Elevated atmospheric [CO2] promotes frost damage in evergreen tree seedlings


J.L. Lutze E-mail: lutze@rsbs.anu.edu.au


Growth under elevated [CO2] promoted spring frost damage in field grown seedlings of snow gum (Eucalyptus pauciflora Sieb. ex Spreng.), one of the most frost tolerant of eucalypts. Freezing began in the leaf midvein, consistent with it being a major site of frost damage under field conditions. The average ice nucleation temperature was higher in leaves grown under elevated [CO2] (– 5·7 °C versus – 4·3 °C), consistent with the greater incidence of frost damage in these leaves (34% versus 68% of leaves damaged). These results have major implications for agriculture, forestry and vegetation dynamics, as an increase in frost susceptibility may reduce potential gains in productivity from CO2 fertilization and may affect predictions of vegetation change based on increasing temperature.


Atmospheric CO2 concentrations have been increasing since the industrial revolution, and are predicted to reach twice present levels late next century (Boden et al. 1994). It is thought that temperature will increase as [CO2] rises, and that warming may be greater in winter than in summer (Plummer, Lin & Torok 1996). This could lengthen the growing season if evergreen species were able to take advantage of warm conditions earlier in spring when below-ground resources are relatively abundant. However, predictions of plant responses to global climate and atmospheric change are complicated by weather variability (Katz & Brown 1992) and the extent to which elevated [CO2] might affect plant responses to temperature (Long 1991), particularly low temperatures. Some studies have predicted that frost damage may increase for trees that break dormancy too early in spring (Cannell & Smith 1986; Repo, Hänninen & Kellomäki 1996). Other studies have found enhanced performance of chilling sensitive plants exposed to low temperatures under elevated [CO2] (Sionit et al. 1981; Potvin 1985; Boese, Wolfe & Melkonian 1997). Clearly, there is no consensus of how elevated [CO2] might affect plant responses to changes in spring temperatures.

Here we report the results of a serendipitous experiment. The experiment naturally occurred during a larger field-based study aimed at understanding the effects of elevated [CO2] on the interactions between grass and trees during spring. The results were unexpected and showed that one of the most frost-hardy of broadleaved, evergreen species suffered greater frost damage when grown under elevated than ambient [CO2].


Plant material and growth conditions

Seeds of Eucalyptus pauciflora Sieb. ex Spreng. were collected from three trees growing along the floor of the Orroral Valley at an elevation of 850 m in New South Wales, Australia. The seeds were cold stratified under moist conditions at 3 °C for 4 weeks before germinating on sand flats in a mist house. Seedlings of similar size were transferred to individual containers (5 cm diameter, 25 cm deep) and grown out of doors for 6 months before the start of the experiment.

In a pasture near Bungendore in southeastern Australia (35° 15’ S, 149° 27’ E; elevation 700 m), 10 open-top chambers were installed as five replicate pairs, flushed with air containing either ambient or elevated CO2 concentrations (350 or 700 cm3 m–3, respectively) during daylight hours. The chambers were essentially as described by Ashenden, Baxter & Rafarel (1992), with two modifications to prevent excessive heating on sunny days. The roof was removed and air lines inside the chambers were painted white, while those outside the chambers were covered with insulating foam topped with aluminium foil. These two changes were sufficient to maintain daytime air temperatures in the zone of plant growth close to those outside the chambers (Fig. 1b). Air temperatures were monitored throughout the study at seedling canopy height (15 cm above ground) in one pair of chambers and in an adjacent unchambered control. Temperatures were measured with copper/constantan thermocouples (42 swg, 64 μm diameter) referenced against a platinum resistance thermometer. Temperatures were recorded on a datalogger every 15 min as the average of measurements made every 10 s during the sampling period. The pasture within each chamber was killed with glyphosate. Two cold-hardened seedlings were transplanted into the dead pasture within each chamber on 8 July 1996. Approximately 1 dm3 of water was supplied to each seedling at transplanting.

Figure 1.

. (a) Daily maximum and minimum air temperatures at seedling height from late winter to early spring. The bar shows the 23 d period with an average minimum temperature of + 2 °C, and the arrow shows the frost on 26 September 1996 which caused a large amount of damage. The diurnal variation in air temperature for that day inside (ambient, Amb; enriched, Enr) and outside (Out) of the chambers is shown in (b). The mid-spring frost on 26 September 1996 resulted in damage visible as necrotic or heavily pigmented areas concentrated around veins (c).

Visual assessment of frost damage

A visual estimate of frost damage was made on 9 October 1996, after the first appearance of visual damage. The proportion of leaves damaged was calculated for each plant, and statistical analysis undertaken on that proportion.

Determination of freezing exotherms

An infrared camera and thermal imaging system (Agema Thermovision 800, Agema Infrared Systems, Danderyd, Sweden) were used to produce freezing exotherms (Wisniewski, Lindow & Ashworth 1997) in whole, detached leaves, cooled at 8 °C h–1 on 24 October 1996. The cooling chamber consisted of a hollow, cylindrical tower made of styrofoam with internal dimensions of 19·5 × 86·5 cm. The lower half of the chamber wall was lined with copper tubing through which coolant flowed from the middle to the bottom of the tower. Coolant also flowed through a flat, disk-shaped aluminium radiator at the bottom of the tower. The camera was mounted at the top of the tower and leaves were suspended on Nylon rigging spanning the width of the cylinder, 6·5 cm above the radiator and 80 cm below the camera lens. The leaves were held flat, with the blade parallel to the lower radiator. Once leaves were installed, the tower could be flushed with dry air from a compressed air cylinder to minimize frost formation. No frost was found on leaf surfaces during measurements. Cooling was provided with a controlled heating and refrigerated circulator (B. Braun Thermomix, Melsungen, Germany).

Images with pixel resolution 0·8 mm2 were acquired every 6 s. Leaf emissivity was determined (Fuchs & Tanner 1966), and an average value of 0·98 was used in calculations of surface temperature. Exotherms were measured concurrently on a pair of leaves, one from each of the two [CO2] treatments. Measurements were repeated on five pairs of leaves, one from each of the five chamber pairs. The ice nucleation temperature was determined as the average of values obtained from five pixels on the midvein and five pixels on the lamina for each leaf. Leaves were matched for size and height above ground and bore no signs of necrosis from frost damage. Preliminary experiments showed no differences in freezing characteristics between detached and attached leaves from the same treatment.

Statistical analyses

The experiment was organized in a split plot randomized block design with five replicates of each CO2 treatment. CO2 enrichment was assigned randomly to one chamber in each of the five chamber pairs, with one chamber pair per block. The results were evaluated by analysis of variance (Genstat 5 Committee 1993).


Winter and spring produced frequent frosts with warm daytime temperatures at seedling canopy height (Fig. 1a). Air temperatures inside the open-top chambers were similar to those outside the chambers during the day, but the minimum temperatures at night were as much as 1·5 °C higher inside the chambers (Fig. 1b). These conditions are similar to many climate predictions (Whetton, Mullan & Pittock 1996) and recent weather patterns (Plummer et al. 1996) showing increases in temperature minima with smaller increases in maxima.

Hardened snow gum seedlings were planted in open-top chambers in early winter, and overwintered with little visible frost damage, even following temperature minima as low as – 14 °C at seedling height. This species is one of the most freeze tolerant of eucalypts (Sakai, Paton & Wardle 1981) and withstands temperatures as low as – 18 °C when fully cold hardened (Harwood 1981). In mid-spring, there was a 23 d period when the minimum temperature averaged + 2 °C at seedling height and decreased below zero to – 2 °C on one night. This warm period was followed on 26 September 1996 by a frost of – 5 °C (Fig. 1a). On that night, air temperatures inside the chambers declined to – 3·5 °C (Fig. 1b) and leaf temperatures would have been at least a degree lower than the air temperature (Leuning & Cremer 1988; Ball et al. 1997). This frost, which occurred before bud break, resulted in damage to a large number of mature leaves. Frost damage occurred in 68 ± 8% and 37 ± 7% of leaves per plant under enriched and ambient [CO2] treatments, respectively. Of the damaged leaves, the effects were visible over 13 ± 2% and 8 ± 1% of the surface area and on 50 and 25% of the midveins in leaves grown under enriched and ambient [CO2] treatments, respectively. Thus, plants grown under elevated [CO2] suffered significantly (P < 0·001) greater frost damage, with the most visible damage concentrated around veins (Fig. 1c).

Freezing exotherms were measured by infrared video thermography as shown in Fig. 2. Conditions in the freezing chamber produced relatively even cooling over leaf surfaces as shown for one pair of leaves from the ambient and enriched [CO2] treatments (Fig. 2a). Ice nucleation occurred at higher temperatures in leaves grown under elevated [CO2]. Freezing typically began in the leaf midvein, consistent with it being a major site of frost damage under field conditions (Fig. 1c), and spread through the leaf lamina (Fig. 2b, d & e). The more rapid rise in temperature in the ambient leaf upon freezing, followed by a return to temperatures similar to those of the enriched leaf, is consistent with a greater degree of supercooling which produced more rapid freezing of a similar fraction of the local tissue water (Fig. 2c & f). There were no significant differences in ice nucleation temperature between midvein and lamina regions of leaves, so ice nucleation temperatures for individual leaves were calculated as the average of ice nucleation temperatures determined from 10 pixels regularly spaced over each leaf surface.

Figure 2.

. Determination of ice nucleation temperatures by infrared video thermography. The distribution of temperatures over the surfaces of leaves from ambient (Amb) and enriched (Enr) [CO2] treatments is shown for four times during the course of cooling. Several seconds before freezing began in the Enr leaf, temperatures averaged – 3·8 ± 0·4 °C and – 3·5 ± 0·4 °C in the Amb and Enr leaves, respectively (a). Six seconds later (b), an increase in surface temperatures showed that freezing had begun in much of the Enr leaf. For several minutes following nucleation, temperatures remained elevated and varied little with time over much of the Enr leaf. This effect is due to the latent heat liberated by the freezing of water, so regions of relatively high (freezing) temperatures indicate regions of ice formation. On the other hand, temperatures in the Amb leaf continued to decrease approximately uniformly, indicating little or no ice formation at this temperature. Thus, 4·6 min after freezing began in the Enr leaf, surface temperatures averaged – 4·7 ± 0·5 °C and – 2·9 ± 0·5 °C in the Amb and Enr leaves, respectively (d). Six seconds later (e) an increase in temperatures indicated that freezing had begun in the midvein area of the Amb leaf, but had not yet spread to the lamina. The surface temperature for one representative pixel in the centre of the Enr and Amb leaves is shown as a function of time in (c) and (f), respectively. Arrows indicate the onset of the increases in leaf temperature due to ice nucleation. Temperatures of nucleation are indicated by dotted lines.

The average ice nucleation temperature was 1·4 °C higher (P < 0·02) in leaves grown under elevated than under ambient [CO2] (– 4·3 °C versus – 5·7 °C, respectively) (Fig. 3b). Given these temperature differences, ice nucleation would have occurred in a greater proportion of leaves growing under elevated [CO2] during the night of 26 September 1996. Indeed, a significantly (Fig. 3a; P < 0·001) greater proportion of leaves were frost damaged in plants grown under elevated [CO2]. Thus, freezing damage was consistent with a greater propensity for ice to form in leaves grown under elevated [CO2]. The basis of this effect is unknown. One possibility is that the phyllosphere of plants grown under elevated [CO2] may support a greater population of ice nucleating bacteria. Whatever the mechanism, a higher ice nucleation temperature may explain the significantly (P < 0·001) greater incidence of frost damage in seedlings grown under elevated [CO2] (Fig. 3a).

Figure 3.

. The effect of [CO2] treatments on (a) frost damage per plant in the open-top chambers, and (b) susceptibility to freezing in the laboratory. Values are mean ± standard error, n = 5. The effect of [CO2] was statistically significant.


While there are uncertainties in predictions of climate change, the increase in [CO2] is beyond debate. The results of the present study show that a doubling in [CO2] predisposes plants to freezing injury. Thus, in a world with high [CO2], a 2 °C increase in temperature minima will not necessarily decrease the incidence of frost damage. Given the importance of freezing temperatures in determining tree species distributions (Davidson & Reid 1985; Woodward 1987) and interactions with grass (Ball et al. 1997), these results have far reaching implications for forest/grassland dynamics in areas subject to frost.


We thank Tim Hobbs for permission to conduct the experiments on his property, Vince Gutschick, Jay Ham and Christian Körner for technical advice to minimize temperature differentials between the air inside and outside of the chambers, Godfried Ashenberger for the construction of open-top chambers, Istvan Zaveczky for the construction of a freezing apparatus, Dennis Vukoja for computer programming, and Eldon Ball, Tara Goodsell and Jeff Wilson for the preparation of the figures.