3 RESULTS
3.1 Bioclimatic niche characteristics of analyzed Cochlearia andIonopsidium accessions show severe differences
The principal coordinate analysis revealed a cumulative proportion of variance of 91% for the first three principal components (PC1,65.9%; PC2,19.2%; PC3,5.9%). As shown in Fig. 2a, four Cochleariaecotype groups were defined, with Ionopsidium forming a fifth and separate group of accessions. Mostly along PC1, alpine accessions (C. excelsa and C. tatrae ) were grouped into cluster 1. Cluster 2 was comprised primarily by inland taxa such as C. polonica and C. pyrenaica , but also included arctic C. groenlandica . A third cluster combined polycarpic coastal taxa (C. anglica and C. aestuaria ). Monocarpic accessions ofC. danica were considered as a separate cluster 4, which shows closest affinities with Mediterranean Ionopsidium representing cluster 5 also defined exclusively by monocarpic taxa. These results are consistent with the findings of Wolf et al. (2021), who identified the same groups based on nine of the 19 bioclimatic variables. These results show that Cochlearia and Ionopsidium accessions are exposed to varying environmental conditions, which is expected considering their geographically distant distribution.Ionopsidium accessions can be aggregated into a single group, suggesting that they inhabit regions with similar precipitation rates and temperatures. Cochlearia accessions, however, were separated into different groups, suggesting a wider distribution and habitat sites that vary in their bioclimatic character along PC1. The orientation and length of loading vectors of variables indicate the method through which variables contribute to principal components, and therefore to separate different clusters (Fig. 2b). Notably, precipitation- and temperature-dominated variables separate clusters on the left (mostly polycarpic Cochlearia accessions) from clusters on the right (monocarpic coastal, Mediterranean Ionopsidium ) in a very similar way. The differentiation of clusters along axis 1 suggests that the habitats of alpine and inland Cochlearia species may be strongly influenced by the amount and regularity of precipitation under cold conditions, whereas for coastal and Mediterranean species, temperature, especially warm temperatures, may play a more prominent role in ecological separation. This raises the question of whether continentality, which is defined by a strong seasonality of temperature and lower precipitation (Bruch et al., 2011), strongly influences the inland distribution of Cochlearia species. Based on these results, it would be expected for different ecological groups to vary in their physiological adaptation to cold; this is especially relevant for groups that were most clearly separated, such as alpineCochlearia and Ionopsidium accessions, as differences in these groups´ bioclimatic niches are supposedly the strongest.Cochlearia danica and coastal Cochlearia species may exhibit a cold response that is more similar to Ionopsidiumspecies, as these groups were positioned closely together in the PCoA analysis.
3.2 Cold acclimation similarly enhanced the freezing tolerances ofIonopsidium and Cochlearia accessions
Cold acclimation for five days at 4°C enhanced the freezing tolerance of all Cochlearia and Ionopsidium accessions.LT50 and LT100 values were significantly lower for acclimated samples than for non-acclimated samples (LT50 : t = -6.5886, df = 61.131,p -value = 5.824e-09; LT100 t = -5.8435, df = 79.099, p -value = 5.437e-08). AcclimatedLT50 ranged from -2.82°C to -12.82°C with a mean value of -7.06°C, whereas non-acclimated LT50values were generally higher, ranging from -2.34°C to -9.24°C with a mean of -4.17°C (Table 1). This was also true forLT100 values, with acclimated values ranging from -4.24°C to -18.09°C (mean: -11.56°C) and non-acclimated values ranging from -2.12°C to -11.78°C (mean: -7.37°C) (Table 2). There was substantial variation in the lethal values for all the measured accessions (Table 2). Cold acclimation potential is indicated by the difference between LT values of acclimated and non-acclimated samples, as it shows how the freezing tolerance of individuals increases through exposing plants to low but non-freezing temperatures for a certain period (in this study, 4°C for five days). This difference was expressed in terms of the LT50 and ΔLT100 values (Tables 1 and 2). The larger this difference, the greater the freezing tolerance of plants through cold acclimation. The LT50 andLT100 values were exclusively positive, which supports the observation that acclimated lethal values were considerably lower than non-acclimated values. There was substantial variation in the cold acclimation potential of different accessions, withLT50 ranging from 0.48°C to 6.59°C (mean:2.91°C) and LT100 ranging from 0.06°C to 10.22°C (mean:4.18°C). Generally, ΔLT100 values were significantly higher than Δ LT50 values (p -value 0.004), which could be expected because freezing damage is not a linear function. Examples of freezing tolerance measurement experiments are provided in Fig. 3, and individual measurements are shown in Suppl. Mat. Table 2. The mean values for our internal controlArabidopsis thaliana Col0 were -8.1°C (SD:0.7) and -4.1 (SD:0.3) (LT50 , acclimated and non-acclimated, respectively), which is also within the range of previously reported values for this ecotype (-9.7°C and -5.5°C; Hannah et al., 2006).
3.3 Freezing tolerance variation within and between species: not taxonomic group specific
Substantial variation in LT50 andLT100 values were observed within species (Tables 1 and 2). The highest range of LT50 values was exhibited by acclimated C. tatrae samples, spanning a difference (min-max) of 6.92°C, and non-acclimated I. abulense samples, at 4.6°C. For LT100 values, C. tatrae showed the highest range of values among species for both acclimated and non-acclimated samples, at 8.21°C and 9.3°C, respectively. If standard deviations are used to compare variability in the data, considering varying sample sizes, there was also substantial variation within (and among) species (Fig. 4). The various species showed, on average, a much lower standard deviation in non-acclimated LT50values compared to all other values. For acclimated samples, C. tatrae had the highest standard deviation for bothLT50 and LT100 values. This species also showed the highest standard deviation forLT50 values of the non-acclimated samples. Although LT50 values of non-acclimated samples presented the lowest standard deviation values, there was also minimal difference among species, with only I. abulense showing the highest value. Notably, a small sample size did not necessarily result in a higher standard deviation, as was expected. Cochlearia pyrenaica (n = 9) and C. tatrae (n = 6) showed higher standard deviation values than some species with lower sample sizes, such asC. excelsa (n = 2). This suggests that, besides varying sample sizes, other differences, such as genetic variation within species, likely influence the variation in the measured data. Because only one accession was measured for I. glastifolium and I. megalospermum , no standard deviation could be calculated for these species. Given that these two species are genetically very similar and are sometimes referred to as subspecies (Vogt, 1987; Koch, 2012) and as they are also distributed in the same regions, they may be compared here as if they were a single species. A similar cold response with very similar lethal values was observed for these two taxa (Fig. 4).
Comparisons among species revealed that C. danica (coasts of the Atlantic Ocean and Northern Sea) and I. abulense (Spain mainland) showed the lowest LT50 values for acclimated samples (-10.72°C and -9.96°C, respectively) (Table 1). Both species are adapted to hot, dry, and in the case of C. Danica , coastal, high-salt conditions (Fig. 1). This result supports the assertion thatC. danica exhibits a similar cold response compared toIonopsidium owing to the similar environmental conditions of their habitats (Fig. 3). This was contrary to the expectation that these species would display the lowest cold tolerance. We expected that species such as arctic C. groenlandica and high alpine C. tatrae and C. excelsa exposed to the lowest ambient temperatures would display the highest frost tolerance and the lowest lethal values. Interestingly, a different species pair showed the lowest acclimatedLT100 values: I. abulense with -15.48°C and C. excelsa with -15.26°C, which shows that even thoughC. excelsa did not show the lowest LT50 value, it may still be able to withstand more extreme temperatures than other species.
Coastal C. danica , which also occurs in Portugal and Spain, may exhibit a cold response similar to that of Ionopsidium species, as these species may be exposed to similar environmental conditions. However, this assumption was not supported by the measured data (Fig. 4). Because lethal values vary considerably within theIonopsidium group, comprehensive comparisons of this group other species proved difficult. Notably, I. abulense showed much lower lethal values (both LT50 andLT100 ) than the remaining species of this group. As stated above, C. danica did exhibit lethal values similar to those of I. abulense . Coastal C. anglica and C. aestuaria exhibited high lethal values that were most similar to those of other Ionopsidium species. However, no obvious distinction between alpine/arctic Cochlearia species and coastal species, such as Ionopsidium , could be identified. ANOVA showed that, generally, there was a significant difference in lethal values between species (LT50 acclim: ***,LT50 non-acclim: **, LT100acclim: **, LT100 non-acclim: **, df = 12). However, multiple t-tests revealed mostly insignificant differences. As Table 3 shows, only nine of the 55 comparisons were significant.
The two evolutionary lineages leading to the sister generaCochlearia and Ionopsidium diverged from each other approximately ten million years ago. However, comparisons of lethal values between the two genera revealed insignificant differences, as shown in Fig. 5 (LT50 acclimated, p = 0.822; LT100 acclimated, p = 0.883;LT50 non-acclimated, p = 0.0599;LT100 non-acclimated, p = 0.249). This further substantiates the finding that Cochlearia andIonopsidium species respond similarly to freezing temperatures. This is contrary to the expectation that, as western MediterraneanIonopsidium forms a separate bioclimatically defined group (Fig. 2), the genus would respond differently to freezing temperatures. Similarly, there were no differences when comparing polycarpic versus monocarpic and diploid versus polyploid accessions (Table 4).
3.3 Freezing tolerance shows weak geographically defined trends
Low temperatures, especially winter minimum temperatures, are important in determining the geographic boundaries of plant species distributions. Therefore, it is expected that a species’ tolerance to freezing temperatures is often correlated with its geographic distribution (Armstrong et al., 2020). There is indeed an environmental gradient strongly associated with temperature, which may create a gradient in natural selection with strong selection pressures for an increased cold tolerance towards the north (Wos & Willi, 2015; Armstrong et al., 2020). A significant correlation between lethal values and longitude has been demonstrated for A. thaliana accessions, suggesting that there may also be a continentality factor influencing the response to cold, as conditions become colder and drier with increasing distance from the coast (Bruch et al., 2011; Zuther et al., 2012). Therefore, geographically distant taxa are expected to differ in their responses to cold (Davey et al., 2018). As shown in the previous chapter,Cochlearia and its sister clade Ionopsidium exhibit similar responses to freezing temperatures, despite being distributed in different geographic regions with varying bioclimatic conditions. To elaborate further on this spatial distribution,LT50 and LT100 values for the different Cochlearia and Ionopsidium accessions have been plotted on a map (Fig. 6). The range of lethal values (acclimated and non-acclimated) is indicated by a color gradient. We determined that accessions at high latitudes would display higher cold tolerances and therefore show lower lethal values than accessions at low latitudes, thereby forming a gradient of cold tolerance. As inland species may be exposed to a colder winter climate than coastal species owing to continentality, a longitudinal gradient may also occur. No such gradient was initially identified for non-acclimated samples, as these plants did not undergo the process of cold acclimation that would naturally occur. Acclimated LT50 values showed a weak trend towards a gradient that spanned from southwest to northeast with lower lethal values, indicating an increase in freezing tolerance; this was partly mirrored by LT100 values for acclimated samples. However, the northwestern arctic C. groenlandica was an outlier in both cases, showing the highest freezing tolerance. In addition, I.abulense accessions located in the southwest showed a comparatively high freezing tolerance, thereby suggesting that the supposed gradient is extremely weak. This also suggests that a combination of latitude and longitude might influence freezing tolerance instead of a single factor. Correlation analysis supports these observations. Both the correlation between lethal values and latitude (LT50 acclim: R = -0.3; p = 0.057,LT100 acclim: R = -0.2; p = 0.22,LT50 non-acclim: R = 0.023; p = 0.88,LT100 non-acclim: R = -0.12; p = 0.44) and the correlation between lethal values and longitude (LT100 acclim: R = 0.056; p = 0.73,LT50 non-acclimated: R = 0.3; p = 0.059,LT100 acclim: R = -0.0094; p = 0.95,LT100 non-acclim: R = 0.21; p = 0.19) were not significant. However, in reviewing the correlation analysis between lethal values and these two factors, the lethal values of the acclimated samples seemed to be slightly more correlated with latitude.