Introduction
The regeneration niche is an important determinant of plant species richness and is defined by the ecological requirements necessary to reproduce and colonize new spaces (Grubb, Reference Grubb1977). One of the most important aspects of the regeneration niche is seed germination, which is primarily controlled by two environmental factors: temperature and water potential (Dürr et al., Reference Dürr, Dickie, Yang and Pritchard2015). In species with a broad distribution, the requirements for seed germination may vary across areas in response to geographical and environmental gradients (Khera and Singh, Reference Khera and Singh2005). For example, a study held by Midmore et al. (Reference Midmore, McCartan, Jinks and Cahalan2015) shows that, in five different provenances of Betula pendula, a widespread European boreal species, seed germination was influenced by latitude. Also, the response to water stress may vary across populations, as demonstrated by Boydak et al. (Reference Boydak, Dirik, Tilki and Çalikoglu2003), for six collections of Pinus brutia originating from different climatic environments.
Populations that grow at the outer edges of the species distribution area may face extreme temperature or drought conditions. Consequently, they might display genetic adaptation to these environments (Levin, Reference Levin1993; Garcia-Ramos and Kirkpatrick, Reference Garcıa-Ramos and Kirkpatrick1997; Sexton et al., Reference Sexton, McIntyre, Angert and Rice2009; Abeli et al., Reference Abeli, Gentili, Mondoni, Orsenigo and Rossi2014; de Dato et al., Reference de Dato, Teani, Mattioni, Aravanopoulos, Avramido, Stojnic and Ducci2020) and may be different, even in the germination requirements, from populations living at the core of the species distribution area. In the present context of environmental and climatic changes, studying these adaptations may be a useful instrument for planning adequate conservation measures (Walck et al., Reference Walck, Hidayati, Dixon, Thompson and Poschlod2011).
In the Mediterranean region, mountainous areas may represent the southernmost suitable habitat for species with boreal distribution. For example, Conopodium majus, widespread in woodlands and oligotrophic meadows in northern Europe, presents a distribution restricted to mountain areas in the Iberian Peninsula (Tutin et al., Reference Tutin, Heywood, Burges, Valentine, Walters and Webb1964) and is differentiated in two subspecies, C. majus subsp. majus and C. majus subsp. marizianum, which represent the southernmost populations (Tutin et al., Reference Tutin, Heywood, Burges, Valentine, Walters and Webb1964). A similar pattern can be recognized for the widespread boreal tree B. pendula Roth. (Ashburner et al., Reference Ashburner, McAllister and Hague2013), which is present in the Southern Mediterranean mountains with two closely related species, B. fontqueri Rothm. in the Iberian peninsula and in the Moroccan Atlas and B. aetnensis Rafin in Sicily.
B. aetnensis is a narrow endemic species of Mount Etna (Sicily, Italy). The Etna birch is a tree morphologically similar to B. pendula, of which it is often considered as a subspecies (Tutin et al., Reference Tutin, Heywood, Burges, Valentine, Walters and Webb1964; Walters, Reference Walters1964; Meusel et al., Reference Meusel, Jäger and Weinert1965; Pignatti, Reference Pignatti1982). However, in this study, we will consider it as a species because of its history of long isolation and poor genetic exchange with other populations of B. pendula. The Etnean birch is a neoendemism and remained segregated on Mt. Etna since the end of the last glacial period. The genetic isolation, together with the selective pressure caused by the volcanic territory and the Mediterranean climate, has brought a new endemic taxon to the differentiation. Nowadays, there are only two populations of B. aetnensis on Mt. Etna located between the 1000- and 2000-m altitude areas (Strano, Reference Strano2010). B. fontqueri has a comparable genesis to B. aetnensis and, as is the latter, is considered by most authors to be a subspecies of B. pendula; but in this study, we will consider it as a species for the same reasons as for B. aetnensis. This Iberian birch has a scattered distribution on the mountains of Central (Sistema Central) and Southern (Sierra Nevada, Sierras de Cazorla, Segura and Las Villas) Spain (Martín et al., Reference Martín, Parra, Clemente-Muñoz and Hernandez-Bermejo2008) and is present with some relict populations in the north of Morocco (Peinado and Moreno, Reference Peinado and Moreno1989). It is included in the red list of plants of the IUCN, classified as endangered due to its reduced distribution and the low density of the existing populations (Martín et al., Reference Martín, Parra, Clemente-Muñoz and Hernandez-Bermejo2008). These two birches possess higher genetic variability than congeneric species from boreal latitudes (Palmé, Reference Palmé2003) and could be considered a genetic reservoir for the entire genus. Unfortunately, both species have a narrow distribution and are threatened by climate and land-use changes. In addition, B. aetnensis has been also subjected, in the last 30 years, to increased volcanic ash deposition, which could prevent its natural regeneration by seed. In this context, it is important to understand the regeneration ecology of these species in order to plan appropriate conservation measures. The adaptation of B. aetnensis and B. fontqueri to a Mediterranean climate could be reflected in their requirements for seed germination and seedling establishment. Currently, there is no published literature on the seed germination ecology of these Mediterranean birches and is fundamental to fill this gap in order to better understand their regeneration strategies. Given the presence of some relict populations of B. pendula in the Apennine range (Italy) (Plini and Tondi, Reference Plini and Tondi1989), which might present a certain degree of adaptation to Mediterranean climatic conditions, it is of interest to compare the response to temperature and water availability between these three species and within populations of B. pendula originating from the core and the margins of the European species’ distribution.
Temperature influences germination and the thermal germination niche can be described by the cardinal temperatures. In fact, there is a minimum (‘base temperature’, T b) and a maximum temperature (‘ceiling temperature’, T c) between which germination occurs and an optimal temperature (T o) at which it occurs at the highest speed (Dürr et al., Reference Dürr, Dickie, Yang and Pritchard2015). Many models of the response to temperature in crops (Garcia-Huidobro et al., Reference Garcia-Huidobro, Monteith and Squire1982; Covell et al., Reference Covell, Ellis, Roberts and Summerfield1986; Ellis et al., Reference Ellis, Covell, Roberts and Summerfield1986; Dürr et al., Reference Dürr, Dickie, Yang and Pritchard2015) and wild species (Pritchard and Manger, Reference Pritchard and Manger1990; Seal et al., Reference Seal, Daws, Flores, Ortega-Baes, Galíndez, León-Lobos, Sandoval, Ceroni Stuva, Ramírez Bullón, Dávila-Aranda, Ordoñez-Salanueva, Yáñez-Espinosa, Ulian, Amosso, Zubani, Torres Bilbao and Pritchard2017; Tudela-Isanta et al., Reference Tudela-Isanta, Ladouceur, Wijayasinghe, Pritchard and Mondoni2018) have been developed using the cardinal temperatures. T b, in particular, is a distinctive species trait (Dürr et al., Reference Dürr, Dickie, Yang and Pritchard2015). Nevertheless, other studies show that there is an intraspecific variation in T b among seed populations according to different environmental conditions during seed development (Daws et al., Reference Daws, Lydall, Chmielarz, Leprince, Matthews, Thanos and Pritchard2004).
The variation in soil water potential, which is the amount of pressure needed by the roots to absorb water from the soil, is another determinant of seed germination (Hegarty, Reference Hegarty1978), especially in environments that are characterized by a dry season, such as the Mediterranean region (Chamorro et al., Reference Chamorro, Luna, Ourcival, Kavgacı, Sirca, Mouillot and Moreno2017). Its effect can be simulated by using artificial solutions (Sharma, Reference Sharma1973; Falusi et al., Reference Falusi, Calamassi and Tocci1983; Chamorro et al., Reference Chamorro, Luna, Ourcival, Kavgacı, Sirca, Mouillot and Moreno2017) in order to calculate the base water potential (Ψb), that is, the condition below which no germination occurs and can be distinctive for a species or population (Kaufmann, Reference Kaufmann1969; Evans and Etherington, Reference Evans and Etherington1990; Choinski and Tuohy, Reference Choinski and Tuohy1991; Daws et al., Reference Daws, Lydall, Chmielarz, Leprince, Matthews, Thanos and Pritchard2004).
In this study, we estimated the base temperature and water potential thresholds for germination and compared them among the three species of Betula. Cold stratification decreases T b in B. pendula (Midmore et al., Reference Midmore, McCartan, Jinks and Cahalan2015) populations from Central and Northern Europe. Here, we compared its effect on the two Mediterranean Betula species and on marginal populations of silver birch. Since seedling establishment and not germination is the most critical stage for plant survival, we also calculated the thermal and water potential thresholds up to the stage of cotyledon emergence.
We wanted to test the following hypotheses:
(1) There is a difference in seed germination response to temperature and water potential among the Mediterranean species and B. pendula due to the long history of isolation of the former ones and to their adaptation to a drier and warmer climate.
(2) As a consequence of isolation and adaptation, the germination threshold results will be influenced by the local climate.
Material and methods
Seed collection
Seeds from eight populations were collected between September and November 2018 across Europe. We sampled five populations of B. pendula from Central Italy, the Balkans and Poland; one population of B. fontqueri from Central Spain and the two known populations of B. aetnensis from Mount Etna (Table 1). All of the populations were natural, except the one in Lazio (Italy), which, according to Plini and Tondi (Reference Plini and Tondi1989), was planted in the beginning of the 20th century. However, de Dato et al. (Reference de Dato, Teani, Mattioni, Aravanopoulos, Avramido, Stojnic and Ducci2020) consider this population to be natural on the basis of pollen fossil records. From here on, we will refer to these populations by using the following abbreviations: ‘CAM’ (Campania), ‘CRO’ (Croatia), ‘GAL’ (Galvarina), ‘LAZ’ (Lazio), ‘POL’ (Poland), ‘SAR’ (Sartorius), ‘SER’ (Serbia) and ‘SPA’ (Spain).
Geographical and bioclimatic data of seed collection sites
T max, maximum temperature of the hottest month; T min, minimum temperature of the coldest month.
From each collection site, the following bioclimatic parameters, which can represent a limiting factor to plant life, were obtained by interpolation using WorldClim Version2 (Fick and Hijmans, Reference Fick and Hijmans2017): precipitations of the driest month (‘Prec. Driest’), maximum temperature of the hottest month (‘T max’) and minimum temperature of the coldest month (‘T min’) (Table 1). The interpolations were calculated at a spatial resolution of 1 km2 and were based on meteorological data collected between 1970 and 2000.
Germination experiments
Temperature
Seed germination was tested at the following constant temperatures: 5, 10, 15, 20, 25 and 30°C with a photoperiod of 12 h of light and 12 h of dark. For each temperature, four replicates of 25 seeds each were sown on 1% agar in 9 mm Petri dishes. In addition, to test the effect of cold stratification on germination, following Cabiaux and Devillez (Reference Cabiaux and Devillez1977), we put four other replicates of 25 seeds each at a temperature of 5°C for 60 d. Light was excluded by wrapping the Petri dishes in a double layer of aluminium foil. After the stratification period, the seeds were moved to the test temperatures and scored every 12 h for the first week and less frequently afterwards. When no further germination or cotyledon emergence was observed for 28 d, the experiments were terminated.
Water potential
Osmotic potentials of −0.1, −0.2, −0.4, −0.6, −0.8, −1 and −1.5 MPa were obtained by dissolving polyethylene glycol (PEG 8000, VWR International srl, Milan, Italy) in deionized water. The quantity of PEG needed for each treatment was calculated according to Michel (Reference Michel1983) for a temperature of 20°C. In the control treatment, only deionized water was added to the germination medium. For each population, except LAZ and SER, for which there were not enough seeds, three replicates of 20 seeds each were sown in 6 mm Petri dishes. In each dish, two layers of Whatman N°1 filter paper (Merck Life Science S.r.l., Milan, Italy) were soaked with 2 ml of solution. Each dish was sealed with Parafilm (Bemis Company, Inc., Neenah, WI, USA) and placed in hermetic plastic bags to avoid changes in osmotic potential due to water evaporation. We refilled each Petri dish every 3 d with 1 ml of the correspondent PEG solution. The experiment lasted for 35 d and was conducted at the constant temperature of 20°C with 12 h of light and 12 h of dark. Germination was defined as 1 mm radicle emergence (henceforth referred to as ‘germination’). In both experiments, germinated seeds were left in the Petri dishes and scored again when the cotyledon opened (henceforth referred to as ‘cotyledon emergence’). At the end of each test, all of the ungerminated seeds were cut and visually inspected to assess their viability by checking the presence or absence of the embryo and its conditions. For each collection, at the end of all of the experiments, the dry weight of 25 seeds was measured, excluding SER, for which there were not enough seeds left.
Statistical analysis
The effects of the experimental treatments and of each treatment level were assessed on the final germination (‘FG’) percentage, the final cotyledon emergence (‘CE’) percentage and on the time to reach 30% of germination and cotyledon emergence (T 30). Germination response was compared between and within species. Time-to-event non-linear models were fitted to the germination and cotyledon emergence using the R package ‘drc’ (Ritz and Streibig, Reference Ritz and Streibig2005). From the fitted models, it was possible to calculate the time, expressed in days, at which each experimental treatment × population combination would have reached the 30 and 50% germination (T 30 and T 50) of the viable seeds. T 30 was used, instead of T 50, for modelling the germination response because its value was more realistic for the treatments with the lowest FG (%) and CE (%). Generalized linear models (GLMs) were fitted to FG (%) and CE (%) data using a binomial error distribution with a logit link. For T 30, we assumed a Gamma distribution based on the error structure, and the identity link was used after comparing the Akaike information criteria (AIC) of models fitted with other functions. Both for temperature and water potential experiments, separated analyses were run, first comparing the three species and then the populations within each species, with the exception of B. fontqueri, which was represented by a single population (SPA).
In the temperature experiment, we used three factors such as temperature, stratification treatment and species/population. In the water potential experiment, in contrast, there were only two factors including the osmotic potential and the species/population. Full models, including all factors and their interactions, were fitted for both the response variables [FG (%), CE (%) and T 30]. To assess the significance of main effects and interactions, an ANOVA with the Wald χ2 test was performed followed by a post hoc pairwise comparison test (Tukey's HSD) to assess the differences between species/populations and treatments.
To obtain the threshold values for temperature and water potential, the germination (or cotyledon emergence) rate was calculated as 1/T 30 and 1/T 50. For each treatment × population combination, a linear model was fitted with the germination speed as a response variable and the temperature or water potential as an independent variable. Once the line equation was obtained, its intersection with the x-axis, which represented the point at which the germination speed is zero, was calculated. The values obtained represent the base temperatures (T b30 and T b50) for unstratified or stratified seeds and the base water potentials (Ψb30 and Ψb50). In the calculation of the T b, the regression lines were fitted from the temperature with the higher germination speed for that treatment to the lower temperature tested (5°C). A two-way ANOVA followed by a post hoc test (Tukey) was performed on the T bs to investigate the effect of cold stratification, species/population and their interaction, while an one-way ANOVA was executed to test the differences in Ψbs among species/populations.
Comparison between environment and seed traits
The geographical (latitude and altitude) and bioclimatic variables were then compared with the seed traits. A data matrix was built, including latitude, altitude, Prec. Driest, T max, T min base temperatures and water potential for both germination and seedlings and seed weight. Data were checked for autocorrelation using the Pearson correlation coefficient in order to avoid variables with a strong autocorrelation. Due to the presence of outliers, the population of LAZ was excluded from this part of the analysis. The normal distribution of each variable was checked using the Shapiro–Wilk test, and the data that did not follow a normal distribution were transformed into their natural logarithm. For each seed trait, a linear model was fitted, including all of the environmental variables as predictors. From the full model, with a backward selection, the least significant variable was omitted, and a new model was fitted and compared with the previous one by a likelihood test. If the two models were not significantly different, the backwards selection proceeded until the model with the lowest AIC was retained. All statistical analyses were performed using R (R Core Team, 2019).
Results
Interspecific variation in germination requirements
FG (%) differed significantly (P < 0.001) among species with temperature and stratification. It was higher at 25 and 30°C and, generally, lower at 10 and 15°C, with the exception of the populations POL, SAR and SPA (Fig. 1). At 5°C, germination was slower but eventually reached high values (Fig. 1). CE was more sensitive to high temperatures: CE (%) peaked at 25°C and decreased at 30°C (Fig. 2). Cold stratification increased FG (%) in all species and treatments. Germination speed (T 30) was influenced by temperature and stratification (P < 0.001 and P = 0.005, respectively, Table 2) but did not differ among species. Stratified seeds germinated faster at all temperatures, except 5°C (Fig. 1). CE was slower at 30°C. It was not recorded for stratified seeds of SPA and CAM due to an experimental error. Base temperatures for FG and CE, calculated at 30 (T b30) and 50% (T b50) of viable seed germination, are significantly different among species (Fig. 3, Table 4). B. fontqueri had the lowest base temperatures for the germination of unstratified seeds, while B. aetnensis had the highest values (Table 4). T b30 and T b50 for the germination of stratified seeds were similar for the two Mediterranean birches and significantly lower in comparison to B. pendula. Stratification alone did not significantly influence the base temperatures, but it did when interacting with the ‘species’ factor (Fig. 3).
Effects of test temperatures and cold stratification on the final proportion (FG %) and the time to reach 30% of germination (T 30). White plots represent unstratified seeds, and grey plots represent stratified seeds. Due to the high range of variation of T 30, to make the figure more readable, its natural logarithm was used instead. FG, final germination.
Effects of test temperatures and cold stratification on the final proportion (CE %) and time to reach 30% of cotyledon emergence (T 30). White plots represent unstratified seeds, and grey plots represent stratified seeds. Due to the high range of variation of T 30, to make the figure more readable, its natural logarithm was used instead. CE, cotyledon emergence.
Boxplots showing the variation of the T b30 for the germination of unstratified and stratified seeds between and within species. At the bottom left of each subplot are displayed the results from a two-way factorial ANOVA: P, populations; S, species; T, treatment; P × T or S × T, interaction between population or species and treatments. The letters on the boxplots indicate differences between the samples (Tukey's HSD post hoc test). NS, not significant, *P < 0.05; **P < 0.01; ***P < 0.001.
Effects of test temperature, stratification, species/population and their interactions on final percentage (FP) and T 30 for germination and cotyledon emergence, based on logit-linked GLM with binomial distribution for FP and gamma distribution with identity link for T 30
Degrees of freedom (d.f.), Wald χ2 and P-values are reported. Temp, temperature; Pop, population; strat, stratification. In bold are the values where p < 0.05.
The reduction of water potential influenced FG (%), CE (%) and germination speed in all species, and no seed germinated below −1 MPa (Fig. 4). Drought tolerance differed among species (Fig. 5). The lowest values of Ψb30 for FG were obtained by B. fontqueri (−1.15 MPa ± 0.03 SE) and B. aetnensis (−1.05 MPa ± 0.01 SE) (Table 4). B. pendula had, on average, higher Ψb30 and Ψb50 with the exception of POL, whose Ψb30 was comparable with the Mediterranean species (Table 4). CE followed a similar pattern but was more affected than germination was by water stress.
Effects of water potential on the final proportions (FG% and CE%) and the time to reach 30% for germination and cotyledon emergence (T 30). Due to the high range of variation of T 30, to make the figure more readable, its natural logarithm was used instead. CE, cotyledon emergence; FG, final germination.
Boxplots showing the variation of the ψb (30) for germination between and within species. At the bottom left of each subplot are displayed the results from one-way factorial ANOVA: P, populations; S, species. The letters on the boxplots indicate differences between the samples (Tukey's HSD post hoc test). NS, not significant, **P < 0.01; ***P < 0.001.
Intraspecific variation in germination requirements
Betula aetnensis
The two existing populations of B. aetnensis presented different responses to temperature and drought. In fact, before stratification, GAL had low or zero FG (%) at medium temperatures (18.3% ± 5.3 SE at 15°C and 0% at 10°C), while SAR reached FG (%) >25% at all temperatures (Fig. 1). Germination speed was significantly (P < 0.001) influenced by temperature, stratification and their interaction but did not differ significantly among populations. CE closely followed germination in all treatments. T 30 decreased after stratification at all temperatures, except 5°C, which significantly slowed germination speed (Fig. 1; Table 4). Unstratified seeds had higher T b than other species (Table 4), but stratification decreased their values. B. aetnensis had high tolerance to water stress, but the two populations differed (P < 0.001) in their response for both FG and CE (Fig. 4), with SAR being the more drought-tolerant (Table 3). Moreover, SAR had the lowest Ψb30 and Ψb50 for CE among all of the populations considered (Table 4).
Effects of water potential and its interaction with species or populations on FP and T 30 for germination and cotyledon emergence based on logit-linked GLM with binomial distribution for FP and gamma distribution with identity link for T 30
Degrees of freedom (d.f.), Wald χ2 and P-values are reported. Pop, population; treat, treatment. In bold are the values where p < 0.05.
Base temperature (T b) for fresh and stratified seeds and base water potential (Ψb) calculated from germination and cotyledon opening speed at 30 and 50%
GAL, Galvarina; SAR, Sartorius; SPA, Spain: CAM, Campania; LAZ, Lazio; CRO, Croatia; SER, Serbia; POL, Poland.
Betula fontqueri
B. fontqueri showed high FG (%), without significant differences among temperatures and stratification treatments (Fig. 1). CE (%) was significantly lower at 30°C for unstratified seeds, and 0 for stratified seeds at 15°C, due to an experimental error. Drought tolerance was comparable to B. aetnensis, with FG and CE (%) decreasing significantly only below −0.8 MPa. Germination speed was positively influenced by temperature and stratification. T 30 for FG decreased with warmer temperatures. In contrast, the speed of CE was significantly lower at 30°C. Base temperatures for FG were not influenced by stratification, remaining below 4.50°C, a value comparable with POL, the northernmost population considered (Table 4). Instead, threshold values for stratified seedlings were significantly lower (P = 0.016 and 0.017 for T b30 and T b50), falling to 0.57°C (±0.97 SE) and 0.16°C (±1.31 SE) for T b30 and T b50, respectively (Table 4). These were the lowest base temperatures calculated for stratified CE among all the populations. Lower water potentials reduced the speed of FG and particularly of CE (Fig. 4). Ψbs of B. fontqueri is among the lowest (Table 4), demonstrating high drought tolerance in this Mediterranean birch.
Betula pendula
The five populations of B. pendula displayed heterogeneous responses to temperature, stratification and water potential treatments, which influenced significantly FG (%) and germination speed (Tables 2 and 3). The interaction between stratification and populations did not influence the speed of germination (Table 2). FG (%) was higher at 20, 25 and 30°C, lower at 10 and 15°C and the thermal germination window widened after stratification (Fig. 1). Only POL reached high FG (%) at 10 and 15°C, while LAZ had significantly lower germination across all treatments. Stratification increased FG (%), even though its effect was small for LAZ and CRO at 10°C (Fig. 1). Cold stratification improved germination speed across all populations and temperatures, with the exception of CRO and SER, whose germination was slower at 5°C after stratification, as in B. aetnensis (Fig. 1). CE followed FG closely, except at 30°C, where the final proportion was lower and T 30 was higher (Fig. 2). T b30 for germination in unstratified seeds ranged from 2.22°C (±0.54 SE) in CRO to 8.94°C (±0.45 SE) in LAZ. The values increased slightly for T b50. After stratification, base temperatures increased in CRO, decreased in LAZ and did not change significantly for the other populations (Fig. 3). T bs were more heterogeneous for CE, and a clear pattern could not be described (Table 4). These results were probably influenced by the low CE speed at high temperatures and should be interpreted carefully. CAM and CRO resulted in drought sensitivity, while POL germinated at 88.9% (±11.1) still at −0.6 MPa, a behaviour closer to B. aetnensis and B. fontqueri (Fig. 4). CRO had the highest threshold values for water potential for both germination stages, while the Ψbs of POL were comparable with the Mediterranean birches (Table 4).
Influence of the environment on germination thresholds
Prec. Driest did not follow a normal distribution and was transformed in its natural logarithm (‘lnPrec.Driest’). All threshold values estimated for 50% FG, and CE were highly correlated (Pearson's correlation coefficient >0.9) to the ones estimated for 30% and were not included in the modelling. The threshold values calculated for CE were correlated with the ones obtained for germination (Pearson's correlation coefficient >0.9 for T b of unstratified seeds and Ψb; Pearson's correlation coefficient >0.6 for T b of stratified seeds) and were, as well, excluded from further analysis. A strong negative correlation (Pearson's correlation coefficient = −0.78) was found between altitude and T max: in fact, despite the lower latitude, the more elevated locations were cooler in summer. A positive correlation (Pearson's correlation coefficient = 0.84) was found between base water potential and base temperature of stratified seeds. Both variables were left in the analysis to investigate the nature of this relationship.
T b30 of unstratified seeds was influenced by T max (P = 0.010): for every 1°C increase in T max, T b30 decreased by 1.45°C. The model that better described the influence of environment on the T b30 of stratified seeds also retained T max, but in this case, its effect was not significant (P = 0.296). However, their relationship was still negative, with a reduction of −0.49°C of T b30 for each degree of increase in T max. T b30 of stratified seeds increased significantly (P = 0.026) with the precipitation of the driest month. Ψb30 was described by the full model, in which altitude, T max, T min and lnPrec.Driest all concurred in explaining the dependent variable, but none of them had a significant effect (Table 5). Finally, seed weight was explained, but not significantly, by altitude (P = 0.091) and T max (P = 0.254).
Summary of GLMs of the influence of environmental variables on the threshold values for germination and seed weight
T b30, base temperature to 30% FG; Ψb30, base water potential to 30% FG; T max, maximum temperature of the hottest month; T min, minimum temperature of the coldest month; ln Prec. Driest, natural logarithm of the minimum precipitation of the driest month; SE, standard error; d.f., degrees of freedom. In bold are the values where p < 0.05.
Discussion
Interspecific differences in germination requirements
B. fontqueri and B. aetnensis, although derived from B. pendula, are species with a mountain Mediterranean distribution and have adapted to a seasonally drier and warmer climate. For all three species, germination was maximum at 30°C, and, for most of the populations of B. pendula and B. aetnensis, it was reduced or even inhibited at 15 and 10°C in the absence of cold stratification. This behaviour could prevent germination in autumn and postpone it to spring. In fact, even though under natural conditions some seeds may germinate in autumn, they are unlikely to survive winter conditions (Vanhatalo et al., Reference Vanhatalo, Leinonen, Rita and Nygren1996). Moist and chilling winter conditions represent the natural way of cold stratification and widen the range of temperatures suitable for germination in B. pendula (Midmore et al., Reference Midmore, McCartan, Jinks and Cahalan2015). Vanhatalo et al. (Reference Vanhatalo, Leinonen, Rita and Nygren1996) observed dormancy release in B. pendula at cold stratification temperatures (2.4 and 5.5°C) and its induction at 12.4°C. A positive effect of cold stratification was reported also for B. aetnensis by Strano and Poli Marchese (Reference Strano and Poli Marchese2011). They found that unstratified seeds reached the maximum FG at 30°C and the lowest at 15 and 35°C. However, after cold stratification, high germination occurred even at these temperatures. Instead, the single population tested of B. fontqueri showed high germination at all temperatures, without need for stratification. A germination temperature of 16°C in the presence of light was reported for this species by Blanca et al. (Reference Blanca, Cabezudo, Henández-Bermejo, Herrera, Molero Mesa, Muñoz and Valdés1999), as well as the absence of seed dormancy. However, the authors affirmed that the germination response varies among individuals.
Average base temperatures and water potential were lower for the two Mediterranean birches, compared to B. pendula. In Picciau et al. (Reference Picciau, Pritchard, Mattana and Bacchetta2019), Mediterranean mountain species had higher T b than lowland species to avoid germination when there is still a risk of freezing. Cold stratification decreases T b (Porceddu et al., Reference Porceddu, Mattana, Pritchard and Bacchetta2013; Dürr et al., Reference Dürr, Dickie, Yang and Pritchard2015; Midmore et al., Reference Midmore, McCartan, Jinks and Cahalan2015), but in our study, this only occurred for B. aetnensis and some populations of B. pendula. Cold stratification contributed to widening the germination window of B. aetnensis, lowering its T b but had no effect on B. fontqueri, which already possesses a low base temperature for the germination of unstratified seeds. However, stratification reduces, in this species, the T b for CE, producing cold-tolerant seedlings, capable of establishing when temperatures are still cold.
Low germination temperature and water potential in B. aetnensis and B. fontqueri represent an adaptive advantage for mountain environments with a seasonally dry and warm climate. For example, summer drought, on Mount Etna, is pronounced for at least 2 months per year (Bagnato et al., Reference Bagnato, La Piana, Mercurio, Merlino, Scarfò, Sciascia, Solano and Spampinato2014), and the anticipation of germination to early spring may increase the possibility, for seedlings, to reach a developmental stage that allows them to tolerate it. Heat tolerance, pronounced in seedlings of B. aetnensis, and a low Ψb could also increase their survival chances through summer. Even though T max on Etna's sites was the lowest (Table 1), the real value could be higher. In fact, our climatic data come from the interpolation (Fick and Hijmans, Reference Fick and Hijmans2017) of air temperature data. But summer temperatures of dark, volcanic soil can easily exceed 30°C even at high altitudes (Blandino, unpublished data). The ability to regenerate at low water potential increased the fitness of B. aetnensis and B. fontqueri in a Mediterranean precipitation pattern.
Intraspecific differences in germination requirements
Intraspecific variation in germination requirements and threshold values was observed in B. aetnensis and, especially, in B. pendula. This has been reported for B. fontqueri (Blanca et al., Reference Blanca, Cabezudo, Henández-Bermejo, Herrera, Molero Mesa, Muñoz and Valdés1999). It appears to be a combination of genetic variability and adaptation to the growing site.
Even though the general trend in B. pendula was of germination inhibition at low temperatures, the response, among the populations tested, was heterogeneous, with POL outstanding for high germination at all treatments, comparable with B. fontqueri. In contrast, LAZ, which grows in the hottest site out of the climatic envelope of the species, on a lowland sulphuric bog of volcanic origin in Central Italy, germinated poorly even at 20°C and presented the highest base temperatures. The divergent behaviour of this azonal population is mirrored by its genetic distinctiveness (de Dato et al., Reference de Dato, Teani, Mattioni, Aravanopoulos, Avramido, Stojnic and Ducci2020). The same authors found a high degree of genetic diversity among Mediterranean populations of silver birch, including B. aetnensis. A similar result was found by Martín et al. (Reference Martín, Parra, Clemente-Muñoz and Hernandez-Bermejo2008) for the extant Spanish populations of B. fontqueri.
The populations of Betula in this study had heterogeneous levels of dormancy, ranging from non-deep physiological dormancy to, apparently, no dormancy. According to Thompson and Ooi (Reference Thompson and Ooi2010), dormancy is a seed-specific characteristic, and its breakage requires slow changes in the seed itself. However, these physiological changes may not ensure germination, meaning that non-dormant seeds still cannot germinate. In fact, germination occurs only when the environmental conditions required by the seed are met. The intraspecific variation in germination requirements and threshold values observed in our populations does not allow for the definition of ‘dormant’ of any of the species studied. The variation of T b with stratification in B. pendula is heterogeneous and may be ascribable to local microclimate adaptations, which must be further investigated. The population subject to lower water stress, CRO, presented the higher Ψb, while POL, our northernmost population, had a Ψb comparable with the Mediterranean species, explainable with its low precipitations in the driest, winter months. In B. aetnensis, the inhibition of germination at 10 and 15°C was more pronounced for GAL than for SAR. These populations grow on different sides of Mount Etna, and genetic exchange is prevented by topography, leading to a certain degree of isolation. The two populations differed also in their Ψb, higher in GAL than in SAR, probably as adaptations to local edaphic conditions. In fact, the eastern side of Mount Etna, where SAR is located, is exposed to a considerable deposition of volcanic ashes, subject to the direction of the prevalent winds (Calabrese et al., Reference Calabrese, Aiuppa, Allard, Bagnato, Bellomo, Brusca, D'Alessandro and Parello2011). Here, the soil is mainly constituted of volcanic sand with a low content of organic matter. Water from precipitation is not retained by this kind of soil, creating the conditions for edaphic drought.
Relationship among seed and environmental traits
T b is a phylogenetically conserved trait, suggesting that adaptation to temperature may be a relatively slow process (Arène et al., Reference Arène, Affre, Doxa and Saatkamp2017). We found a negative relationship between the T b of unstratified seeds and the T max. However, higher summer temperatures were experienced mainly by the populations of B. pendula (Table 1), situated at lower altitudes. Nevertheless, cold stratification changed this relationship, broadening the germination window of B. aetnensis by lowering its T b.
The base temperature of stratified seeds was influenced by the intensity of annual drought, being lower where the minimum precipitation of the driest months was less.
Depending on the population, this relationship may have different interpretations. In fact, Mediterranean mountain populations do experience drought in summer months, after cold stratification occurs, while in the south of Poland, there is less precipitation in winter. If, in the first case, a low T b can anticipate germination before the onset of the dry season, the selective pressure on seeds from POL is direct, to avoid germination in the coldest months. However, the base temperature of POL did not change significantly with stratification and is high enough to avoid germination in the winter months.
None of the environmental parameters considered significantly influenced Ψb, although the best predictive model included all four variables. However, the differences in its value among the three species were significant. Base water potential is a more variable trait than T b and can be easily modified by adaptation to local conditions (Arène et al., Reference Arène, Affre, Doxa and Saatkamp2017). A stronger correlation with edaphic drought may have been demonstrated by using in situ collected data.
Conclusion
B. aetnensis and B. fontqueri survive as relict populations in glacial refugia at low latitudes and developed an adaptation to the mountain Mediterranean climate. Morphological, histological and ecophysiological adaptations to the Mediterranean climate were already described for the Etna birch (Pavari, Reference Pavari1956; Biondi and Baldoni, Reference Biondi and Baldoni1984; Leonardi et al., Reference Leonardi, Rapp, Failla and Komaromy1994). We provided evidence for an adaptation of the regeneration niche of these two species to a warmer and drier climate. Both Mediterranean birches are today endangered by land-use change, herbivory, fungal infection (Blanca et al., Reference Blanca, Cabezudo, Henández-Bermejo, Herrera, Molero Mesa, Muñoz and Valdés1999; Bagnato et al., Reference Bagnato, La Piana, Mercurio, Merlino, Scarfò, Sciascia, Solano and Spampinato2014; Morales-Molino et al., Reference Morales-Molino, Tinner, Perea, Carrión, Colombaroli, Valbuena-Carabaña, Zafra and Gil2019) and, in the case of B. aetnensis, increased deposition of volcanic ashes (Calabrese et al., Reference Calabrese, Aiuppa, Allard, Bagnato, Bellomo, Brusca, D'Alessandro and Parello2011).
Cotyledon emergence always followed germination at low temperatures, but seedlings do not tolerate excessive warmth and drought. The stage of seedling establishment is more critical than germination, especially in light of increasing average temperatures and drought. Protocols for ex situ cultivation and in situ interventions (e.g. translocations) should be developed to preserve all of the populations of the rare B. aetnensis and B. fontqueri and the highly diverse marginal populations of B. pendula.
In a scenario of climatic change, the Betula species adapted to the Mediterranean mountains appears to be greater prepared to face increasing summer temperatures and drought. Therefore, their conservation is important because they represent a genetic reservoir for the affine B. pendula and could prove to be a useful resource for the mitigation of climate change effects in the Northern European populations of silver birch.
Acknowledgements
We thank Adriano Stinca, Giovanni Salerno, Maria Belèn Luna Trenado, Salvatore Cambria and Zygmunt Kacki for the seed collection. This study was possible with the help of a scholarship from the Cutgana (University Center for the Protection and Management of Natural Environment and Agricultural Ecosystems) of the University of Catania.
