1. Introduction

Seeds can be classified into two broad categories: orthodox and recalcitrant (Roberts, 1973; Pritchard, 2004). Orthodox seeds can be dried to water contents (WCs) as low as 0.02−0.06 g H₂O g^-1 (dry weight basis, hereafter g g^-1) without losing viability, enabling them to tolerate low temperatures (Roberts, 1973; Pritchard, 2004). Recalcitrant seeds maintain high metabolic rates and when dried to WCs of ca. 0.25 g g^-1, are killed due to the loss of intracellular 'free water', which causes vacuoles to collapse, leading to structural damage and death (Vertucci and Farrant, 1995; Farrant et al., 1997; Pritchard, 2004; Berjak and Pammenter, 2013).

The requirement for recalcitrant seeds to maintain high WCs means that at subzero temperatures they face the dual challenges of dehydration and freezing. When these hydrated seeds are exposed to subzero temperatures, intracellular water relocates to the intercellular spaces, leading to dehydration, and large ice crystals form within and between cells causing cell structural damage and death (Hubel et al., 2007; Saragusty et al., 2009; Hyman and Simons, 2012; Wesley-Smith et al., 2014, 2015). While recalcitrant-seeded species are most abundant in humid tropical environments, where these risks are extremely low (Tweddle et al., 2003; Li and Pritchard, 2009), some occur in temperate regions where seeds experience severe winter cold (e.g., Daws et al., 2006; Xia et al., 2022).

The genus Quercus (Fagaceae) contains ca. 450 species and is widely distributed throughout forest ecosystems of the Northern Hemisphere. Apart from Q. emoryoi (SID, 2023), all species studied to date have been reported to have recalcitrant seeds (Xia et al., 2012a, 2022; Amimi et al., 2020). Quercus seeds may display specialized physiological properties and strategies to minimize the likelihood of seed desiccation and freezing over winter (Xia et al., 2012b, 2014, 2022). For example, seeds of subalpine evergreen oaks, from the Himalaya-Hengduan Mountains at altitudes between 2500 and 4300 m, and some temperate oak species from regions with freezing winters (e.g., Q. mongolica) germinate rapidly at low temperatures ensuring germination prior to the onset of extreme winter cold (Xia et al., 2022). In contrast, seeds of subtropical evergreen oaks have relatively high optimum temperatures for germination and may germinate slowly (Xia et al., 2022).

In the Mediterranean region, oak seeds disperse in the relatively mesic autumn and early winter. However, in higher-altitude Mediterranean areas (ca. 600 m), seeds of Quercus ilex and Q. canariensis that often experience frost after dispersal can withstand temperatures as low as -12 ℃, and exhibit greater cold tolerance than seeds of the lower-altitude (ca. 390 m) species Q. coccifera and Q. suber (Amimi et al., 2020). Similarly, under ultra-low temperature stress (in liquid nitrogen LN₂, -196 ℃), different groups of oak species appear to exhibit distinct responses: embryonic axes of the North American species Q. rubra and Q. gambellii, which experience average monthly temperatures as low as -8 ℃ after seed dispersal, survived partial drying and freezing while axes of Q. schottkyana and Q. franchetii from the subtropical regions did not (Xia et al., 2014).

These studies imply that Quercus seeds from cooler regions have ecological strategies (such as rapid germination), that allow seeds to avoid or withstand subzero-temperatures and may possess mechanisms that enable relatively greater cold tolerance. In an intra-specific study of Acer pseudoplatanus across a broad climate gradient, Daws et al. (2008) reported climate-related co-variation in responses to desiccation and subzero-temperatures: the more desiccation-sensitive seedlots were most sensitive to subzero-temperatures. However, we are not aware of other intra- or inter-specific studies of possible climate-related co-variation between responses to desiccation and subzero-temperatures in recalcitrant seeds.

Globally, 31% of oak species are at risk of extinction (Carrero et al., 2020). Cryopreservation (in LN₂) of excised embryonic axes is considered to be the only way to achieve long-term ex situ preservation of these species. Excised axes, unlike whole seeds, may achieve sufficiently rapid cooling rates that reduce ice crystal formation (Walters et al., 2013). However, there are few examples of successfully regenerated oak plants following cryopreservation (e.g., Xia et al., 2014; reviewed in Ballesteros and Pritchard, 2020). Globally, China has the second-highest diversity of Quercus species, and the most endangered species. Among the threatened Chinese Quercus species, more than 77% are Asian subtropical evergreen oaks, including the critically endangered Q. sichourensis (investigated in this study), for which only 10 individuals are known in the natural environment (Xia et al., 2016; Carrero et al., 2020). However, cryopreservation of Chinese species has received little attention; thus, comparative studies of low-temperature responses of Chinese Quercus seeds may have significant importance for ex situ conservation of a large proportion of global oak diversity.

In this study, we investigated the responses of embryonic axes from 13 Quercus species collected across China to ultra-low temperatures. We tested the prediction that the responses of axes to desiccation and cold stress will co-vary and be related to patterns of climate variation. Specifically, we predicted that (1) axes of species from cool locations (i.e., temperate and subalpine oaks) will have a relatively higher degree of both desiccation and cold tolerance than those from warm/subtropical locations, which will enhance the chances for successful regeneration of these plants after cryopreservation, and (2) that species growing in more mesic or warmer habitats (i.e., subtropical evergreen oaks), will be more sensitive to desiccation and subzero temperatures, posing additional challenges for cryo-conserving these species.

2. Materials and methods 2.1. Seed lot details and initial viability

Fully mature fruits (acorns; hereafter seeds) of 13 Quercus species were collected in China (Tables 1 and S1). According to Deng et al. (2018) and Hipp et al. (2020), the species spanned three species groupings: (1) subalpine oaks - Himalayan subalpine lineage, section Ilex, (2) subtropical evergreen oaks, section Cyclobalanopsis which is further subdivided into compound trichome bases (CTB) and single-celled trichome bases (STB) lineages and (3) temperate species from sections Cerris and Quercus.

Seeds of all species were collected from at least five individual trees at the period of peak seed dispersal in late autumn between 2017 and 2021 (Table 1). Seeds were stored in loosely tied plastic bags at either 15 ℃ (subtropical species) or 5 ℃ (subalpine and temperate species). All experiments commenced within one week of seed collection. Only seeds that were not visibly infested by fungi or weevils were used in experiments. All seeds were randomly selected for experimentation. Axes and cotyledons from 10 to 20 seeds of each species were separately oven-dried at 103 ℃ for 17 h to determine dry mass (DM) and WC on a dry weight basis.

Initial axis viability was assessed using the methods described in Xia et al. (2014). Axes were excised and kept moist in a solution of citric and ascorbic acid (0.2 g L^-1) until sufficient numbers were accumulated for study. This took approximately 2 h for each species. Axes were then surface-sterilized in 1% sodium hypochlorite for 10 min, rinsed twice in sterile water and transferred to Woody Plant Medium with 0.3% charcoal. Axes were subsequently incubated at 20 ℃ (subalpine and temperate species) or 25 ℃ (subtropical species). A light/dark photoperiod of 12/12 h was used with axes kept in darkness for the first 48 h. Viability monitoring started after 3 days. Every week, axes were categorized based on whether they had: (1) expanded, turned green or formed callus, (2) showed normal development of either roots, or (3) shoots, or (4) produced a normal seedling with both root and shoot development. Survival of just roots (2) or shoots (3) was recorded separately because shoot survival has been reported to be lower than root survival in previous studies on Quercus survival both post desiccation and post freezing (Xia et al., 2014). In addition, shoots of subalpine species can exhibit epicotyl dormancy (Yang et al., 2020). For each treatment × species combination, aliquots of ten axes each, for a total of 30–60 axes (repeated on two or three separate days) were used.

2.2. Dehydration treatments and relative desiccation tolerance of axes

To rapidly dehydrate axes, hydrated axes were placed on blotter paper over freshly regenerated silica gel in fully sealed boxes at room temperature for different durations (15−1400 min). Drying durations were adjusted to achieve a range of final WCs. The maximum WC at the beginning of drying, the WC at each drying interval, and DM for all WC levels were measured for three to five individual axes. Following drying, axes were slowly rehydrated on damp blotter paper for 1 h. These rehydrated axes were then surface-sterilized and placed on the growing medium (described above) to assess survival and growth.

Drying time course curves were calculated following Xia et al. (2014). The drying rate model was calculated from the linear relationship between drying time t and ln (WC at time t – final WC). The intercept of the model was constrained by the average maximum WC for the tissue; this value corresponds to the WC of hydrated or freshly excised axes prior to drying treatments. For each species, the slope of the model is regarded as the axis drying rate coefficient.

For each species, post-desiccation viability percentage of axes versus WC was analyzed by two linear lines. The horizontal lines represent the proportion of normal axis growth before the onset of desiccation related stress symptoms. Diagonal lines were the regression of water content versus growth at lower water contents. The intersection of the two lines was calculated as the damaging WC, the point during drying at which the onset of abnormal growth or organ mortality was observed.

The relationship between WC and water potential (WP) was determined for axes and cotyledons using pressure–volume relationships as described in Xia et al. (2014, 2022). Both initial WCs and the damaging WCs were converted to WPs using these relationships. Damaging WC and WP were used to describe the relative desiccation tolerance among species which was quantified by the WC or WP at which loss of viability or reduced growth was apparent.

2.3. Cryo-exposure and relative cold tolerance

Freshly excised axes or axes dried to a range of WCs were exposed to LN₂ (-196 ℃) using cooling methods described in Xia et al. (2014). Axes were enclosed in either (1) 27 mg lightweight aluminium foil packets (2 cm × 3 cm) (two axes per packet) and forcibly plunged into LN₂ slush, which gave cooling rates of 30–80 ℃ s^-1, or (2) 1.2 ml screw-cap polypropylene cryovials (ten axes per cryovial), which were submerged into LN₂ with cooling rates of 3–8 ℃ s^-1. These two methods have been shown previously to be effective for three Quercus species (Xia et al., 2014). Axes were held at LN₂ temperatures for 18 h and then rapidly warmed by immersion in a 0.5 M sucrose solution at 42 ℃ (Xia et al., 2014). Re-warmed axes were surface-sterilized and placed on growth medium to assess survival and growth. Across all the cryo-exposure treatments, the absolute highest viability, encompassing all four recorded survival categories, was used as an indicator of the relative cold tolerance of each species. The WC at which axes had the absolute highest viability after cryo-exposure (hereafter optimal cryo-exposure WC) was also noted.

2.4. Correlations between axes traits and climate variables

Data for responses of the 13 Quercus species, from this study, to desiccation and cold stress were combined with data for four additional Quercus species from Xia et al. (2014). This included data for the Chinese species Q. schottkyana (subtropical oak), Q. franchetii (section Ilex, temperate oak group), as well as the temperate American oak species Q. gambelii (two seedlots; section Quercus) and Q. rubra (section Lobatae). This resulted in data for a total of 17 species and 18 seed lots for the current analysis.

For the 17 species, climate data for each species was accessed from WorldClim v.2.1 (Fick and Hijmans, 2017) as described in Xia et al. (2022). Seven climatic variables with biological significance were used: mean annual temperature, the minimum temperature of the coldest month, total annual precipitation, the seasonality of precipitation, total precipitation of the coldest quarter, the mean diurnal temperature range and temperature seasonality.

The 17 (species) × 6 (axis traits: axis DM, initial axis WP, axis drying rate, damaging WP, relative cold tolerance and optimal cryo-exposure WC) and 17 (species) × 7 (climatic variables) matrices were subjected to principal components analysis (PCA) to explore intercorrelations among axis traits or climate variables. Pairwise Pearson's correlation coefficients were calculated for all combinations of either axis traits or climatic variables. Linear regression was used to test for relationships between the first two principal components from the axis traits PCA and those from the climate PCA. Differences in axis traits and climatic variables between the three species groupings were tested using PERMANOVA and one-way ANOVA followed by Fisher's least significant difference (LSD) test.

3. Results 3.1. Axis size, water status and initial viability

Subalpine oak axes had significantly higher initial WCs (mean = 2.29 g g^-1) than those of both subtropical (mean = 1.45 g g^-1) and temperate oaks (mean = 1.46 g g^-1) (One-way ANOVA, P < 0.05; Fig. 1a and Table S2). Initial cotyledon WCs ranged from 0.61 to 1.66 g g^-1 and did not differ significantly across the three groups (P > 0.05; Table S2).

Across the study species, axis and cotyledon DM varied from 0.92 to 6.18 mg and 375 to 4465 mg, respectively (Table S2). There were no significant differences between the three groups for axis DM (One-way ANOVA, P > 0.05).

The initial WPs of axes and cotyledons ranged from -0.05 to -3.92 MPa and -0.01 to -5.84 MPa, respectively (Fig. S1 and Table S2). Subalpine oaks had significantly less negative initial axis WPs (mean = -0.62 MPa) than the temperate oaks (mean = -3.32 MPa) (One-way ANOVA, P < 0.05; Fig. 1a). Axis WPs for the subtropical oaks (mean = -1.71 MPa) did not differ significantly from either the subalpine or temperate oaks (Fig. 1a). Initial WPs for the cotyledons did not differ significantly between the groups (Table S2).

The initial axis WP of all subalpine species was either similar to, or less negative than, the initial cotyledon WP (Table S2). For the temperate and subtropical evergreen oak species only Quercus mongolica, Q. sichourensis and Q. fleuryi had initial axis WPs that were less negative than the cotyledons (Table S2). The remaining five species had initial axis WPs that were more negative than those of the cotyledons (Table S2).

Freshly excised and non-dried embryonic axes had high viability in culture. Across all species at least 86% of axes were viable (Fig. 1c). Root development in culture was high for most species (75.9–100%), except Q. annulata and Q. sichourensis, which had root development of 30% and 58.6%, respectively. Across species, shoot development was generally lower than root development (One-way ANOVA, P < 0.05; Fig. 1c and Table S2). Subalpine species had significantly lower shoot development (mean = 17.1%) than temperate oaks (mean = 63.9%; One-way ANOVA, P < 0.05; Fig. 1c). Shoot development of subtropical evergreen oaks (mean = 53.5%) did not differ significantly from either the subalpine or temperate oaks (P > 0.05).

3.2. Axis response to desiccation

Across species, axis drying rates varied from 0.005 g g^-1 min^-1 for Quercus serrata to 0.024 g g^-1 min^-1 for Q. senescens (Fig. S2 and Table S3). Drying rates did not differ significantly between the three oak groups (One-way ANOVA, P > 0.05; Table S4).

For most species, slight drying had no effect on survival or root and shoot development (Fig. S3). As drying time increased, resulting in lower axis WCs, normal seedling development was reduced. With further drying, signs of viability (i.e., expansion and greening of axes) were lost (Fig. S3). Root and shoot growth were most sensitive to reductions in WC, with axis survival (defined as the presence of at least one of the following: expansion and greening, root growth, shoot growth, or normal seedling growth) occurring for most species following drying to lower WCs (Fig. 1b and Table S3). Across species, the average WC at which shoot growth, root growth and axis survival were prevented (damaging WCs) were 1.22, 1.05 and 0.70 g g^-1, respectively. Temperate oak axes survived drying to significantly lower WCs (mean = 0.32 g g^-1) and more negative WPs, (mean = -15.07 MPa) than both subalpine (mean = 0.71 g g^-1 and -5.97 MPa) and subtropical oaks (1.06 g g^-1 and -2.96 MPa) (One-way ANOVA, P < 0.05; Fig. 1a and b, Table S3). Damaging WCs and WPs did not differ significantly between subalpine and subtropical oaks (One-way ANOVA, P > 0.05).

The drying times for axis viability loss (i.e., a 50% decline in axis survival) varied from 108 min for Quercus delavayi to 2090 min for Q. mongolica (Table S3 and Fig. S4). Drying times for viability loss of individual organs (i.e., root or shoot growth) were shorter and ranged from 20 min for Q. sichourensis to 1685 min for Q. mongolica (Table S3 and Fig. S4). Between the three species groups, drying times for axis or organ tissue viability loss did not differ (One-way ANOVA, P > 0.05; Table S4).

3.3. Response to low-temperature stress

Axes cooled in cryovials (cooling rate 3–8 ℃ s^-1) generally had higher survival and more normal development over a wider range of WCs than those cooled in foil packets (cooling rate 30–80 ℃ s^-1; Fig. 2). In both cryovials and foil packets, subalpine oak axes had significantly higher survival (mean = 68.9% and 53% for cryovials and foil packets, respectively) and root development (mean = 48.4% and 33.8%) than most temperate oak species (mean = 26.7% and 16.1% for the axes and 10% and 3.3% for the roots) and subtropical oaks (≤ 5% for all treatments) (One-way ANOVA, P < 0.05; Fig. 2). Except for Q. mongolica, shoot development was consistently lower than root development post cryo-exposure (One-way ANOVA, P < 0.05). For subtropical oak axes, both methods were mostly lethal across a broad range of WCs (0.10–3.15 g g^-1). The only exception was a low percentage of Q. annulata axes cooled in cryovials, which showed evidence of callus formation and root development (Fig. 2).

For most subalpine species (except Quercus guyavifolia), axes survived cryo-exposure at their initial WC or high WCs (≥ 1.4 g g^-1). For example, Q. monimotricha had 100% survival and 90% organ tissue development after cryo-exposure at an initial WC of 3.2 g g^-1 (Fig. 2). Partially desiccated Q. monimotricha and Q. spinosa axes had lower survival and organ tissue growth following cryo-exposure than fresh axes. However, for Q. mongolica and other temperate species, survival and organ development tended to increase as the WC of cryo-exposed axes decreased to levels close to (0.10–0.24 g g^-1 for Q. mongolica) or higher than the damaging WC (ca. 0.7 g g^-1) at which desiccation damage occurred (Fig. 2 and Table S3). Across the temperate oaks, only Q. mongolica had high survival (100%) and organ growth (90%) following cryo-exposure at low WCs (ca. 0.1 g g^-1). Axes of the additional temperate oaks had low (e.g., Q. serrata and Q. variabilis) or zero survival (e.g., Q. acutissima) after cryo-exposure at all WCs.

3.4. Correlations between axis traits and responses and climatic variables

Axis traits of the 17 species (including four species from Xia et al., 2014) were characterized by PCA axes 1 (PC1_trait) and 2 (PC2_trait), which had eigenvalues of 1.84 and 1.51, explaining a combined 55.8% of the variance (Fig. 3a). Initial axis WPs were positively and significantly associated with optimal cryo-exposure WC and PC1_trait (Fig. 3a and Table S5). The damaging WP was negatively and significantly associated with axis relative cold tolerance and PC2_trait but positively and significantly associated with PC1_trait. The axis relative cold tolerance and optimal cryo-exposure WC were positively and significantly associated with PC2_trait and PC1_trait respectively (Fig. 3a and Table S5). Axis DM and drying rate were not correlated with any other traits or either Principal Component.

The overall suites of axis traits differed significantly between the three species groupings (PERMANOVA, P < 0.05). Axis traits of temperate species from relatively warmer areas [i.e., Quercus variablis (11), Q. serrata (13) and Q. franchetii (14)] overlapped with subtropical species. Subtropical species clustered along PC1_trait and were separate from the subalpine species and other temperate species (Fig. 3a). Axes of subtropical species had less negative damaging WP (i.e., more desiccation sensitive) and limited survival after cryo-exposure.

Axes of subalpine species had higher initial WP and less negative damaging WP but their viability after cryo-exposure was higher. The temperate oaks had a much wider 75% confidence ellipse. Except Quercus variablis, Q. franchetii and Q. serrata being closer to the subtropical oaks, other species typically Q. mongolica, Q. gambelii and Q. rubra, were clustered along PC2_trait and the negative direction of PC1_trait and separate from the subalpine and subtropical species. Temperate oaks had lower initial WP, were less desiccation-sensitive (lower damaging WP), and Q. mongolica, Q. gambelii and Q. rubra from the coldest areas in this group had high viability after cryo-exposure.

The climate PCA axis 1 (PC1_climate) and 2 (PC2_climate) had eigenvalues of 3.72 and 1.90 explaining 80.2% of the variance in the seven climatic variables (Fig. 3b). Along PC1_climate, temperatures and precipitation increased while temperature seasonality and mean diurnal range declined (Fig. 3b and Table S6). Along PC2_climate precipitation of the coldest quarter increased while precipitation seasonality declined. The three species groups were distributed in habitats with differing climates (PERMANOVA, P < 0.05; Figs. 3b and S5). Subtropical oaks were distributed in hotter (MAT = 17.2 ℃) areas than the subalpine (MAT = 9.2 ℃) and temperate oaks (MAT = 9.9 ℃) (Fig. S5). Annual precipitation for subtropical species (Mean = 1116 mm) was significantly higher than for temperate species (Mean = 766 mm), but did not differ from the subalpine species (Mean = 982 mm). The lowest monthly temperatures experienced by subtropical oaks (Mean = 3.9 ℃, range 1.9–6.7 ℃) were significantly higher than for subalpine (Mean = -3.47 ℃, range -6 to 0.2 ℃) and temperate oaks (Mean = -7.2 ℃, range -19.2 to 4.7 ℃).

PC1_climate was positively correlated to PC1_trait but negatively correlated to PC2_trait (Fig. 3c and d). Therefore, initial axis WP, damaging WP and optimal cryo-exposure WC were positively related to annual mean and minimum temperature, and annual precipitation, but negatively related to mean diurnal range and temperature seasonality. Conversely, relative cold tolerance showed the opposite relationships. Axis dry mass and axis drying rates were unrelated to the climatic variables.

4. Discussion 4.1. Different cold resistant strategies between oak groups

Our study is the first to investigate responses of subalpine oak axes to subzero-temperatures and demonstrates distinct responses compared with the oak taxonomic groups studied previously. Axes of both subalpine and temperate species from regions with temperatures below freezing in winter survived cryo-exposure. However, despite high initial WCs and being unable to survive much water loss, axes of subalpine oaks had generally higher post cryo-exposure survival than temperate oaks. Surprisingly, subalpine oak axes could be frozen at high WCs without lethal freezing injury.

In contrast, for temperate oaks only axes of species with comparatively higher desiccation tolerance survived cryo-exposure, and axes of this group typically only survived freezing when dried to lower WCs. Among this group, Quercus mongolica axes were the least desiccation sensitive and achieved 100% survival after cryo-exposure at a comparatively low WC of 0.10 g g^-1. Across recalcitrant seeded species, surviving exposure to low temperatures is generally attributed to a balance between the ability to remove enough water, which reduces the likelihood of damaging ice crystal formation, without causing too much desiccation related damage (Pammenter and Berjak, 2014; Xia et al., 2014).

When axes are rapidly dried, the WC at which axes experience desiccation related mortality is lowered which also increases the likelihood of surviving cryopreservation (Pammenter and Berjak, 2014; Walters and Pence, 2021). Comparing desiccation responses of rapidly dried axes in this study with intact seeds of the same species dried slowly, Xia et al. (2022) showed that while the relative desiccation tolerance for rapidly and slowly dried axes were correlated (P < 0.05; Fig. S6), there was a downward shift in the critical WC and WP (WC and WP at which axes lost 50% viability) with rapid drying (mean WC of 0.28 and 0.82 g g^-1 and mean WP of -18.59 and -5.26 MPa for rapidly dried and slow dried axes, respectively) (P < 0.05; Fig. S6). Rapidly cooling axes can also enhance their survival of cryo-exposure by reducing the likelihood of ice crystals forming (Wesley-Smith et al., 2015; Walters and Pence, 2021). The temperate oak species in our study exhibited this general pattern of increased post cryo-exposure survival following non-lethal reductions in WC; subalpine species did not.

For temperate oaks, axis WP was typically less negative than that of the cotyledons promoting water movement away from the axis and potentially reducing the likelihood of lethal freezing injury to the axis. Consequently, Xia et al. (2014) regarded this as an adaptive mechanism for the temperate species Quercus rubra and Q. gambelii, which both experience cold winters (mean minimum temperatures around -8 ℃) in North America. In our current study we broaden this pattern, and potential adaptive mechanism, to other temperate oaks including the Chinese species Q. mongolica. In contrast, subalpine oaks had the opposite WP gradient between the axis and cotyledons; water was most likely to flow towards the axis. While this may reduce the likelihood of desiccation damage it may increase the likelihood of freezing damage. However, our data suggest that mechanisms have evolved to prevent lethal freezing injury at high WCs of subalpine oak axes. These mechanisms may be expected to either (1) prevent ice crystals forming, (2) control the location of ice crystal formation, (3) control the structure and size of ice crystals within the axis or (4) protect critical cellular machinery by freeze-desiccation or vitrification mechanisms (Wesley-Smith et al., 2014).

A similar ability to survive cryo-exposure at high WCs has been reported for the temperate recalcitrant-seeded species Acer saccharinum (Wesley-Smith et al., 2014). Fresh embryonic axes with high WCs (1.5–1.9 g g^-1) had ca. 90% survival for axes cooled, at 3.3 ℃ s^-1, to -196 ℃ (Wesley-Smith et al., 2014). For A. saccharinum, ultrastructure studies demonstrated that axes survived because intracellular ice crystals were small and at low density (Wesley-Smith et al., 2014). In fully hydrated axes cooled at an intermediate rate, the interior of many organelles appeared ice-free (Wesley-Smith et al., 2014). To address these issues, we are currently undertaking ultrastructural studies of axes of Quercus species post cryo-exposure.

Subtropical oaks from both lineages had similarly high initial WCs, WPs and desiccation sensitivity as the subalpine oaks but had minimal survival following cryo-exposure. Subtropical oaks generally occur in warmer and wetter environments than the subalpine oaks, with a relatively low risk of cold injury and seed desiccation. Evidence from the fossil records and molecular analyses suggests subalpine oaks originated in tropical or subtropical forest habitats and colonized cold habitats during the Himalaya-Hengduan Mountains uplift and associated climatic cooling since the mid-Miocene (Meng et al., 2017). Presumably this shift was associated with the evolution of seed cold tolerance. A similar relatively rapid evolution of cold tolerance since the mid-Pleistocene in vegetative tissue has been reported in a live oak clade in the United States (Quercus section Virentes; Koehler et al., 2012).

4.2. Axis cold responses and climates

Our analysis, for a range of Quercus species, of the statistical relationships between axis traits and responses to desiccation and freezing and their habitat climates, demonstrated a range of consistent patterns. Thus, species from colder and drier places (i.e., temperate species) had axes with lower initial WCs and greater relative desiccation tolerance. This enabled WCs to be reduced sufficiently to enable cryo-exposure survival. Conversely, species from warmer and wetter habitats (i.e., subtropical and some temperate oaks, e.g., Q. acutissima, Q. franchetii and Q. variabilis) were more desiccation and cold sensitive. Seeds of these species were susceptible to either ice damage at higher WC during cryo-exposure or desiccation, leading to their demise. However, several species of subalpine oaks, (a group that only contains 7 species in total) deviated from these general patterns (Fig. 3c and d), and they survived freezing at high WC, despite having other axis traits (e.g., initial WCs/WPs and desiccation-sensitivity) that were most similar to the subtropical oaks.

A limited ability, compared to temperate species, to survive both sub-zero and ultra-low temperatures has been suggested to be a feature of recalcitrant species from warmer climates (Berjak and Pammenter, 2014; Walters and Pence, 2021). Our data for con-generic species from a broad climatic range provide empirical support for this notion. These climatic related differences in response to cryo-exposure may relate to differences in the ability to control the formation of ice crystals and partition where they occur within cells (e.g., Wesley-Smith et al., 2014). Alternatively, the ability to tolerate cryo-exposure may relate to differences in axis lipid composition between tropical and temperature/subalpine species. For example, Dussert et al. (2001) observed a negative correlation between the percentage of unsaturated fatty acids and seedling recovery in nine species of oil rich Coffea seeds after exposure to LN₂.

4.3. Cryo-conservation of the genus Quercus

As the only option for long-term storage of recalcitrant-seeded species (Walters et al., 2013), cryo-conservation of threatened Quercus species is increasingly important. Despite a number of studies, root and shoot recovery of Quercus species after exposure to LN₂ is relatively low (reviewed in Ballesteros and Pritchard, 2020). Our results support earlier studies and suggest that pre-drying axes to levels near or above the damaging water content for desiccation followed by rapid cooling and re-warming can improve survival of temperate Quercus species. Our results also demonstrate a novel pattern of responses to cryo-conservation in subalpine Quercus species. In conclusion, the limited success in our study with subtropical Quercus species, a group with a significant number of threatened species most of which occur in China, suggests that this group requires significant additional research to develop long-term conservation options.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (NSFC) (Grant Nos. 32260332 and 31770358) and the Youth Talent Program of Xingdian Yingcai Support Plan, Yunnan Province (Grant No. XDYC-QNRC-2022-0023) to Ke Xia, and Major Program for Basic Research Project of Yunnan Province (202101BC070002). We thank Lei Fan, Shan-Shan Hu and Jia Liu for technical assistance and Dr. Min Cao, Dr. Wen-Yun Chen, Prof. Lin Cao, Dr. Xiao-Guo Xiang, Dr. Lin-Bo Jia and Hua-Jie He for assistance with seed collection. We thank the Germplasm Bank of Wild Species, Kunming Institute of Botany, Chinese Academy of Sciences for the help during the laboratory work.

CRediT authorship contribution statement

Ke Xia: Study design, Investigation, Formal analysis, Data interpretation, Writing – Original Draft, Writing – Review & Editing, Funding acquisition. Matthew I. Daws: Data interpretation, Writing – Review & Editing. Zi-Qi Zhu: Investigation.

Declaration of competing interest

The authors declare no conflicts of interests.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.04.004.