetic differences were important for germination, as the environmental conditions of both the population origin and the transplant site had an effect on germination (Figure 2).

| High germination at and beyond the southern range margin, but low germination of marginal populations
Across all sites, the populations from the driest winters, that is from

| Environmental parameters affecting the earliest life stages of beech
Germination success increased with moister winter conditions at the population origin as well as with warmer winter and spring conditions at the transplant site. Moist winters may result in a higher risk of seeds being damaged by pathogens such as fungi, which is known to increase with moisture in winter at least for grass species (Mordecai, 2012;Schafer & Kotanen, 2003;Wagner & Mitschunas, 2008). Thus, moist winter conditions at the origin could possibly trigger an investment into seeds that are more resistant against pathogens and lead to higher germination success at all sites compared to the seeds originating from dry winter conditions. The increased germination success at transplant sites with warmer spring conditions follows basic biological expectations of temperature enhancing physiological processes. However, the lower germination success with cooler soils might also be induced by a higher number of freeze-thaw cycles in the seeds, as freezing events occur more often at these sites than at sites with a soil temperature of more than 1°C in winter (Supporting Information Figure S9). and hence storage of carbohydrates might be hampered by warm air temperatures in autumn. Moreover, cold temperatures during autumn are required to build up cold hardiness and thus lower autumn temperatures may better prepare young trees for the coming winter (Larcher, 2005).

TA B L E 2
Germination success explained by each factor: transplant site, population origin, and their interaction (quasibinomial generalized linear model fit).

| Importance of plasticity, genetic differentiation, and local adaptation in early lifehistory traits
We found that the germination success differed among populations. The variation among populations could potentially be explained by genetic differences between the populations or by maternal effects, as germination was positively affected by seed weight. A positive influence of seed weight on germination was also found in Pinus and Quercus species in Spain (Gómez, 2004;Urbieta et al., 2008;Zas et al., 2013). Seed provisioning is of high importance for the heterotrophic growth of seedlings. However, also other maternal effects (i.e. epigenetic modifications) could possibly influence the germination of beech (Donohue, 2009).
Studies on maternal effects on germination in trees are scarce and should be increasingly conducted to better understand the mechanisms explaining the differences in germination success between populations.
F I G U R E 5 Differences in germination success in 2017 (germinated seeds/planted) between population (effect of the population origin) and transplant sites (effect of the environment). Populations and sites are ordered from the margin and beyond the margin to the core with respect to drought (a,b) and winter cold (c,d). The lower-case letters above the graphs represent significantly different groups in a leastsquares means post-hoc test over a quasi-binomial generalized linear model fit including population and site (without interaction) as model predictors. See Table 1 for site ID codes and Supporting Information Table S1 for population ID codes [Colour figure can be viewed at wileyonlinelibrary.com]

(a) (b) (c) (d)
No sign of local adaptation in germination success was detected as the germination success did not increase with a shorter climate transfer distance and no significant interaction between population and transplant site occurred. Likewise, establishment in the first and survival in the second year were only affected by the environment at the transplant sites and not by the population origin, revealing a high plasticity and no indications of local adaptations in these traits. This finding corresponds to our expectations that local environmental conditions, and thus plasticity, mainly shape the natural regeneration of beech. A high plasticity in beech was also found by other studies for growth, young tree survival, phenology, and in root traits (Gárate-Escamilla et al., 2019;Kramer et al., 2017;Meier & Leuschner, 2008;Vitasse et al., 2010). However, it is important to note that high plasticity is not inherently positive for local population persistence (cf. Kreyling et al., 2019), as it implies large differences in survival with changing environmental parameters, that is, limited survival under certain conditions.
The conditions of our sites along the large climatic gradient were highly diverse. Thus, we were able to study the responses of the earliest life stages in beech to various climatic conditions that are representative of the whole species distribution range. Moreover, a continent-wide masting event occurred in 2016 (Ascoli et al., 2017) and ensured high seed quality and quantity for all sites. Here, we were not able to investigate the phenology of germination. Studies by Arana et al. (2016) and Varsamis et al. (2020)

| Potential for regeneration across the distribution range from the south to the north
Germination was favoured by warmer and drier conditions and maximized even further south than the current contracting edge, but juvenile survival decreased strongly with heat and drought. Hence, germination and survival are driven by different, even opposite environmental factors. The lack of regeneration at the rear edge could exacerbate the range loss (Peñuelas et al., 2007).
Low survival was also found in parts of the distribution centre, as the year 2018 had been one of the driest and hottest years ever recorded in Switzerland and Germany (Buras et al., 2020;Schuldt et al., 2020). This indicates that marginality is not just induced geographically but rather ecologically (e.g. by climatic and soil characteristics; Bolte et al., 2016;Vilà-Cabrera et al., 2019). A vulnerability of beech to drought in the central part of its distribution was also found for adult trees during the summer drought in 2018 (Schuldt et al., 2020). This is supported by other studies that found that adult beech trees were sensitive to drought in the centre of the distribution range (Cavin & Jump, 2017;Muffler et al., 2020). Thus, parts of the distribution range centre of beech, for example our site BA in south-western Germany, are likely becoming ecologically marginal for adult trees and for natural regeneration, as the likelihood of drought events in summer is expected to increase in the near future at mid latitudes in Europe . In addition, species persistence at marginal sites likely also depends on the competition with better adapted species, such as the less drought sensitive evergreen tree species Quercus ilex L. at the southern distribution margin (Barbeta et al., 2019).
At cold marginal sites, the high plasticity in the early life-history traits may allow for increasing germination success with increasing temperatures and may thus facilitate natural regeneration even beyond the current cold margin in times of climate change. However, the potential for natural regeneration (survival) was three times higher at the northern than the north-eastern distribution margin in our study (Supporting Information Table S3). Water stress over summer and exceptionally moist conditions in autumn may explain the low survival at the north-eastern margin, as too moist soil conditions are stressful for beech growth (Leuschner & Ellenberg, 2017).
These findings call for subsequent regional studies with more study sites ranging from cold-dry to cold-wet sites, in order to better explain the potential for natural regeneration for the projected expansion of beech beyond the current cold distribution margin (Kramer et al., 2010;Saltre et al., 2015).

| Implications and outlook
Although no sign of local genetic adaptation for early life-history traits was detected, we cannot exclude genetic moderation of the observed early-stage plasticity in beech. Genetic variation in plastic responses is suggested to be an important predictor of species vulnerabilities to climate change (Kelly, 2019). Overall, the insights of our study into natural regeneration patterns of beech under changing climatic conditions are a valuable contribution to improve species distribution models. The potential for natural regeneration under warmer and drier climatic conditions has so far been understudied and requires more experimental investigations. Moreover, increasing temperature and occurrence of drought are not only threats for beech but also for other tree species worldwide (Allen et al., 2010;Anderegg et al., 2013;Brodribb et al., 2020).
Thus, we call for similar and long-term studies on other economically and ecologically important tree species that investigate the response of natural regeneration to changing climatic conditions. An ability for acclimatization or adaptation in the earliest life stages of a tree species could mitigate the effects of increasing heat and occurrence of drought by enabling local populations to persist under these conditions or by facilitating a range expansion.

ACK N OWLED G M ENTS
We