Introduction

Low temperature is an important determinant of the distribution of plants, and it is the first environmental filter that species have to pass to become high-elevation (Körner, 2003). Thus, plants inhabiting high-elevation environments are adapted to cope with the extreme low temperatures that characterize these habitats (Billings, 1974). Predictable frost events, as they occur during winter, are usually not critical for plant survival as cold hardening and snow cover provide protection against frost damage in most high-elevation species (Körner, 2003). However, when freezing temperatures occur during the growing season, high-elevation plants may lose a substantial fraction of their above-ground tissues or die as a result of frost damage (Körner, 2003; Taschler et al., 2003).

The ability to survive freezing temperatures (i.e. freezing resistance) is closely related to the ambient temperature that plants experience (Beck et al., 2004; Bannister et al., 2005). This has led some authors to propose that early in the growing season (i.e. after snow melt) high-elevation plants are fully active and the ability to resist freezing temperatures is minimal. Conversely, late in the growing season (i.e. after fruit ripening and/or the seed dispersal phase) plants are hardened to cope with the unfavorable cold season (Sakai & Larcher, 1987; Larcher, 2003). However, freezing temperatures can occur both early and late in the growing season (Greenland & Losleben, 2001), suggesting that, from an evolutionary point of view, the freezing resistance of high-elevation plants should be high both early and late in the growing season.

Climate change is affecting the length of the growing season, and subsequently the survival and reproduction of high-elevation plants (Inouye & McGuire, 1991; Inouye, 2008; Kudernatsch et al., 2008). Some climatic models predict a reduction of snow cover duration and increases in the frequency and intensity of sudden frost events during the growing season (Menzel & Fabian, 1999; Easterling et al., 2000; Cannone et al., 2007), increasing the risk of damage by frost (Neuner et al., 1999). Thus, it seems important to understand the ability of high-elevation plants to withstand freezing conditions and how this changes as the growing season progresses.

Although there have been many studies on freezing resistance in high-elevation plants, most of them were carried out on mesic mountains, where plants do not experience shortages in water availability (Sakai & Ötsuka, 1970; Taschler & Neuner, 2004; Márquez et al., 2007). By contrast, little is known about summer freezing resistance in xeric mountains (Tyurina, 1957 in Sakai & Larcher, 1987) with studies in Mediterranean-type mountains being particularly scarce (Loik & Redar, 2003; Loik et al., 2004). Studies in xeric mountains are particularly important because freezing temperatures and drought share similar physiological responses. For example, plants can counteract cellular dehydration and turgor loss by osmotic adjustment via synthesis of carbohydrates and proline (Blödner et al., 2005; Nakashima & Yamaguchi-Shimozaki, 2006; Beck et al., 2007). Therefore, the induction of similar physiological responses suggests that the abilities of plants to survive freezing temperatures and drought can be closely connected.

The altitudinal decrease in air temperature may affect the freezing resistance of plants from different elevations (Goldstein et al., 1985; Rada et al., 1987; Halloy & González, 1993; Loik & Redar, 2003). For instance, high-elevation species survive freezing temperatures 4 K lower than low-elevation species in the Northern Chilean Andes (Squeo et al., 1996). Nevertheless, Márquez et al. (2007) found that seven out of nine grass species from the Venezuelan Andes did not show higher freezing resistance at higher elevations, suggesting that altitudinal trends in freezing resistance are not universal. Whether freezing resistance is found to vary with elevation can be determined by the time of the year at which altitudinal comparisons are made. For example, Bannister & Fagan (1989) found that the altitudinal difference in freezing resistance for the fern Blechnum penna-marina measured in summer was smaller than that measured in winter. This is attributable to an increase in the water deficit in B. penna-marina individuals from low elevations, which increases the freezing resistance of the fronds by several degrees (Bannister & Fagan, 1989). Hence, altitudinal trends in the freezing resistance of plants inhabiting xeric mountains may be altered by drought episodes at the end of the growing season, particularly through their effects on plants from low elevations.

The main objectives of this study were to explore the freezing resistance of 24 alpine plant species inhabiting the Mediterranean-type climate of the high Andes of central Chile, and to determine their freezing resistance mechanisms (i.e. avoidance or tolerance). In particular, we assessed whether freezing resistance changes within the growing season, and whether freezing resistance increases with elevation. We hypothesize that freezing resistance of Andean plants does not change within the growing season, but increases with elevation. This information is important for assessing how changes in the length of the growing season and in the frequency and intensity of frost events can affect the survival and distribution of these high-elevation species, and hence for inferring responses in other xeric high-elevation areas.

Materials and Methods

Results

Early in the growing season, the majority of the studied species could resist temperatures as low as −11°C (LT₅₀). Exceptions were Quinchamalium chilense Mol. and Haplopappus schumannii (Kuntze) G.K. et W.D. (both with LT₅₀ = −10.3°C). In this period, Nastanthus spathulatus (Phil.) was the most freezing-resistant species (LT₅₀ = −18°C) (Table 2). Twenty-two out of the 24 investigated species were classified as FT, while only two species (Hordeum comosum (J. Presl.) Löve and Q. chilense) were classified as freezing avoidant (FA) (Table 2). Late in the growing season, most of the studied species resisted temperatures below −10°C (LT₅₀), except for Astragalus looserii I.M. Johnst. (LT₅₀ = −8.2°C) and Calceolaria arachnoidea Graham (LT₅₀ = −8.6°C). Erigeron andicola D.C. was the most freezing-resistant species (LT₅₀ = −19.5°C). In this period, 16 species were classified as FT and eight species as FA (Table 3). Thus, the ability to resist freezing temperatures changed during the growing season, and several patterns emerged depending on the variable analyzed (Fig. 1). Contrary to expectations, no intraseasonal changes in LT₅₀ were detected in only seven out of the 24 studied species (29.2%). Twenty-five per cent of species (six species) were more freezing resistant (more negative LT₅₀) late than early in the growing season, while 45.8% of species (11 species) showed the opposite trend (Fig. 1a). In this last case, the average of freezing resistance early in the growing season was significantly higher than that late in the growing season (LT₅₀ = −15.5 ± 0.6°C vs −10.9 ± 0.6°C; t = 6.05 (paired t-test), P < 0.0001).

Regarding FP, 8.3% of the species showed apoplastic freezing at lower temperatures early than late in the growing season. Conversely, in 62.5% of the species apoplastic freezing occurred at lower temperatures late than early in the growing season (Fig. 1b). Mean FP significantly decreased from −2.1 ± 0.4°C early to −5.6 ± 0.6°C late in the growing season (t = 6.91 (paired t-test), P < 0.0001). Twenty-nine per cent of the species did not change FP within the growing season.

Altitudinal differences in LT₅₀ were found in four out of seven studied species (57.1%) (Table 4). Mean freezing resistance was c. 2 K higher (lower LT₅₀) at high than at low elevations. Three species (Colobanthus quitensis (Kunth) Bartl., E. andicola, and Phacelia secunda J.F. Gmel) did not change their LT₅₀ with elevation.

In six studied species (85.7%) apoplastic freezing (FP) occurred at lower temperatures in plants from low than from high elevations (Table 5). Only C. quitensis did not change its FP with elevation. However, a significant interaction between elevation and time of measurements within the growing season was detected in some species, both for LT₅₀ and for FP (Tables 4, 5), suggesting that altitudinal differences in freezing resistance changed through the growing season.

Discussion

Although our results show a wide range of summer freezing resistance in high-elevation species from the Mediterranean-type zone of central Chile (from −8.2 to −19.5°Cl Tables 2, 3), most of the species were found to be able to resist temperatures below −10°C. Thus, our results indicate that the 24 species studied here are among the high-elevation species with the highest summer freezing resistance reported to date. High-elevation species from Mt Kurodake (Japan) and Mt Patscherkofel (Austria), mountains located in mesic areas, resist freezing temperatures from −3 to −7°C and from −4.5 to −14.6°C, respectively (Sakai & Ötsuka, 1970; Taschler & Neuner, 2004). In mountains located in xeric areas, it has been found that high-elevation species resist temperatures from −7.5 to −16°C in East Pamir (Tyurina, 1957 in Sakai & Larcher, 1987) and from −4.7 to −20°C in the Doña Ana Mountains, northern Chile (Squeo et al., 1996). While the mean summer freezing resistance for lowland species in the Northern Hemisphere is c. 4 K greater than that for species in the Southern Hemisphere, high-elevation plants from the Southern Hemisphere are more freezing resistant than high-elevation plants from the Northern Hemisphere (Bannister, 2007), and our results support this trend. Some authors have suggested that freezing resistance assessed in attached and detached leaves varies depending on the techniques employed (see Bannister, 2007 for a complete review). In particular, Taschler & Neuner (2004) found that the freezing resistance of Austrian Alp species measured on detached leaves was underestimated with respect to attached leaves. Our estimations were based on detached plant material, suggesting that the actual freezing resistance of the high-elevation species of the central Chilean Andes may be even greater than the high values found here.

Most of the species studied here tolerated some extracellular ice formation in their leaves, as indicated by the significant difference between NT and LT₅₀ (Tables 2,3). Therefore, freezing tolerance is the main freezing resistance mechanism in the species studied. Given that freezing avoidance mechanisms can only be effective for a few hours (Goldstein et al., 1985; Rada et al., 1987), these types of mechanism may be less effective than freezing tolerance mechanisms in coping with the extremely low and long lasting freezing temperatures that occur during the growing season in the central Chilean Andes, and therefore, FA species were poorly represented in the species studied. This result is consistent with those of Sakai & Larcher (1987), who suggested that, in areas with severe frosts during the growing season, plants must be capable of tolerating extracellular ice formation even when they are metabolically active.

Given that freezing temperatures can occur early and late in the growing season (Greenland & Losleben, 2001), we expected that the freezing resistance of Andean plants would be similar at the two extremes of the growing season. However, our results indicated that only 29.2% of the studied species resist similar freezing temperatures early and late in the growing season. By contrast, while 25% of the studied species were more freezing resistant late than early in the growing season, c. 45% of the studied species showed higher freezing resistance early than late in the growing season. Data from a climate station located at 2650 m above sea level showed that the mean minimum air temperature increased from 3.6 ± 0.5°C in November 2004 to 7.9 ± 0.4°C in March 2005 (early and late growing seasons, respectively). Similar variation of the mean minimum air temperature was reported by Casanova-Katny et al. (2006), suggesting that the decrease in freezing resistance towards the end of the growing season in the majority of studied species was related to the higher temperature conditions that plants experienced during this period. In addition, a high proportion of the species in our study site lose part or all of their above-ground tissues before the onset of the unfavorable season (e.g. Perezia carthamoides, Quinchamalium chilense and Taraxacum officinale), suggesting that they invest resources to protect the regenerating tissues located below-ground and/or to accumulate reserves in below-ground tissues instead of cold hardening the above-ground organs.

In contrast to freezing resistance trends (LT₅₀), an FP depression at the end of the growing season was observed in 62.5% of the studied species, suggesting that plants are also responding to environmental factors other than temperature. In the central Chilean Andes, soil moisture decreases during the growing season (Table 1), exposing plants to prolonged drought episodes at the end of this period (Cavieres et al., 2006, 2007). FP and NT depend on specific properties of the plant tissues and may vary according to the cell sap concentration and/or the accumulation of water-binding substances inside the cell (Sakai & Larcher, 1987). Several studies have reported that water-soluble carbohydrates depress FP, and their accumulation is positively related to drought survival (Streeter et al., 2001; Merchant et al., 2006; Monson et al., 2006). Moreover, other studies reported that the lower FP of water-stressed plants result from a higher of solute concentration in their tissues (Chen et al., 1977; Goldstein et al., 1985; Anisko & Lindstrom, 1996). Hence, decreases in the FP at the end of the growing season are likely to be related to plant responses to drought rather than to ambient temperatures alone. Although further studies are needed, it seems that the loss of above-ground tissues shown by many plant species in this high-elevation zone is in response to drought occurring at the end of the growing season. Accelerated senescence and leaf abscission are associated with drought in perennial plants, as a strategy that contributes to the survival of the plant and the completion of the plant life cycle under drought stress (Munné-Bosch & Alegre, 2004; Rivero et al., 2007).

As expected, most of the plants from higher elevations resisted 2 K more than plants from lower elevations, which is consistent with results reported in previous studies (Goldstein et al., 1985; Squeo et al., 1996; Taschler & Neuner, 2004). However, our altitudinal difference in LT₅₀ is lower than the altitudinal difference in mean minimum and maximum temperatures reported by Cavieres et al. (2007) (see also Table 1). Our altitudinal difference in freezing resistance is also narrower than the 5.6 K expected from the altitudinal lapse rate, which is c. 7°C km⁻¹ in the summer in the central Chilean Andes (Cavieres & Arroyo, 1999). In the Venezuelan Andes, with an altitudinal lapse rate of 5.8°C km⁻¹, Espeletia schultzii plants resist temperatures 5.9 K lower at 4200 m than at 2600 m (Rada et al., 1987), indicating that the altitudinal difference in freezing resistance in this species is narrower than the 9.3 K expected. Loik & Redar (2003) found that freezing resistance of Artemisia tridentata seedlings differed c. 0.9 K between low and high sites, in Bishop Creek, California where the altitudinal lapse rate is 2.6°C km⁻¹. They attributed the narrow altitudinal difference to microclimatic effects, whereby plants at higher elevations avoid low temperatures beneath the snow pack, while plants at lower elevations are directly exposed to frost events occurring through the growing season. In our case, plants from lower elevations are exposed to drought, which reinforces plant responses to frost events, mitigating the expected altitudinal differences in freezing resistance during the summer. For instance, in the majority of the studied species (85.7%), low-elevation plants had a lower FP compared with those from the higher elevation, with the magnitude of this altitudinal difference increasing at the end of the growing season (Table 5). Hence, the lower FP in plants from lower elevations at the end of the growing season suggests that FP is reflecting plant responses to the water shortage conditions, which are greater at lower elevations (Cavieres et al., 2006). This is reinforced by the statistical interaction between elevation and time of measurement within growing season for FP (Table 5). For example, E. andicola from the low elevation had an FP 1.6 K lower than plants from the high elevation, early in the growing season. However, this altitudinal difference increased to 5 K for plants measured late in the growing season. Additionally, C. quitensis, the only species that did not show altitudinal changes in FP, is a perennial herb inhabiting bogs that provide water availability throughout the entire growing season.

The change in the freezing resistance mechanism from FT to FA observed both in intraseasonal and in altitudinal comparisons was mainly related to a decrease in NT, indicating that tissues have less water to freeze, reducing the probability of ice nucleation (Pearce, 2001). As far as we know, changes in the freezing resistance mechanism within the same species have not been previously reported, suggesting that future studies should consider the time during the growing season at which freezing resistance measurements are conducted before drawing final conclusions. As discussed above, changes from FT to FA are more closely related to plant water status than to low temperatures per se, suggesting that water shortage conditions are altering patterns of freezing resistance in several ways in plants inhabiting xeric mountains.

In conclusion, our results demonstrate that Andean plants have a high ability to resist freezing temperatures during the growing season, and that the main freezing resistance mechanism is freezing tolerance. Freezing resistance of the Andean species varies within the growing season and with elevation, and both intraseasonal and altitudinal variations seem to be strongly affected by water availability. Further studies are needed to separate drought from thermal effects on summer freezing resistance in high-elevation plants from the xeric Andes. Climate models predict an increment of frost events and severe droughts for Mediterranean-type mountains, and thus a better knowledge of the mechanisms underlying freezing and drought resistance will improve any prediction of plant responses to global warming in this and other xeric high-elevation ecosystems.

Acknowledgements

We thank the staff of La Parva and Valle Nevado Ski Resorts for their help with access to our study sites, and also Mauricio Castro and Valeria Neira for their technical assistance in the field and with lab work. We thank C. Lortie and Jürgen Hacker for providing useful comments which improved the manuscript. We also thank Victor and Angélica Rojas from Valparaiso Lodge, our second home. This study was supported by FONDECYT 1060710, 1060910 and P05-002 ICM (Center for Advanced Studies in Ecology and Research on Biodiversity). AS-A is supported by a CONICYT Doctoral Scholarship.