Intraspecific variation of Quercus ilex L. seed morphophysiological traits in Tunisia reveals a trade-off between seed germination and shoot emergence rates along a thermal gradient

Abstract

Key message

Quercus ilex populations from cold habitats display a large lag between seed germination and shoot emergence time, favouring avoidance of late frost events. Populations from mild habitats show the fastest seed germination and shoot emergence rates at moderate temperatures, enabling them to synchronize germination in the late winter-early spring period and a rapid seed-to-seedling transition, during the favourable rainy period.

Context

Quercus ilex is the most abundant and representative Mediterranean oak species. Identifying and describing intraspecific variation in seed traits is necessary to characterize the germination niche, and to elucidate drivers of species’ range.

Aims

In order to identify adaptations to local environments that may reflect ecological strategies for stress avoidance and seed survival, we tested under common and optimal conditions whether seed functional traits vary, in Quercus ilex subsp. rotundifolia Lam., along climatic gradients within its distribution range in Tunisia.

Methods

We have explored variations in seed morphological traits, desiccation sensitivity level, germination and shoot emergence rates under different controlled temperature conditions, among 15 populations of Q. ilex sampled throughout the Tunisian distribution of the species.

Results

Significant between-populations differences were observed for morphological seed traits but no relationships could be established with the climate of the sampling sites. In contrast, key physiological traits varied significantly with elevation and temperature. Specifically, mild temperatures in lowland regions were associated with higher seed moisture content, fast germination and shoot emergence rates at moderate temperatures (13 °C) for germination. Seeds of Q. ilex populations from cold sites displayed the fastest germination rates at low temperatures (5 °C) as well as the greatest lag between seed germination and shoot emergence time.

Conclusion

Intraspecific variation in seed physiological traits is significantly associated with local climate. This functional diversity should be considered when evaluating germplasm and predicting suitability for reforestation and assisted migration programs.

1 Introduction

Quercus ilex L. (holm oak) is the most abundant and representative Quercus tree species in the Mediterranean forests and is known for its widest ecological amplitude among Mediterranean oaks, from semi-arid to per-humid bioclimates, and from warm to cold conditions depending on its elevation (Barbero et al. 1992). This sclerophyllous evergreen tree species is found from Turkey to Spain on the European side of the Mediterranean Sea, and from Morocco to Tunisia on the African side, and has also colonized most of the Mediterranean islands, with large morphological and genetic variations, as well as East–West differentiation (Lumaret et al. 2002; López De Heredia et al. 2007).

Q. ilex distribution is thought to be delimited by aridity in the southern area (Terradas and Savé 1992) and by low winter temperatures and freezing stress in northern and high-altitude areas (Nardini et al. 2000). Intraspecific variations exhibited by Q. ilex across its circum-Mediterranean distribution have been largely documented for morphological, structural and functional traits (Barbéro et al. 1992; Gratani et al. 2003; Lumaret et al. 2002; Michaud et al. 1995; Peguero-Pina et al. 2014; Sanchez et Retuerto 2007), including those associated with tolerance to drought and cold stresses (Gimeno et al. 2009; Peguero-Pina et al. 2014). Q. ilex shows two main morphological types, associated with genetic differentiation, which are considered as two subspecies (Lumaret et al. 2002; Michaud et al. 1995). Quercus ilex subsp. rotundifolia Lam. (Q. ilex subsp. ballota being a heterotypic synonym, Ferrer-Galego and Sáez 2019; Le Floc’h et al. 2010) displays the ‘rotundifolia’ morphotype, characterized by small and rounded thick leaves with high vein density, and is the exclusive morphotype in Tunisia, Algeria and Morocco, and the dominant one in the Iberian Peninsula (Barbéro et al. 1992; Ferrer-Galego and Sáez 2019; Peguero-Pina et al 2014). The ‘ilex’ morph (Q. ilex subsp. ilex L.), distinguished by elongated and large leaves with low vein density, is present in mild coastal areas from Greece to France, while an ‘intermediate’ morphotype for trees displaying intermediate characteristics between the two morphs and dominating coastal areas of eastern Spain and south-eastern France (Lumaret et al. 2002).

In the context of global climate change, circulation models predict periods of prolonged drought in the Mediterranean basin and more frequent extreme events, such as heat waves and late-winter frosts, which have already been observed (Giorgi and Lionello 2008; Ruffault et al. 2014; Tramblay et al. 2020). Mediterranean oaks are usually seen to share the capacity to cope with water deficits through specialized adaptive features such as sclerophylly, restricted leaf area, and thick cuticles. However, climate change could have dramatic consequences for the survival and distribution of Mediterranean oak species, especially on the southern edge of the Mediterranean region (Matesanz and Valladares 2014; Ruiz-Labourdette et al. 2012). Indeed, both migration and genetic adaptation are relatively slow processes in species that have a long generation time such as oaks (Bussotti et al. 2014; Garcia and Zamora 2003). Many research studies have been conducted to determine the capacity of adult oak trees to survive summer drought and to understand how vegetative functional traits contribute to the ecological ranges of the different Mediterranean oak species (Acherar and Rambal 1992; Adams et al. 2017; Castagneri et al. 2017; Cavender-Bares et al. 2005; Limousin et al. 2012; Lloret et al. 2016; Quero et al. 2011; Niinemets and Keenan 2014; Ramirez-Valiente et al., 2022; Solé-Medina et al. 2022). However, less attention has been paid to Mediterranean oak fecundity and sexual reproduction (Le Roncé et al. 2021), as well as seed traits and seedling ecology, especially in North Africa (Gomez-Aparicio et al. 2008; Gonzales-Rodrigues et al. 2011; Joët et al. 2016; Urbieta et al. 2008). Yet seed germination is a vital process in a plant life cycle affecting seedling establishment and survival, which can subsequently determine species’ distribution and population persistence (Carta et al. 2022; Cochrane et al. 2015a; Donohue et al. 2010; Rosbakh and Poschlod 2015). Sensitivity to abiotic stresses makes regeneration, which is the transition from seed to seedling, a serious bottleneck in population recruitment and the most critical stage for survival in a Mediterranean-type community (Cochrane 2016; Lloret et al. 2004; Perez-Ramos et al. 2013). For instance, changes in precipitation have been noticed to have dramatic effects on Holm oak recruitment in woodlands in southern France (Perez-Ramos et al. 2013). Considering that the seeds of Mediterranean oaks are recalcitrant, i.e. they are short-lived and desiccation-sensitive (Amimi et al. 2020; Ganatsas and Tsakaldimi 2013; Joët et al. 2013), the first type of stress they may encounter during winter is a prolonged dry spell between shedding in autumn and the return of favourable conditions for germination in spring. Although summer drought is widely considered to be the main limiting factor for plant survival in Mediterranean ecosystems, winter stresses are crucial when interpreting species distribution (Larcher 2000). Desiccation was shown to be the major cause of in situ mortality of Q. ilex seeds at the end of winter and a long period of drought can lead to the loss of an entire annual seed cohort (Joët et al. 2013, 2016). The need to remain fully hydrated during winter also exposes recalcitrant seeds to a second type of hazard which is freezing.

Recently, key seed traits that govern seed persistence and germination niche breadth, and that may influence the geographical ranges and ecological strategies, have been shown to diverge significantly among four Mediterranean oak species co-occurring in Tunisia (Amimi et al. 2020). Indeed, when assessed on 4–5 different representative populations for each species, the seeds of Q. ilex and Quercus canariensis Willd., which occur at relatively high elevations where frost events are frequent, displayed the highest freezing tolerance while acorns of Quercus coccifera L., which is frequent in warm and arid environments, showed the highest germination rate and synchrony (Amimi et al. 2020). However, the extent of intraspecific variation for key germination traits remains to be determined for each species across a large number of populations and is the focus of the present study for Q. ilex.

Many studies have revealed that among-population variation in germination response to environmental conditions resulted in different responses to climate change within species, which could mitigate the species’ vulnerability to changing climate and provide opportunities for species adaptation and conservation (Chamorro et al. 2017, 2018; Cochrane et al. 2015a). The sensitivity of germination to climate variability depends on the species’ phenotypic plasticity, local adaption, and geographic distribution (Cochrane et al. 2015b; Nicotra et al. 2010). However, to date, only a few studies have examined intraspecific variation in seed traits in Mediterranean oaks, and these studies have remained primarily focused on seed size. This latter is a key functional trait determining the reserves of carbohydrates and nutrients seedlings need during germination, which contributes to the ecological strategy and distribution of species (Jimenez et al. 2016; Moles and Westoby 2004). In Mediterranean Quercus species, the significant interspecific and intraspecific variation in seed size may facilitate their establishment in a heterogeneous environment (Quero et al. 2007). In a comparative study of Quercus suber L. populations, positive correlations between seed mass and xerothermic indices have been observed (Ramirez-Valiente et al. 2009). Variability, within species, in acorn morphological and chemical traits, including starch, lipids and phenolic compounds, has also been reported for Q. ilex (Caliskan 2014; López-Hidalgo et al. 2021; Valero Galvan et al. 2012). Determining within-species variation in seed physiological and germination traits is essential to understand how trade-offs constrain adaptation to contrasted environmental conditions, and to elucidate the drivers of species’ ranges (Garcia-Nogales et al. 2016). Exploring how seed germination traits from different close local provenances respond to climate proxies will be helpful to understand the strategies of Q. ilex to adapt along local xerothermic gradients.

Here we test whether the functional traits of seeds vary along a climatic gradient within its range of distribution in Tunisia. Holm oak occurs mainly in the two orographic regions in Northern Tunisia, High-Tell and the Tunisian Ridge, and their surroundings. Influenced by the proximity of the Mediterranean Sea, the High Tell region is the wettest one, while the Tunisian Ridge is characterized by a drier climate, and low winter temperatures (Nabli 1989, 1995). To improve our understanding of the effects of climate and topographic factors on the major seed traits, the aims of the present study were to (1) inventory the distribution of the main Q. ilex populations in Tunisia and document the climate factors associated with their habitats; (2) measure intraspecific variations in the main seed morphophysiological traits, including seed germination and response to desiccation and assess their intraspecific variations; and (3) explore the relationships between observed variations in seed traits and climate factors in order to identify adaptations to local environments that may determine seed survival and stress avoidance in holm oak.

2 Material and methods

2.1 Study area, and climatic data

This study covered the entire Tunisian distribution area of Q. ilex (Fig. 1; Amimi et al. 2022). The inventory of the distribution of Q. ilex in Tunisia was carried out over a period of several months in 2020 using systematic transect searches on forest roads and pastoral paths, in and around the previously identified areas of occurrence in woodlands of the two Tunisian orographic regions, High-Tell and the Tunisian Ridge, and their surroundings. On the highest peaks (Chaâmbi, Kesra, Serj, Semmama) holm oak forms pure stands. Below 1000 m, under meso-Mediterranean bioclimates, holm oak mixes with Pinus halepensis Mill. (Aleppo pine; El Hamrouni et al. 2020). Species is also present in the thermo-Mediterranean bioclimates of the Tunisian High Tell, north of Ghardimaou (Ouled Ali forest) and north of Téboursouk (Nabli 1989). The sampling area, from 501 to 1078 m in elevation, ranged in latitude from 35° 42′ 36″ N to 36° 25′ 48″ N and in longitude from 8° 23′ 38.4″ E to 10° 5′ 34.8″ E. The geographic information about each sampling site was collected by a GPS navigator upon sampling. Information on habitat was inferred from the environmental layers of 19 bioclimatic variables (bio1-19), integrated over the years 1970–2000, obtained from the Worldclim version 2 database (Fick and Hijmans 2017). The resolution used for all climatic layers was 30 arc-seconds (1 km²).

Fig. 1

Distribution map of Quercus ilex populations in Tunisia. The current Q. ilex distribution (dark green areas) was estimated in this study, while the past distribution area (light green surfaces) is based on the last national forest inventory of 1990 (Nabli 1989). All populations of Q. ilex are distributed in the two mountainous regions of the High Tell and the Tunisian Ridge. Locations of the 15 woodlands sampled are indicated by red dots

2.2 Biological material

This study was carried out on mature acorns collected in November 2020 from 15 holm oak woodlands (Q. ilex). Twenty trees were randomly selected from each population, the sampled trees being located at a distance of at least 20 m from each other. The acorns were collected by shaking the tree when the acorns reached full maturity, according to the indexes of maturity described by Bonner and Vozzo (1987). In each site, about 2000 acorns were collected and pooled (100 acorns per tree), immediately enclosed in hermetically sealed plastic bags, and stored in the dark at ambient temperature for a maximum of 2 days before being transported to INRGREF (Tunis, Tunisia). In the laboratory, the acorns were first immersed in water to identify insect-damaged or infected floating acorns. After additional visual screening for abnormal and defective acorns, sound acorns were stored in hermetically sealed plastic boxes in the dark at 4 °C, as described previously (Bonner and Vozzo 1987), for a short period of a maximum of 10 days until physiological characterization was initiated.

2.3 Acorn and seed morphological traits

The seed and pericarp moisture contents, expressed on a fresh weight basis, and their dry mass were determined gravimetrically after oven-drying for 17 h at 103 °C, using 50 randomly selected acorns per sampled site. The ratio of pericarp mass to acorn mass (pericarp/acorn ratio) was also calculated from these measurements. Acorn length and maximal width as well as pericarp thickness were measured on 100 randomly selected acorns per sampled site, using a digital calliper.

2.4 Germination and shoot emergence

Germination kinetics were assessed in the dark in climate chambers fixed at five different constant temperatures (5, 9, 13, 17 and 21 °C) to determine the germination base temperature (Pritchard and Manger 1990). For each condition and location sampled, the pericarp was removed and three batches of 30 seeds were placed in plastic boxes (170 × 105 × 60 mm), containing a layer of cotton moistened with distilled water. Germination and shoot emergence were surveyed daily for a period of 20 weeks and recorded when the radicle had grown at least 5 mm and curved and shoot length reached 5 mm, respectively. Time needed to achieve 50% of maximum germination (t₅₀) and the specific parameter Ga (germination asynchrony, i.e. the slope of the tangent line at the point of inflection) were determined using the following germination function and least-squares regression as computed by the quasi-Newton method:

$$G=\frac{Gmax}{1+\mathrm{exp}(Ga \left(t-t{50}\right))}$$

where G is the percentage of germination at time t, and Gmax is the maximum germination. The estimation of t₅₀ further enabled the calculation of the germination rate, GR, defined as 1/t₅₀. Time needed to achieve 50% of maximum shoot emergence (t₅₀SE) and the specific parameter GaSE shoot emergence asynchrony were determined using a similar function where SE is the percentage of shoot emergence at time t, and SEmax is the maximum shoot emergence.

2.5 Desiccation sensitivity

To measure seed desiccation sensitivity, 10–12 batches of 50 mature acorns per population were spread on benches in a ventilated room (15 °C, 35–40% RH) and air-dried for 30 days. By using one batch every 3 days, seed water content was assessed gravimetrically on 10 seeds while seed viability was assessed on 40 others following the germination criterion, i.e. protrusion of the radicle and geotropic growth, after 2 weeks of culture at 21 °C in the dark. Desiccation sensitivity was quantified using the quantal response model (Dussert et al. 1999):

$$\mathrm{V}=\frac{\mathrm{Vi}}{1+\mathrm{exp}(\mathrm{b}\left(\mathrm{WC}-\mathrm{WC}50\right))}$$

where V is seed viability, Vi is the initial viability, WC₅₀ is the water content at which half of the initial viability is lost, and b is seed lot-specific parameter.

2.6 Statistical analysis

ANOVA, PCA and linear and non-linear regressions were carried out using Statistica software (Statsoft, USA). Differences in seed traits were tested using one-way ANOVA with a fixed effect and the Bonferroni post hoc test. Residuals were checked for normality and homoscedasticity using log-transformed data. Correlations between variables were analyzed by linear regression using Pearson’s correlation coefficient. A significance threshold of P = 0.05 was retained for both ANOVA and linear regression. Principal component analysis (PCA) was performed to analyse the covariations between climatic factors and morpho-physiological traits of seeds from the different surveyed sites, latitude, longitude and elevation being supplementary variables. Hierarchical Clustering Analysis (HCA, Ward’s grouping method, Euclidian distance) was then applied to factorial scores obtained from PCA analysis on climate variables in order to pinpoint clusters and similarities in bioclimatic envelopes associated with the different Q. ilex populations. Non-linear regressions (desiccation sensitivity and germination time) were performed using least-squares regression computed by the quasi-Newton method (Dussert et al. 1999).

3 Results

3.1 Bioclimatic conditions associated with Quercus ilex populations in Tunisia

The inventory of Q. ilex in Tunisia identified 15 major areas where holm oak is present, including 4 sites as dominant species, and 11 others in association with either Aleppo pine or other Mediterranean oaks, Q. coccifera and Q. suber (Fig. 1; Appendix Table 1; Amimi et al. 2022). The structure of climate variability among habitats was analyzed by PCA, and the first two principal components explained 81% of the overall variance (Fig. 2a). The first Principal Component (PC1) was highly correlated with average temperature Tmean (R = 0.981) while negatively correlated with elevation. PC1 represented a thermal gradient associated with elevation, the average temperatures varying significantly from 13.8 °C to 16.8 °C between sites. The second component (PC2) was positively correlated with annual rainfall (R = 0. 879) and highlighted a gradient of rainfall seasonality as it separated sites with the rainiest winters and annual rainfall from sites with the broadest annual temperature range and maximal temperatures in summer. The 15 sampled locations displayed high variability in cumulative annual rainfall, ranging from 407 to 816 mm (Appendix Table 1). Each sampling population can be characterized by its factorial score co-ordinates on the first two PC, highlighting differences for bioclimatic variables associated with the different Tunisian holm oak populations (Fig. 2b). Hierarchical clustering analysis of such co-ordinates further revealed a discontinuous structure with two main clusters, I and II, which differentiate warm and cold sites, respectively (Fig. 2c). Clusters were further split into two subclusters, subclusters II-a and II-b corresponding to elevated sites associated with wet and dry climates, respectively. This analysis therefore defined a climatic typology of holm oak woodlands, with four clusters of sites that are differentiated along the thermal and rainfall gradients.

Fig. 2

Multivariate analysis of the climatic characteristics of the 15 Quercus ilex sampling sites. a Principal Component Analysis (PCA) correlations (factor loadings) of climatic variables with the first two principal components (PC). Longitude, latitude and elevation, in grey, were treated as supplementary variables. b PC1–PC2 score plot of the 15 populations studied. Each population sampled was characterized by its factorial score co-ordinates on the first two PCs. c Hierarchical Ascendant Classification (HCA) of holm oak sites according to their climatic variables. HCA analysis revealed 4 differentiated clusters, which colour code have been reported on the upper PC1-PC2 score plot

3.2 Intraspecific variation in seed morphophysiological traits

All measured morphological traits, i.e. seed size, seed mass, seed water content, pericarp thickness, and the pericarp mass to acorn mass ratio (PAR), showed large individual variation within a population, but also statistically significant variation among sampled populations (P < 0.05; Appendix Table 2). The largest variation between sites was observed for seed dry mass that ranged from 1.55 to 4.45 g. The seed water content at shedding also displayed large variations and ranged from 33.82 to 41.98% (on a fresh mass basis).

Seed germination was always very high, since it ranged between 95 and 100%, independent of the population and the germination temperature within the range 5–21 °C (Fig. 3), demonstrating the high vigour of all seed lots. Tunisian holm oak displays very low temperature requirement for germination, since all populations tested were able to complete germination at 5 °C. By contrast, large intraspecific variation was observed for the germination time at such low temperatures. Indeed, for the four populations with the fastest seed germination, germination was completed in about 30 days (800 h), while 70 days (1700 h) was required for the two populations with the slowest germination (Fig. 3).

Fig. 3

Seed germination time courses at 5 °C, 13 °C and 17 °C (a) and shoot emergence time course at 13 °C (b) for the 15 Tunisian Quercus ilex populations sampled

Between the two extreme populations, the difference in germination time, as quantified by t₅₀, i.e. the time to reach 50% of maximum germination, is 54 days (1316 h). A significant correlation (P < 0.05) was found between t₅₀ and germination asynchrony Ga at 13 °C and 17 °C (R = 0.76 and 0.59 at 13 °C and 17 °C, respectively), indicating that populations with the longest germination time were also those with the highest seed-to-seed variability in germination time (Appendix Table 3). Time for germination at 9 °C and 17 °C was much shorter than that at 5 °C, with smaller variations in germination time between populations (Fig. 3). Furthermore, seed germination traits (t₅₀ and Ga) determined at temperatures higher than 5 °C (9 °C, 13 °C, 17 °C and 21 °C) displayed some positive correlations between them (Appendix Table 3).

The germination rate (1/t₅₀) increased linearly between 5 °C and 21 °C in all populations (Fig. 4), enabling the determination of the base temperature T_b, i.e. the minimum temperature for germination, in the different populations. Q. ilex showed a relatively low intraspecific variation in germination rate response to temperature, with 5.08 ± 0.91 °C average T_b, except for one population (#8) which showed a T_b of 2.5 °C. This narrow window of variation among populations for T_b nevertheless displayed significant correlations with seed mass (R = 0.57) and germination time t₅₀ at 21 °C (R =  − 0.69) (Appendix Table 3). In addition to germination time at low temperature, a large intraspecific variation was also observed for the time of emergence of shoot and primary leaves after germination at a favourable temperature of 13 °C (Fig. 3b). Indeed, the SEt₅₀, the time to reach 50% of maximum shoot emergence ranged from 26 to 57 d. This trait was positively correlated with germination time at the same temperature (R = 0.60; Appendix Table 3). Finally, large between-population variations were observed for WC₅₀, the water content at which half of the initial viability is lost, that varied from 26.4% to 38.4% (Fig. 5). While not significantly correlated to seed mass (R = -0.39; P = 0.19), WC₅₀ was positively correlated with initial moisture content of mature seeds at shedding (R = 0.68; P = 0.01) (Appendix Table 3).

Fig. 4

Effect of germination temperature on seed germination rate (1/t₅₀) in the 15 Tunisian Quercus ilex populations studied and determination of seed base temperature, Tb (ordinate) for seed germination

Fig. 5

Survival of seeds of the different Quercus ilex populations studied after desiccation to various water contents. Viability was estimated as the percentage of germination

3.3 Identification of key climate-seed trait associations

PCA performed on the complete dataset, including climatic variables and seed morphophysiological traits, led to similar results to those obtained with climatic variables alone (Figs. 2a and 6). Variance of climatic variables was mainly explained by the two first principal components PC1′ and PC2′, which accounted for 47% of the overall variance. Like PC1, the first factor (PC1′) was highly correlated with average temperature Tmean (R = 0.971) and represented a thermal gradient negatively associated with elevation, while the second factor (PC2′), as observed for PC2, was positively correlated with rainfall (R = 0. 774) and latitude. None of the morphological traits, including seed mass and pericarp thickness, was collinear with bioclimatic variables and correlated with either PC1′ or PC2′ (Fig. 6). As for seed physiological traits, it is worth noting that only a few of them were correlated with PC1’, and strikingly, none of them was correlated with PC2′. PC1′ was positively correlated with seed water content (R = 0.664), desiccation sensitivity (0.734), and germination time at low temperatures (t₅₀ at 5 °C; R = 0.631), but negatively correlated with the time for shoot emergence when germinated at an optimal temperature of 13 °C (R =  − 0.736). This result suggests an adaptive trade-off between the capacity of rapid germination at low temperatures, which is significantly higher for all populations distributed in cold climates, and seedling vigour, i.e. rapid stem emergence at more favourable temperatures (13 °C) for populations distributed in areas with the mildest climates. One may also note that most of evidenced seed trait-climate associations revealed by PCA were also significant (P < 0.05) when tested individually using Pearson correlation coefficients (Appendix Table 4). Indeed, germination time at low temperatures (t₅₀ at 5 °C) and seed water content were significantly correlated with mean annual temperature (R = 0.60 and 0.57, respectively) while WC₅₀ was positively correlated with mean annual temperature (R = 0.65) and inversely correlated with annual precipitation (R =  − 0.56) (Appendix Table 4). Similarly, germination asynchrony at 5 °C was positively correlated with temperature (R = 58), while shoot emergence at 13 °C was negatively correlated (R =  − 0.68). Finally, none of the major morphological traits, including acorn size, pericarp thickness, or seed mass, showed variation among populations that correlated significantly with climate variables (Appendix Table 4).

Fig. 6

Principal Component Analysis (PCA) correlations (factor loadings) of climatic variables (black circles) and seed morphophysiological traits (pink circles, red circles for correlations higher than 0.5) with the first two principal components (PC1′ and PC2′). Longitude, latitude and elevation, in grey, were treated as supplementary variables

4 Discussion

In this study, we have explored variations for seed functional traits among populations of Q. ilex sampled over the entire Tunisian distribution of the species. While all measured morphological traits, including acorn size, seed mass and pericarp thickness, showed within and between-site variations, no significant relationship could be established between variation in these traits and major climatic factors along environmental gradients. When studied at large spatial scales, significant relationships have been detected between Q. ilex seed mass and geographical factors, such as latitude or altitude (Garcia-Nogales et al. 2016; Valero Galván et al. 2012), and similar trends have been observed in Q. suber along a xerothermic gradient (Ramírez -Valiente et al. 2009; Matías et al. 2018). Our study, conducted on a smaller scale, on two mountain ranges in Tunisia, did not reveal such relationships between seed mass and climate or geographical variables. In this context, it is worth noting that the seed mass of Q. ilex is a highly variable trait whose variations may be associated with different local factors such as selective pressure from seed predators, intensity of summer drought during seed development, or age of mother trees (Alonso-Crespo et al. 2020; Celebias and Bogdziewicz 2022; Le Roncé et al. 2021). Since variation in Q. ilex seed mass affects seedling fitness as well as the probability of attack by post-dispersal seed predators (Gómez et al. 2004), it has been suggested that large intraspecific variation in this trait is the result of a bet-hedging strategy facilitating seed survival and seedling establishment in a heterogeneous environment (Quero et al. 2007). Furthermore, we did not detect any significant correlation between seed mass and germination traits associated with the different Q. ilex populations. The absence of relationships between key morphological traits, including seed mass and pericarp thickness, and germination traits has also been observed in many other oak species (Xia et al. 2015).

By contrast, several key seed physiological traits, including seed water content at maturity and desiccation sensitivity, as well as germination time at 5 °C, displayed variations significantly correlated with mean annual temperature at the site of sampling. These results obtained on Q. ilex seeds corroborate several recent studies emphasizing the primary role of local climate in shaping intraspecific variation in temperate and Mediterranean plant species with respect to key seed germination traits (Carta et al. 2022; Chamorro et al. 2018; Cochrane et al. 2015a). At this stage, we cannot separate genetic and environmental aspects (involving either local genetic adaptation or phenotypic plasticity associated with maternal or environmental effects), but such seed trait variations highlight the ecological strategies adopted for adaptation to different climatic constraints. Q. ilex populations from the coldest locations, i.e. high elevation sites of the Tunisian ridge, displayed the fastest germination rates at low temperatures. By contrast, populations distributed within the warmest sites, i.e. low elevation sites surrounding the High-Tell and the Tunisian ridge, displayed the fastest seed germination and shoot emergence rates at mild temperatures. The observed opposite variation of these traits along a thermal gradient suggests the existence of population-specific germination niches and ecological strategies adapted to local Tunisian climates. The rapid germination rate at mild temperatures, with synchronized shoot emergence, enables seeds from mid- and low-altitude Q. ilex populations to align germination and seed-to-seedling transition schedules with early spring, a favourable rainy period with limited risk of a late frost. Seeds of Q. ilex populations from elevated sites show fast germination in the cold, enabling also germination in late winter-early spring, in mountainous climatic conditions. However, those populations display higher uncoupling between seed germination and shoot emergence time, favouring avoidance for the plantlets of the freezing stress that is associated with frequent late frost events. Early development of a large root system is characteristic of all oak species (Johnson et al. 2009), and a temperature-dependent lag between radicle development and shoot emergence, or epicotyl dormancy, has been described for different temperate European and Asian oaks (McCartan et al. 2015; Sun et al. 2021; Xia et al. 2022). For instance, for the European temperate oak Q. robur under natural conditions, winter ambient temperatures are generally sufficient for seed germination, but not favourable for shoot emergence (McCartan et al. 2015). Our data also strongly suggest a population-specific variation of this key seed trait along a temperature gradient. The antagonism between populations for the ability to germinate rapidly at cold or optimal temperatures suggests physiological or molecular trade-offs for the acquisition of these traits. Specifically, we showed that mild temperatures of lowland regions are associated with higher seed moisture content and fastest germination and shoot emergence rates at optimal temperatures for germination. Being dispersed with high relative water content, these populations may have higher metabolic activity and vigour (Xia et al. 2022).

For recalcitrant seeds, air temperatures during development can influence the seed maturity state and consequently causes subtle variations in the relative level of seed desiccation sensitivity (Daws et al. 2005). In the case of Q. ilex in Tunisia, the variation in seed desiccation sensitivity observed in different populations correlated well with the initial seed water content at the time of dispersal and surprisingly not with xerothermic gradients. Such absence of relationships between desiccation sensitivity and climate proxy has also been observed for several Asian oaks (Xia et al. 2015, 2022), and deserves further study, including measurement of tissue water potential.

5 Conclusion

Our study provides a comprehensive picture of the seed functional diversity of Q. ilex in Tunisia. We highlighted the contribution of temperature regimes in shaping seed germination traits differentiation among sites. Although the genetic component of seed trait variation associated with the provenance sites remains to be determined, the evidence for large functional diversity associated with local climate supports the existence of population-specific germination niches and ecological strategies and underlines the importance of conserving the genetic resources that peripheral populations harbour at the edges of distribution (Fady et al. 2016). As previously highlighted for Q. robur in Europe (McCartan et al. 2015), the knowledge gained in this study will be of tremendous importance to predict the impact of reforestation and assisted migration programs on the synchronization of germination and shoot emergence of holm oak, and therefore in guiding the selection of populations to be chosen for different target areas.

Appendix

Table 1 Locations of the 15 source populations for Quercus ilex in Northern Tunisia, and their main associated climate characteristics. Lat, latitude (decimal); Long, longitude (decimal); Alt, altitude (m), MAT, mean annual temperature (°C); MAP, mean annual precipitation (mm); ET0, mean annual potential evapotranspiration (mm)

Table 2 Variation in morphological traits of seeds and acorns among the 15 Tunisian Holm oak populations studied. Differences between populations were tested using one-way ANOVA and post hoc Bonferroni test form comparison of means. Two means displaying no letters in common are significantly different (P < 0.05)

Table 3 Cross-correlation matrix between seed traits. R values in bold indicate adjusted P values  < 0.05. Ga, germination asynchrony parameter at different temperatures; PAR, pericarp/acorn mass ratio; Tb, base temperature for germination; T50, germination time at different temperatures; SE, shoot emergence; WC₅₀, seed water content for 50% mortality

Table 4 Correlation matrix between seed traits and climatic variables associated with population sites. R = Pearson’s linear-correlation coefficients. R values in bold indicate adjusted P values < 0.05. Ga, germination asynchrony; Q, quartile; t₅₀, germination time, R, rainfall; SE, shoot emergence; T, temperature; Tb, base temperature for germination