Summer frost resistance and freezing patterns measured in situ in leaves of major alpine plant growth forms in relation to their upper distribution boundary

Authors


Dunja Taschler. Fax: +43 512 5072715; e-mail: dunja.taschler@uibk.ac.at

ABSTRACT

Summer frost resistance and ice nucleation temperatures for 33 alpine plant species were measured in situ to avoid the shortcomings of laboratory tests. Species were selected to investigate the relationship between plant stature and upper distribution boundary, and frost resistance and freezing patterns. The species tested in situ were on average 1.1 K (± 0.2, SE) frost hardier than in laboratory tests. Frost resistance (LT50) ranged from −4.5 to −14.6 °C and appeared insufficient to protect against air temperature minima, corroborating reports of natural frost damage. All species tolerated extracellular ice formation (recorded at −1.9 ± 0.2 °C; E1). Initial frost damage occurred at average temperatures 4.9 K below E1. In 64% of the species a second exotherm (E2) and frost damage were recorded between −3.7 and −9.4 °C. In the highest ranging species E2 was not detectable. Frost resistance increased with increasing upper distribution boundary (0.4 K per 100 m), corresponding well with the altitudinal decrease in air temperature minima. No relationship between plant stature and frost resistance was found. Graminoids were significantly frost hardier than other growth forms. Frost survival at high altitudes will depend not only on altitudinal increase in frost resistance but also on freezing avoidance strategies, snow cover protection and a high recuperation capacity.

INTRODUCTION

Predictable frost events, as occur during winter, are usually not critical to frost survival as hardening provides sufficient protection against frost damage in most species. Alpine plants, in particular are usually protected from low-temperature extremes by snow cover (see Sakai & Larcher 1987). In winter some fully hardened alpine plants have been shown to survive dipping in liquid nitrogen (Larcher 1980). The situation is different throughout the growing period at higher altitudes when a comparatively low frost resistance coincides with unpredictable freezing temperatures – under such conditions alpine plants may suffer frost damage and lose a substantial fraction of their above-ground tissues (Larcher 1985; Sakai & Larcher 1987; Körner 1999; Taschler, Beikircher & Neuner 2003).

Summer frost resistance (LTI) of alpine plants has already been studied extensively and ranges from −3 to −9.5 °C in dwarf shrubs and herbaceous plants (for reviews see Sakai & Larcher 1987; Körner 1999). In tropical alpine plants two different frost resistance mechanisms are reported, i.e. freezing avoidance by supercooling (Larcher 1975; Goldstein & Meinzer 1983; Goldstein, Rada & Azocar 1985; Rada et al. 1987; Squeo et al. 1991) and freezing tolerance (Beck et al. 1984; Azocar, Rada & Goldstein 1988; Lipp et al. 1994; Rada et al. 2001). Frost resistance mechanisms in temperate alpine plants are less well understood (Sakai & Larcher 1987).

In temperate alpine plants available frost resistance data have been determined on detached plant material. Small tissue volume or detached leaves are known to give erroneous determinations of ice nucleation temperatures (Ashworth, Davis & Anderson 1985; Neuner & Bannister 1995; Neuner, Bannister & Larcher 1997), for example, for the cushion plant Saxifraga caespitosa detachment of leaves decreased ice nucleation temperature by 4.5 K (Robberecht & Junttila 1992). Therefore, measurements on detached leaves are not always accurate predictors of ice nucleation temperatures in intact plants in the field. Increased supercooling in detached leaves, particularly in summer, could also result in artifactual frost resistance data. Higher survival of seedlings at the same low temperature but at increased altitude has been reported (Halloy & Gonzalez 1993) implying that where experimental freezing treatments of alpine plants are conducted may be important and that accurate determination of frost resistance requires in situ measurements. Therefore, our first aim was to determine ice nucleation temperatures and frost resistance in situ in intact alpine plants at their natural growing site.

Our second aim was to investigate the effect of plant growth form on summer frost resistance in temperate alpine plants. In alpine environments during snow free periods in summer it is not clear whether small or tall plants are at greater risk of frost damage because of the complex interaction between convection and radiative cooling (Körner 1999). In open vegetation small plants are often colder at night than tall plants (Wilson et al. 1987; Jordan & Smith 1994) but in the understorey it can be vice-versa. In tropical mountains a vertical stratification of frost resistance was found and ground level plants showed freezing tolerance which was different to arborescent forms exhibiting freezing avoidance (Squeo et al. 1991). These observations indicate that plant height can be important in determining frost resistance and freezing mechanisms.

Selecting all major plant growth forms typical of temperate alpine environments, namely timberline trees, woody dwarf shrubs, graminoids, herbs and cushion plant species also gave a full range of upper distribution boundaries. This allowed us to evaluate the relationship between the upper distribution boundary of a species and the extent of summer frost resistance. We hypothesized that the extent of summer frost resistance would increase with increasing upper distribution boundary in response to decreasing air temperature minima.

MATERIALS AND METHODS

Experimental sites and plant species

In situ freezing experiments were conducted at two altitudinally different investigation sites. The first site was located at the alpine timberline (1950 m a.s.l) on Mt. Patscherkofel near the University of Innsbruck's Alpine Garden laboratory (47°12′ N, 11°27′ E). The second experimental site was situated in the foreland of the Hintertuxer Glacier (2660 m a.s.l.; 47°04′ N, 11°40 E).

The species were selected to include the most important growth forms of the alpine life zone (Körner 1999): timberline trees, woody dwarf shrubs, herbs, cushion plants and graminoids (Table 1). The upper distribution boundaries of the 33 species investigated ranged from 2000 to 3400 m. In situ frost treatments were conducted between June and August of two successive growing periods (2001 and 2002). Frost resistance was measured exclusively on leaves developed during the investigation year. Frost resistance data presented are mean values of leaves in different developmental stages throughout summer.

Table 1.  The alpine plant species investigated differ in their growth form and their upper distribution boundary
Plant growth formSpeciesUpper distribution boundary*Site
  1. Investigations were carried out at two altitudinally different sites: Mt. Patscherkofel (P) at the alpine timberline (1950 ma.s.l) and the Glacier foreland of the Hintertuxer glacier (H) at 2660 ma.s.l. *After Landolt (1992) and Adler, Oswald & Fischer(1994).

TreesAlnus alno-betula2300P
Betula pendula2000P
Larix decidua2400P
Picea abies2100P
Pinus cembra2400P
Sorbus aucuparia2400P
Woody dwarf shrubsCalluna vulgaris2600P
Juniperus communis sps. nana2600P
Loiseleuria procumbens2800P
Rhododendron ferrugineum2500P
Vaccinium gaultherioides2700P
Vaccinium myrtillus2600P
Vaccinium vitis-idaea2700P
HerbsArtemisia genipi3000H
Cerastium uniflorum3400H
Geum reptans3200H
Oxyria digyna2900H
Primula minima2800P
Ranunculus glacialis3200H
Sedum alpestre3000H
Soldanella pusilla2800P
Tanacetum alpinum2800H
Veronica bellidioides2800H
Cushion plantsAndrosace alpina3000H
Saxifraga bryoides3300H
Saxifraga moschata3000H
Saxifraga oppositifolia3200H
Saxifraga paniculata3200P
Silene acaulis2900H
GraminoidsJuncus trifidus3000P
Nardus stricta2800P
Phleum alpinum2600P
Poa alpina3200H

In situ determination of frost resistance and freezing exotherms

Two in situ field freezing systems were used for field freezing experiments. These systems have been developed recently by B & K Elektronik (MCC-6; B & K Elektronik, Natters, Austria) and a detailed description of them is available at http://www.bk-elektronik.com/pages/e_develop.htm. The control unit of the freezing system includes a programmable data-logger (CR10X; Campbell Scientific, Logan, UT, USA), a mains adapter and relays. The data-logger can be programmed to run independent temperature profiles with precise cooling rates, exposure times and thawing rates in each of the six freezing chambers (inner diameter: 11 cm × 11 cm × 15 cm). A ventilator prevents temperature gradients inside the freezing chamber. Air temperature is measured by NTC-sensors and monitored, controlled and recorded by the data-logger. Cooling is provided by Peltier-units that are turned on or off when the recorded temperature deviates from the pre-programmed target temperature. In each freezing chamber leaf temperatures were also recorded with six fine wire copper-constantan thermocouples (see below).

The bottom of the freezing chamber is open. During measurements of trees and dwarf shrubs it was closed by a special lid through which twigs could be inserted. Alpine cushion plants, herbs and graminoids were enclosed by putting the freezing chamber upside down over intact plants. The contact area between the soil surface and the freezing chamber wall was thermally insulated with rubber foam. During the freezing treatment the freezing chambers were pinned down to hold them in a fixed position.

The most significant results, ecologically, are obtained by cooling that imitates naturally occurring frosts (see Sakai & Larcher 1987). In field freezing experiments the leaf temperature was lowered at a constant rate of 2 K h−1 into the subzero temperature range as would naturally occur. Samples then remained at target temperatures for 4 h. Constant thawing rates (2 K h−1) were also employed as thawing rates may also influence frost survival (Gross et al. 1991).

To detect freezing exotherms leaf temperatures were measured with fine-wire copper-constantan thermocouples (welding spot diameter: 0.15 mm) during the controlled in situ freezing at 2 K h−1. Thermocouples were positioned with thermally insulated leaf clips to attach the welding spot closely to the leaf surface. During each in situ freezing treatment 36 leaf temperatures (six in each freezing chamber) were recorded on a CR10X data-logger. Ice nucleation temperatures were determined graphically as the starting point of the sudden leaf temperature increase at the beginning of freezing exotherms. As two freezing systems were available 72 leaf temperature records were obtained during each in situ freezing test. For each of the 33 species an average of 12 freezing runs were carried out during the main growing periods of 2001 and 2002.

Viability assay

Frost damage was assessed in situ 5 d after the frost treatment. As leaves of most of the tested alpine species showed either no initial damage or were completely killed, the percentage of the leaf area frost damaged could easily be assessed visually at the investigation site. In the unusual event that the percentage damaged was not clear-cut frost damage was documented by digital photographs and further analysed using image analysis software (optimas 6.5; see Neuner & Buchner 1999).

LT0 is the lowest treatment temperature sustained without frost damage. LTi is the temperature where incipient frost damage could be observed. LT100 is the highest treatment temperature causing 100% frost damage. To calculate the temperature at 50% frost damage (LT50), the percentage damage was plotted against the treatment temperature for each replicate separately (n = 6). A classic logistic function was then fitted to the viability data using P.Fit software (Biosoft, Ferguson, MO, USA). LT50 was calculated as a parameter of the logistic function and read from the curve-fitting protocol.

Statistical data analysis

Frost resistance and ice nucleation temperatures were usually determined on leaves of more than 20 individuals but for trees six individuals, chosen at random, were sampled. After the data had passed the Kolmogorov–Smirnov test the statistical significance of differences between means of frost resistance (LTi−100) and ice nucleation temperatures were tested by analysis of variance and the Duncan's multiple range test (P < 0.01) using SPSS software (SPSS Inc., Chicago, IL, USA).

RESULTS

In situ frost resistance

Summer frost resistance (LT50) in leaves of alpine species ranged from −4.5 (Vaccinium myrtillus) to −14.6 °C (Poa alpina) (Fig. 1), indicating a wide range for frost resistance within the investigated species. The foliar frost resistances of the woody dwarf shrubs tested did not differ significantly (P < 0.01). This was also the case for the graminoids (Table 2). In contrast, within trees, herbs and cushion plants two significantly different homogeneous subsets with respect to frost resistance were detected (tested by anova and the Bonferroni test). The more frost-sensitive subsets within trees and herbs were the most frost sensitive of the species investigated. Graminoids were the most frost-tolerant group.

Figure 1.

Mean summer frost resistance measured in situ in leaves of various alpine plant species from June to August over two successive growing periods (2001 and 2002) at two alpine investigation sites (1950 and 2660 m a.s.l). Mean frost resistance values for each plant growth form group are shown or when within a plant growth form group significantly different subsets with respect to frost resistance were detectable by anova and the Bonferroni test (P < 0.01), mean values for these subsets are shown separately. Bars range from the lowest temperature experienced without sustaining frost damage (LT0) to the highest temperature with 100% frost damage (LT100). LTi and LT50 are also shown. The significance of differences between mean values of frost resistance was tested with Students t-test (P < 0.01) and is indicated by different capital letters. The mean sample size per plant species was 30.

Table 2.  The significance of differences between mean frost resistances (LTi ± SE) of the investigated species within each plant growth form group were tested by anova and the Bonferroni test (P < 0.01). Within trees, herbs and cushion plants two significantly different homogenous subsets with respect to frost resistance were detectable (subsets a and b)
Plant growth formSpecies(LTi) Mean ± SESubset
TreesAlnus alno-betula4.8 ± 0.5a
Betula pendula4.1 ± 0.5a
Larix decidua8.0 ± 0.6b
Picea abies5.1 ± 0.3a
Pinus cembra6.7 ± 0.5b
Sorbus aucuparia6.8 ± 0.5b
Woody dwarf shrubsCalluna vulgaris8.3 ± 0.1a
Juniperus communis sps. nana9.0 ± 0.6a
Loiseleuria procumbens6.6 ± 0.6a
Rhododendron ferrugineum4.7 ± 0.5a
Vaccinium gaultherioides5.6 ± 0.5a
Vaccinium myrtillus4.1 ± 0.4a
Vaccinium vitis-idaea5.5 ± 0.3a
HerbsArtemisia genipi6.9 ± 0.4b
Cerastium uniflorum8.5 ± 0.3b
Geum reptans7.1 ± 0.2b
Oxyria digyna4.5 ± 0.3a
Primula minima4.6 ± 0.3a
Ranunculus glacialis7.9 ± 0.1b
Sedum alpestre6.6 ± 0.1b
Soldanella pusilla5.7 ± 0.4a
Tanacetum alpinum6.9 ± 0.1b
Veronica bellidioides5.0 ± 0.4a
Cushion plantsAndrosace alpina8.6 ± 0.2b
Saxifraga bryoides8.5 ± 0.2b
Saxifraga moschata6.0 ± 0.1a
Saxifraga oppositifolia7.8 ± 0.2b
Saxifraga paniculata5.8 ± 0.2a
Silene acaulis6.0 ± 0.3a
GraminoidsJuncus trifidus7.7 ± 0.3a
Nardus stricta10.3 ± 0.6a
Phleum alpinum10.8 ± 1.5a
Poa alpina−9.9 ± 1.4a

Species differed in the temperature span over which frost damage occurred (LT0–LT100). This span varied from 0.9 to 7.8 K. While a narrow temperature span indicates a sudden lethal freezing event, a wider span indicates gradually occurring frost damage. In the frost-sensitive alpine herb group (subset a) the span was only 2.3 K. The widest temperature span was found in graminoids (on average 7.0 K).

Summer frost resistance of leaves measured in situ was on average 1.1 K (± 0.2, SE) higher (frost hardier) than frost resistance reported earlier by various other authors (Table 3). Of the 21 species for which earlier laboratory freezing experiments were reported 70% were more than 0.5 K frost hardier when tested in situ, in 12% the difference was less than ± 0.5 K and only 18% of the species were frost hardier in the laboratory freezing tests.

Table 3.  Summer frost resistance [°C] of leaves of the investigated alpine plant species measured by various authors were compared with results of our in situ field freezing experiments from June to August 2001 and 2002
SpeciesFrost resistance from various authorsΔ to in situ min ΔLT/max ΔLT
LT0LTiLT50LT100
  1. +ΔLT means that a species was frost hardier in the in situ experiments. LT0 is the lowest treatment temperature sustained without frost damage, at LTi– temperatures incipient frost damage could be observed, LT50 is the temperature at 50% frost damage and LT100 is the highest temperature causing 100% tissue death. MinLTx and MaxLTx are separated by (/). aPisek & Schiessl (1947); bPisek, Larcher & Unterholzner (1967); cKemnitzer (cited in Pisek et al. 1967), dMoser (1965); eSchwarz (1970); fPack (cited in Pisek et al. 1967), gKainmüller (1975); hLarcher & Wagner (1976); iTranquillini (1979); jLarcher & Bauer (1981); kChristersson, von Fircks & Sihe (1987); lJuntilla & Robberecht (1993); mRepo (1992); nKörner (1999), oNeuner et al. (1999); pHacker, unpublished; qBannister & Polwart (2001).

Alnus alno-betula −3b/−3.5c,b −5c1.3/1.8
Betula pendula −3.5b −5b0.6/2.2
Calluna vulgaris−4j/−5j −6q 1.9/2.9
Geum reptans −4d −6d3.1/5.6
Larix decidua −4.5f,b/−5.0k,b −7f,b2.5/3.5
Loiseleuria procumbens−5j/−6j,h  −8h0/0.6
Nardus stricta −9.5n  0.8
Oxyria digyna −6d −7d−1/−1.5
Picea abies −3.0k/−3.5m  1.1/1.3
Pinus cembra−2.0i/−3.5a,g   2.4/3.9
Poa alpina −8.5n−9.5n 1.4/5.1
Primula minima−3h  −4h0.6/1.7
Ranunculus glacialis−7h−6d −7.5d0.5/1.9
Rhododendron ferrugineum −3.8o/−4e  0.7/0.9
Saxifraga oppositifolia−4h/−5h,g  −8h/−10g0.4/2.8
Saxifraga paniculata−5.9p−7.1p−8.5p−10.7p−1.3/−2
Silene acaulis−5g/−6l −7l/−9l−10g−3/0.1
Soldanella pusilla−4h  −6h0/1
Vaccinium gaultherioides−4j/−6j   −2.3/−0.3
Vaccinium myrtillus−4j−3b −5.5b−1.4/0.5
Vaccinium vitis-idaea5j−3.5b −5.5b−1.2/1.9

In situ freezing pattern

Two different freezing patterns were observed. Typical examples of each are depicted in Fig. 2. In 64% of the investigated species, such as Vaccinium gaultherioides (Fig. 2a–f) two distinct intrinsic freezing events were recorded. For the other 36%, such as Silene acaulis (Fig. 2g–l) only one intrinsic freezing event was recorded. The first intrinsic freezing event (E1) was found in all species at an average temperature of −1.9 °C (± 0.2 SE). In the two highest  ranging  species  groups mean  values  were  lower (−2.6 °C, −3.5 °C), however, the difference was only significant at P < 0.05. E1 was in all cases non-injurious even when, following the occurrence of E1, plants were exposed to temperatures slightly above LTi for more than 4 h (see Fig. 2a & b). Extracellular ice formed in the leaves after the occurrence of E1 as indicated by infiltrations after thawing. Thus, all investigated species tolerated extracellular ice formation to some extent. When a second freezing exotherm (E2) was recorded it always coincided with frost damage (see Fig. 2c–f).  The  E2  values  were  recorded  at  between −3.7 and −9.4 °C depending on the species. After rainfall, when leaves were wet a small extrinsic exotherm (E0), namely the freezing of water at the leaf surface, at −0.2 °C (± 0.1, SE) was often recorded.

Figure 2.

Leaf temperatures recorded during controlled in situ frost treatments on leaves of V. gaultherioides (a–f) and S. acaulis (g–l). Six different exposure temperatures were usually employed to measure frost resistance in situ . Mean summer frost resistance of the species (LT0 − LT100) is indicated by dotted lines. In all species intrinsic ice nucleation (E1) was recorded at significantly (P < 0.01) higher temperatures than initial frost damage. In 64% of the investigated species (as shown by V. gaultherioides) a second exotherm (E2) at lower temperatures matching those at which frost damage occurred was registered. The number in parenthesis indicates resultant percentage frost damage.

Plant growth forms with the highest upper distribution boundary, namely cushion plants and the less frost-susceptible herbs (subset b, see Table 2) suffered frost damage without the occurrence of a second freezing exotherm (Fig. 3). In leaves of trees, woody dwarf shrubs, graminoids and frost-susceptible herbs (subset a) E2 was always recorded and occurred at the same temperature (P < 0.01) as frost damage. In trees and woody dwarf shrubs that have a comparatively low upper distribution boundary, E2 coincided with 50% frost damage. In frost-susceptible herbs (a) and graminoids E2 occurred at the same temperature as initial frost damage.

Figure 3.

Patterns of in situ freezing and summer frost resistance (LT0−100) for the investigated alpine plant growth form groups including their subsets (see Table 2). The different groups were arranged according to their mean upper distribution boundary (± SE; •) as shown in the bottom half of the figure. The first exotherm (E1, ▿) is connected with a solid line to the second exotherm (E2, ▴), which occurred at a significantly lower temperature. The mean sample size per plant growth form was 100.

Trees, herbs and cushion plants exhibiting a higher frost resistance (subset b) had a significantly higher upper distribution boundary than those with a lower frost resistance (subset a). However, the herb group (a) is as frost susceptible as tree group (a) yet the herb group (a) extends up to 2900 m whereas the tree group (a) is limited to 2300 m. Graminoids have the highest frost resistance but they are not the highest ranging plant growth form.

When frost resistance is plotted against a species upper distribution boundary (Fig. 4), frost hardiness tends to increase with increasing upper distribution boundary. The mean increase in frost resistance (LT50) with increasing upper distribution boundary was 0.4 K per 100 m. Strikingly this corresponds well with the altitudinal decrease in absolute air temperature minima (0.4 K per 100 m). However, it should be noted that this tendency may not hold true for individual species, particularly, within the graminoid group in which species upper distribution boundaries range from 2400 to 3000 m despite their consistently high frost resistance.

Figure 4.

Means of summer frost resistance (LT0 − LT100) for alpine plant species with similar upper distribution boundaries (200 m altitude classes; 2.0 = 1.901–2.100 km). Absolute air temperature minima were recorded during 2001 and 2002 in June (•) and in July and August (○) at different altitudes between 1938 and 3400 m. Longer term (8–40 years) absolute air temperature minima for June (▾) and for July and August (▿) are also shown. Temperature data were provided by ZAMG Austria and derive from various climate stations [Obergurgl 1938 m (40 years), Galzig 2081 m, Patscherkofel 2247 m (30 years), Ischgl-Idalpe 2323 m, Pitztaler Gletscher 2850 m (8 years); Sonnblick 3105 m (40 years), Brunnenkogel 3400 m].

Absolute air temperature minima during the growing period are low enough to explain the occurrence of some frost damage. Radiative cooling during clear nights is more pronounced at high altitudes and may expose alpine plants to leaf temperatures distinctly lower than air temperature (up to 8 K; see Jordan & Smith 1994). This means that the risk of frost damage may be even higher. Particularly during June, snow-free plants may suffer severe frost damage above 2000 m within the alpine zone. Although at the timberline on Mt. Patscherkofel plants are snow free from about the end of May onwards, at higher altitudes most species may still be protected by snow cover. During July and August, at altitudes lower than about 2500 m, summer frost damage is unlikely to occur as absolute air temperature minima are higher than frost resistances. In contrast, above about 2500 m most of the investigated species may suffer frost damage even in July and August, which is the main growing period at high altitudes. In these 2 months the absolute air temperature minimum (40 years) recorded at 3105 m (ZAMG, Sonnblick) was −9.8 °C, which is below the LT0 of all investigated species. In 71% of the species ranging higher than 3105 m this is even lower than LT50.

DISCUSSION

In situ frost resistance

In our field freezing experiments the investigated alpine plant species were on average 1.1 K frost hardier than reported by various other authors (see Table 3) in earlier laboratory studies using detached plant material. Artifactual supercooling noticed in detached plant parts and its effect on frost damage (Ashworth et al. 1985; Sakai & Larcher 1987; Robberecht & Junttila 1992; Flinn & Ashworth 1994; Neuner et al. 1997; Pearce 2001) could be one explanation for the differences observed. Direct comparisons of frost resistance determined on detached and attached twigs of Picea abies resulted in a similar 1.2 K difference (Taschler et al. 2003). On the other hand our results could also corroborate the findings of Halloy & Gonzalez (1993) who reported higher seedling survival at higher altitudes than at lower altitudes. Our results further suggest that given a low frost-hardening capacity in alpine summers there may be an insufficient safety margin between minimum temperatures and actual frost resistance. This is consistent with repeated observations of frost damage to alpine plants in summer (Larcher 1981; Sakai & Larcher 1987; Körner 1999; Taschler et al. 2003). The observed 1.1 K difference in the frost hardiness results must therefore be considered ecologically highly significant and important in estimating the future risk, frequency and severity of frost damage in the alpine life zone during summer.

Species could escape frost damage by frost hardening or frost tolerance. With the exception of the early observations of Tyurina (1957, cited in Sakai & Larcher 1987) who found frost hardening by 1–5 K at temperatures slightly above 0 °C within 1 d in high mountain plants of East Pamir, little is known about the frost-hardening capacity of alpine plants during summer.

Leaves of all investigated species tolerate at least some extracellular ice formation as indicated by the absence of frost damage in infiltrated leaves removed from the frost chambers at temperatures below E1. Only if temperatures dropped by on average 4.9 K below this first intrinsic freezing event (E1) did initial frost damage occur. E1 was found in all species at an average temperature of −1.9 °C (± 0.2, SE), corroborating earlier in situ observations (from −0.6 to −2.6 °C; for review see Pearce 2001). This is in the line with the suggestions of Sakai & Larcher (1987) that in areas with frosts of below −10 °C during the growing season, at least certain plants must be capable of tolerating extracellular ice formation even when metabolically active. This is the only possible explanation for the relatively high summer frost resistance (LT0: −7.5/−16.0 °C) of arid mountain plants in East Pamir (Tyurina cited in Sakai & Larcher 1987) and for temperate alpine plants being found frozen stiff in the morning, but entirely undamaged after thawing (Moser 1965; Körner 1999).

Plant growth form and frost survival

There was no correlation between plant growth form and frost resistance per se; for example, some timberline trees (subset a, see Table 2) were among the most frost-sensitive species whereas other trees (subset b) were among the most frost hardy. In low-stature plants excluding graminoids frost resistance (LT50) ranged from −4.5 to −10.5 °C. Graminoids were the most frost-resistant species group (from −8.6 to −14.6 °C). This corroborates the reported tendency for graminoids to resist lower temperatures better than broad-leaved species (Poa alpina: LTi: −9.5 °C; Körner 1999 and Stipa glareosa: LT0: −16.0 °C cited in Sakai & Larcher 1987).

Frost survival, however, is more complex than frost resistance as frost survival will be co-determined by the frost avoidance and recuperation properties of a species. The position of the regenerating buds is the best known and most obvious means of avoiding low-temperature extremes (Larcher 1985). In many alpine plants, namely sessile rosettes, graminoids and some dwarf shrubs, vegetative shoot apices, leaf primordial tissues and premature reproductive organs are buried several centimetres below ground, and thus are not exposed to otherwise deleterious freezing temperatures (Körner 1999). In cushion plants delayed night time cooling due to the heat capacity of moisture inside the cushion also helps to avoid critical freezing temperatures (Sakai & Larcher 1987). Trees and taller dwarf shrubs evidently lack such a morphological frost avoidance mechanism yet some trees (subset b) were among the frost-hardiest species groups, exceeded only by graminoids and cushion plants (subset b). Little is known about the frost resistance of seedlings of alpine plants that lack the freezing avoidance mechanism described above. Comparable data are only available for Loiseleuria procumbens. This data suggests that seedlings have similar frost resistance (LT50) to adults (M.Th. Wildner-Eccher cited in Sakai & Larcher 1987). However, LT0 values were 1–2 K higher for seedlings, indicating that they could still be a critical developmental stage in determining frost survival.

Repair and replacement of vegetative tissue losses after frost damage has rarely been investigated but is likely to be affected by seasonal timing, the severity of frost damage and the existence of undamaged regeneration buds and sufficient storage reserves (Körner 1999). The root system contains substantial carbohydrate and lipid reserves (Körner 1999), is protected from frost damage and is commonly larger in alpine plants than lowland plants (Körner & Renhardt 1987). Repeated loss of much of the annual leaf crop by browsing did no obvious long-term harm to some high altitude plants (Diemer 1996) as recuperation was made possible by root resources. Preliminary results of a recuperation experiment at 2660 m after frost show an extraordinary regeneration capacity particularly in the frost-sensitive herb group (subset a). Oxyria digyna showed an extremely fast recuperation of a functioning structure after complete loss of the above-ground tissues despite unfavourable environmental conditions during the recovery period. In contrast after losing this year's growth coniferous trees do not replace foliage again until the next growing season.

Altitude and frost resistance

Plotting frost resistance against the upper distribution boundary revealed a significant increase of 5.1 K in frost resistance (LT50) with increasing upper distribution boundary. Species with the same growth form but a higher frost resistance (subset b) had a significantly higher upper distribution boundary than those with a lower frost resistance (subset a). The highest ranging species, namely cushion plants and frost-hardy herbs (subset b), showed a different freezing pattern in situ to that of the other species groups. In these high-ranging species no E2 was found. For the other species (64% of those investigated) E2s indicated a lethal freezing process. A second exotherm at temperatures slightly higher than initial frost damage was also recorded in ground-level plants of tropical mountains (Squeo et al. 1991). E2 values usually mark freezing of damaged protoplasts with the initial target for frost damage likely to be the plasma membrane (Pearce & Willison 1985) or homogenous ice nucleation after supercooling (Olien 1978, 1981). We hypothesize that plasma membranes of the high-ranging species remain intact and there is a different initial target for frost damage.

The increase in frost resistance with altitude does not fully protect alpine plants against frost damage as at altitudes above 2500 m absolute air temperature minima lower than frost resistance are possible throughout summer. Occasional frost damage is likely during alpine summer and has been repeatedly observed. Frost resistance is not the only factor in determining frost survival as, in addition, this will depend on frost avoidance strategies, potential snow cover during cold spells and the ability to recuperate from frost damage. These additional factors will increase in importance with increasing altitude. In alpine environments, as stated earlier, frost damage may not affect the survival of the species at a population level but is likely to alter the contribution of the damaged species to cover and biomass production (Körner & Larcher 1988).

ACKNOWLEDGMENTS

This study was financially supported by the FWF (project P14524-Bot) and a grant from the ‘Verein zur Förderung von Südtirolern an der Landesuniversität Innsbruck’. We further wish to thank D. Buchnor from B&K Electronik, the Hintertuxer Gletscherbahnen, J. Klausner from the Spannagelhaus and P. Schröcksnadel (Patscherkofel Bergbahnen) for free transportation and the Hintertuxer Gletscherbahnen additionally for free electricity supply at 2660 m.

Ancillary