In situ growth stimulation of a temperate zone liana (Hedera helix) in elevated CO2


†Author to whom correspondence should be addressed. E-mail:


  • 1Lianas, which arguably benefit more than other growth forms from elevated CO2, have been associated with increasing turn-over rates in tropical forests as observed in recent decades. Although rarely as prominent outside the tropics, an increase in abundance of climbing plants is likely to affect forest dynamics in the temperate zone as well. Hedera helix is the most abundant climber in western Europe and previous work in controlled conditions suggested CO2 effects similar to those observed in tropical lianas.
  • 2Here we present an in situ test of the hypothesis that this abundant climber will benefit from increasing CO2 concentrations primarily under light limitation in the forest understorey, but much less in the forest canopy. To this end, we studied growth responses to elevated CO2 for an entire growing season at the Swiss Canopy Crane site on the forest floor and 20–25 m above the ground in the forest canopy.
  • 3The relative stimulation of length and biomass increment of shoots by elevated CO2 (about 600 µl l−1) was indeed very pronounced in deep shade (c. +60%), about twice as much as in the subcanopy (c. +30%). Given the rapid depletion of non-structural carbohydrate (NSC) pools at canopy closure in spring, carbon limitation under current ambient CO2 concentrations must be substantial in the understorey. In contrast to the understorey, where CO2 enrichment had no effect on NSC pools, there was a sharp increase in NSC concentration in subcanopy leaves exposed to elevated CO2.
  • 4We conclude that rising CO2 concentrations will allow Hedera to explore light-limited understorey microhabitats more vigorously, which should increase the likelihood of reaching the forest canopy. There, however, Hedera benefits less from elevated CO2. Our results add one more explanation to Hedera's current success in European forests.


Repeated large-scale surveys in tropical forests suggest increasing turn-over rates in recent decades, which have been causally linked to an increasing dominance of lianas (Phillips et al. 2002; Wright et al. 2004). These structurally dependent plants are known to increase tree mortality and to suppress tree growth (Putz 1984; Stevens 1987; Schnitzer & Bongers 2002). Increased turnover will most likely result in a reduction of carbon sink activity of tropical ecosystems, because early successional tree species commonly store less carbon per unit land area than late successional species (Körner 2004). While generally not as prominent a feature as in the tropics, lianas can also play an important role in temperate forests, particularly in gallery forests (Schnitzler 1995) and in moist, oceanic settings (Dawson 1988; Kelly 1981). English Ivy (Hedera helix L., Araliaceae) is a common root-climbing liana native to Europe, western Asia and northern Africa. Its potential to impact virtually all strata of forested areas is particularly well documented in North America where it has been introduced, originally as an ornamental, and is now regarded a serious threat to native vegetation (Swearingen & Diedrich 2004). This aggressive invader competes with tree foliage for light, but also makes infested trees more susceptible to blow-over owing to the added weight and increased wind resistance (compare Siccama, Weir & Wallace 1976). In the understorey, where ivy frequently forms dense carpets, it may inhibit the regeneration of co-occurring native herbs or tree seedlings.

The current rise in CO2 concentrations may increase the potentially detrimental effects of this and other temperate liana species in their native ranges as well. It is well established both theoretically and experimentally that deep shade does not preclude elevated CO2 from stimulating photosynthesis. In fact, the relative effect of increased CO2 concentrations on growth is greatest as one approaches severe light limitation, owing to a reduction in photorespiration, which in turn lowers the light compensation point of photosynthetic carbon uptake (Long & Drake 1991; Hättenschwiler & Körner 2000). Lianas, which invest a relatively high proportion of their biomass in leaves, may benefit more than other plants in the understorey from increased carbon gain and increase in abundance in the canopy in the long run as well (Granados & Körner 2002; Forseth & Innis 2004; Körner 2004). On the other hand, the absolute stimulation of carbon gain should be much stronger as light limitation decreases towards the forest canopy, leading to increased growth compared with the understorey. Previous research with Hedera helix in a spruce/beech model ecosystem built in large open top chambers had documented a gradual increase in relative growth stimulation with diminishing light (Körner 2003b). Here, we tested these predictions in situ, in a mixed, old-growth forest stand in western Europe by subjecting naturally grown individuals of Hedera helix to elevated CO2 both in the understorey and in the subcanopy at c. 25 m for an entire growing season and compared growth rates with those of plants growing under ambient CO2.

Materials and methods

site description

This study was conducted in 2003 in a diverse mixed forest stand ∼15 km south of Basel, Switzerland (47°28′ N, 7°30′ E; elevation 550 m a.s.l.). The forest is about 100 years old, tree height ranges from 30 to 35 m, tree density (diameter at breast height = 0·1 m) is 415 trees ha−1 and stem basal area is 46 m2 ha−1 (Körner & Zotz 2003). The stand is characterized by a dominance of Fagus sylvatica L. and Quercus petraea (Matt.) Liebl., but there are also other broad-leaved species including Carpinus betulus L. and Acer campestre L. as well as four species of conifers. The shrub layer is composed of tree saplings and shrubs such as Lonicera xylosteum and Corylus avellana. In the herb layer tree seedlings are found together with forbs (e.g. Mercurialis perennis) and woody creepers (Rubus spp.; Hedera helix).

The typical humid temperate zone climate is characterized by mild winters and moderately warm summers. In the study year, the growing season of deciduous trees lasted from mid-April to mid-November. Leaf expansion of broad-leaved trees was complete by mid- to late May. Mean January and July air temperatures are 2 °C and 19 °C, respectively. Long-term average annual precipitation for the region is 990 mm, two-thirds of which commonly fall during the growing season. In 2003, however, precipitation during the growing season was only about half the 1989–99 average and temperatures were several degrees above average.

plant material and experimental design

Ivy (Hedera helix L., Araliaceae) is an evergreen woody liana that is common and widespread in Europe and adjacent Asia (Grivet & Petit 2002). Together with Rubus spp., creeping juveniles of this species are abundant in the understorey at the study site, frequently forming dense carpets. Adventitious roots, which are produced at the nodes of creeping stems, are used for climbing. A large proportion of the trees at the study site are infested with adult individuals of ivy, which occasionally climb up the trunks for up to 30 m. The species is found on a variety of host trees, but for ease of access only individuals growing in Larix decidua were included in this study. CO2 enrichment began on 25 April 2003, just around bud break of trees and ivy plants, and ended on 26 September.

In the understorey, we employed a similar set-up to a previous study with tree seedlings at the same site (Hättenschwiler & Körner 2000) using 35 cylindrical open-top chambers (OTCs) with a diameter and height of 0·38 m. In each OTC, we inserted two to five leading shoots of naturally growing juvenile creepers, randomly assigning each OTC to either ambient (c. 360 µl l−1; 12 OTCs) or two elevated CO2 concentrations (500 µl l−1 and 650 µl l−1; 11 and 12 OTCs). Three large blowers (DFV 20750, Meidinger Ag, Allschwil, Switzerland) produced a steady but very gentle flow of air sampled at 1 m above the forest floor. The air was distributed to the chambers by manifolds. Two blowers received additional CO2 to the airstreams to achieve elevated CO2 concentrations. The CO2 flow rate was manually regulated. Hättenschwiler & Körner (2000) provide more details on system performance.

Access to the forest canopy was achieved by a 45-m free-standing tower crane equipped with a 30-m jib (Körner & Zotz 2003). To control CO2 concentrations around the leading shoots of climbing ivy plants we installed eight transparent open-top plastic tunnels (OPTs) around the trunks of eight Larix trees, c. 20–30 m above the ground. Each tunnel was about 2·5 m long, measured c. 0·8 m in diameter, and was stabilized by loops of wire, which were attached to Larix branches. The tunnels were closed at the bottom and open at the top. The enclosed ivy plants had between 3 and 5 m of Larix canopy above them. Several leading ivy shoots were enclosed in each tunnel, but it was impossible to determine whether shoots on a given tree belonged to one or several genets owing to the dense tangle of shoots. All shoots in an OPT were exposed to either ambient (c. 360 µl l−1) or elevated (set point of 650 µl l−1) CO2 concentrations. To achieve these concentrations, we again used large blowers allowing airflows of c. 0·3 m3 min−1. One blower received CO2 additions at its outlet that was attached to a large supply tube subsequently divided by manifolds into the final four tubes reaching the four OPTs with elevated CO2. A similar set-up using ambient air was used for the four controls. The volume of air in each OPT was exchanged approximately once per minute.

co2 monitoring

After the initial set-up in early spring, the CO2 concentrations in two randomly chosen OTCs and in all four OPTs with elevated CO2 were checked about every 2 weeks using a portable infrared gas analyser (EGM-1; PP Systems, Hitchin, UK). If necessary concentrations were adjusted to the set points by manually regulating CO2 flow rates. In the understorey, the average concentrations of the 17 determinations made during the study period were, respectively, 525 ± 17 µl l−1 (mean ± SE; set-point: 500 µl l−1) and 666 ± 21 (mean ± SE; set-point: 650 µl l−1). Ambient controls averaged 363 µl l−1. The long-term average CO2 concentrations in the four OPTs were also close to the set-point with 609 ± 26 µl l−1 (mean ± SE; set-point: 650 µl l−1).

environmental conditions

The temperature, humidity and light regime in the understorey at our study site has already been described in detail by Hättenschwiler & Körner (2000). Located in the deep shade of the forest understorey, microclimatic conditions inside and outside the OTCs did not differ. The total daily PFD in the forest understorey ranged from 0·4 to 2·2 mol m−2 day−1, i.e. 0·8–4·8% of the amount reaching the top of the forest. Temperatures near the forest floor and in 20 m height were monitored continuously with two TidBit temperature data loggers (StowAwayTidBit, Onset Computer Corporation, Bourne, MA) during the entire growing season. The average air temperate at 20 m was almost 2 °C higher than at the forest floor (15·5 °C vs 13·7 °C). Air temperatures inside the OPTs were determined in two measurement campaigns from 26 April to 22 May and from 26 July to 8 August 2003, again using several TidBit data loggers. Each logger inside one of four randomly chosen OPTs was matched by another logger outside. Temperatures were logged in 30-min intervals. Occasionally, air temperature inside the OPTs could be up to 3 °C higher than outside during the day (during intense sunflecks) and up to 3 °C lower at night, but the long-term averages were very similar (ΔT inside vs outside <0·2 °C).

The light environment in the canopy was studied with a portable photon flux density (PFD) sensor (LI-189, LI-COR Inc., Lincoln, NE) on 2 days in June and September, 2003. Instantaneous readings were taken at several times during the day at the height of the leading shoots at each of the eight OPTs, with simultaneous reference measurements using an additional light-sensor installed at the top of the tower of the crane (‘above canopy’). Even at noon on a bright day in September, PFD close to the ivy plants was only 185 ± 58 µmol m−2 s−1 (mean ± SE), which corresponds to 12 ± 4% transmittance. A simultaneous determination of PFD in the understorey was more than a magnitude lower, i.e. 6·0 µmol m−2 s−1, which represented 0·4% transmittance. Later measurement from mid-May to July 2005 at the original growing sites of the leading shoots integrating light over 8 weeks using HOBO Pendant Light-Loggers (Onset Computer Corporation, Bourne, MA) showed a tight correlation with the original spot measurements (r = 0·84, data not shown).

growth measurements

This study rests on the assumption of branch autonomy for carbon (Sprugel, Hinckley & Schaap 1991), i.e. our experimental units consisted of small portions of entire plants that responded autonomously to the experimental conditions. At the beginning of the study, two to five ivy shoots per OTC or OPT were marked and apical shoot increment was measured monthly from 26 April to 26 September. While all leading shoots (initial length 14–46 cm) in the understorey were unbranched, similar unbranched shoots were rare in the subcanopy and shoot hierarchy was not unambiguous. Thus, principal and lateral shoots of the apical 40 cm of a leader were treated as one module: in addition to the monthly length increment of the principal shoot we also measured the total, cumulative branch lengths of these ‘modules’ at the end of the study.

In addition to these linear measurements, we determined the total biomass of new 2003 growth in late September. The entire new apical portion of each shoot that had developed during the study period was harvested and dried at 65 °C for 48 h. Dry mass of leaves and stems was determined separately.

To improve the statistical power of our analysis of subcanopy plants, we also checked for possible a priori differences in vigour among the experimental plants. Individual vigour was estimated by determining the average annual radial increment during the 2–3 years prior to the experiment in a distance of 1 m from the shoot apex.

leaf quality

To study changes in specific leaf area (SLA) and non-structural carbohydrate (NSC) concentrations leaf samples were taken with a cork borer (diameter 0·7 cm2) at regular intervals at the beginning of April, June, August and September 2003. For these traits, we collected four replicates in the understorey and four in the subcanopy for each CO2 concentration, each sample being from a different OTC/OPT. The first set of samples in April represents the condition before the CO2 treatment. After collection, samples were microwaved to stop enzymatic activity, and dried in a drying oven at 65 °C for 24 h. Then, leaf disks were weighed and SLA was determined. The quantification of NSCs, i.e. free, low molecular weight sugars (glucose, fructose and sucrose) plus starch, followed the enzymatic digest and glucose assay as described by Wong (1990) and Körner & Miglietta (1994). The NSC concentrations are expressed as a percentage of leaf dry matter.

statistical tests

Statistical analyses were done using R (Version 1·7.1, with the exception of the repeated measures analyses that were done with JMP version 3·1 (SAS Institute, Cary, NC). We treated individual OTCs/OPTs as sample units. Thus, before analysing absolute and relative growth of apical shoots, values of individual shoots within each OTC and OPT were averaged.


Both in the understorey and the subcanopy, shoots increased in length continuously throughout the study period. Plants in the understorey never developed side shoots nor flowers, while subcanopy plants invariably did. Flowers and fruits were, however, produced more distally and not on any of the shoot sections included in this study. In the understorey, we tested CO2 responses for two above-ambient concentrations. However, for all traits tested, the response was non-linear, i.e. there were no significant differences between plants exposed to 500 and 650 µl l−1 (repeated-measures anovas, Newman–Keuls post-hoc tests, P > 0·1). While a remarkable result in itself, this simplified the following analyses and allowed us to pool results for both elevated CO2 concentrations.

shoot length increment

In the understorey, shoot length increments were already significantly higher in elevated CO2 1 month after the start of the experiment in late May (Fig. 1) and remained higher until the end of the experiment. In relative terms, however, this initial stimulation was slightly lower (+41%) than in the following 4 months (+60 to +62%).

Figure 1.

Apical shoot length increment of Hedera helix at ambient (A, open bars) and elevated (E, closed bars) CO2. Data are means +1 SE. The upper panel shows shoots in the subcanopy (n = 4 for both A and E), the lower panel shows shoots in the understorey (n = 12 [A]; 23 [E]). Asterisks indicate significant differences (repeated-measures anova, Newman–Keuls post-hoc test; NS = not significant; *P < 0·05; **P < 0·01).

In contrast, shoot length increment was not significantly increased in elevated CO2 in the subcanopy (+27% at the end of the experiment; repeated-measures anova, P = 0·6). However, length increment in the subcanopy showed a significant, negative correlation with local light conditions (P = 0·03). When accounting for this variation by including local PFD as covariate in an ancova, with total length increment after 5 months as response variable, there was no significant CO2 effect either (F1,5 = 1·31, P = 0·3). However, the inclusion of individual differences in vigour (pretreatment radial growth) as an additional covariate did yield a significant CO2 effect (F1,4 = 12·54, P = 0·04).


In understorey plants, biomass production in shoots was consistent with the length growth signal. By the end of the season, total biomass of new shoots in elevated CO2 exceeded that under ambient CO2 by more than 80% (Fig. 2). This difference was primarily due to an almost twofold increase in leaf biomass. Although absolute biomass increments in the subcanopy were almost 10-fold higher than in the understorey (3·1 g vs 0·4 g per shoot), the relative stimulation by elevated CO2 of almost 40% in the subcanopy was not statistically significant (one-way anova, P = 0·36 for total shoots and P = 0·16 for stems). This surprising lack of a treatment effect was due to the great variation in PFD and the pronounced individual differences in vigour, because the inclusion of both factors as covariates in an ancova yielded a significant CO2 effect (F1,4 = 12·54, P = 0·03 for total shoots). Thus, as expected, elevated CO2 stimulated growth in the subcanopy, but to a much lesser relative extent than in the understorey.

Figure 2.

Biomass production of Hedera helix over one full growing season at ambient (open bars) and elevated CO2 (closed bars). Data are means +1 SE. Biomass production is given for the total shoot or separately for leaves and stems. Statistical significance values refer to three ancovas with local PFD and plant vigour as covariates (subcanopy) and three one-way anovas (understorey). Sample sizes as in Fig. 1.

non-structural carbohydrates and sla

NSCs accounted for more than one-quarter of the total dry mass of understorey leaves at the beginning of the growing season, in late April, probably reflecting carbon assimilation during late winter and early spring, when the forest canopy was largely leafless. This value dropped to <10% total dry mass in new leaves during the remaining study period, with no differences related to treatment conditions (Fig. 3, lower panel). In leaves in the subcanopy, which had similarly high NSC concentrations at the onset of the season, there was a consistent and significant difference in leaves growing in ambient vs elevated CO2: NSC was about 60% higher in elevated CO2 (t-tests, P = 0·03; Fig. 3, upper panel). Specific leaf area was about twice as high in the understorey than in the subcanopy (215 ± 6 cm2 g−1 vs 106 ± 5 cm2 g−1; means ± SE), but did not differ with treatment conditions (anova, P > 0·05).

Figure 3.

Seasonal changes in non-structural carbohydrate concentrations in Hedera helix in the understorey and the subcanopy of the forest. Open bars represent ambient (A), closed bars elevated (E) CO2. Data are means +1 SE. Sample sizes in the subcanopy are n = 4 (for both A and E), in the understorey n = 4 [A]; 8 [E]. Asterisks indicate significant differences (repeated-measures anova, Newman–Keuls post-hoc test; NS = not significant; *P < 0·05; **P < 0·01).


The field data presented here for the most abundant liana species in forests of western Europe highlight the potential impact of rising CO2 concentrations, in particular for the juvenile phase in the forest understorey. There, we observed a 60% stimulation in length increments of leading shoots (Fig. 1) and an even more pronounced increase in biomass production (Fig. 2), which was primarily due to a doubling of leaf biomass. This finding is fully consistent with earlier observations on small seedlings of the same species (Hättenschwiler & Körner 2003; Körner 2003b) or with results from a comparative study with several tropical lianas (Granados & Körner 2002). Other life-forms, e.g. tree seedlings, show similar in situ responses in deep forest shade (Würth, Winter & Körner 1998; Hättenschwiler & Körner 2000). However, a comparative study with seedlings of various woody plants (Hättenschwiler & Körner 2003) showed that shoot extension in the liana Hedera helix was much more stimulated in elevated CO2 than in four tree and shrub species. We currently lack the database to decide whether lianas in general show a stronger response to elevated CO2 compared with co-occurring trees, but such a suggestion does not seem unreasonable. Lianas invest a higher proportion of their biomass in leaves, and allocation to mechanical support tissue in stems is also less of a constraint. Massive shoot extension, as demonstrated for our study species, increases the likelihood of contact with a tree trunk, which in turn should lead to higher infestation rates of trees.

In the subcanopy of the forest, absolute growth of Hedera helix was almost an order of magnitude higher compared with the understorey, whether expressed as length increment of shoots or as biomass production (Figs 1 and 2). There was also a consistent trend towards higher growth in the subcanopy when exposed to elevated CO2 concentrations, but this trend was significant only when local differences in light and individual vigour were accounted for (Fig. 2). Thus, the overall results of this study are fully consistent with our general expectation of higher relative growth, but lower absolute growth stimulation in the understorey compared with the subcanopy (Bazzaz & Miao 1993; Körner 2000).

Inevitably, however, our analysis confounds plant responses to environmental differences between forest understorey and canopy with possible ontogenetic differences. Plants in the understorey were all juveniles, while subcanopy plants were all adults. In Hedera, juveniles and adults show conspicuous differences in leaf morphology, anatomy and physiology (Bauer & Bauer 1980; Hoflacher & Bauer 1982; Bauer & Thöni 1988). Canopy plants also feature a different growth habit and become more shrub-like with less pronounced apical dominance (Passecker 1977) and substantial reproductive investments.

The pronounced accumulation of NSC in subcanopy leaves (Fig. 3) reflects an imbalance between carbon gain by photosynthesis and sink activity (growth) at high light, while growth in the understorey was obviously still carbon-limited at our study site. This is in line with a recent global analysis of carbon limitation in woody plants (Körner 2003a), which concluded that deep shade was likely to be the prime situation under which carbon limitation plays a significant role under current CO2 concentrations. In contrast, substantial increases in NSC in subcanopy leaves suggest other constraints at more exposed sites, e.g. physiological controls within the leaf or nutrient limitations (Körner 2000). The high NSC concentrations at the beginning of the growing season in April (Fig. 3), which are also typical for co-occurring evergreen trees (Hoch, Richter & Körner 2003), support this interpretation: photosynthesis during winter and early spring before bud break probably increases NSC pools in the absence of active sinks (see also Hansen et al. 1997).

The negative correlation of length increment and local PFD in subcanopy plants suggests that the increased investment into vertical growth (= increase in height) is triggered by low light. It is noteworthy that the assumption of branch autonomy (Sprugel et al. 1991) does not preclude the possibility that study shoots in the subcanopy exported a certain amount of carbon, either to subapical horizontal branches or to reproductive structures. Our method of measuring growth may thus underestimate a possible CO2 effect. However, irrespective of the magnitude of growth stimulation of Hedera helix in the subcanopy, the documented increase in growth with elevated CO2 in the understorey (Figs 1 and 2) should improve the likelihood of Hedera plants reaching the forest canopy in the first place. All studies that explored CO2 responses in deep shade under realistic growth conditions, i.e. without adding nutrients, suggest a significant stimulation of plant growth (see the above references). There is evidence that lianas already influence the dynamics of entire tropical forest ecosystems (Phillips et al. 2002; Wright et al. 2004). Considering the lower importance of climbing plants in most temperate forests, a similarly strong impact in the temperate zone seems unlikely, although we do expect Hedera helix to become a more prominent component in a CO2-rich atmosphere. There are, however, more liana-rich temperate forests, e.g. in New Zealand (Dawson 1988), in which more aggressive lianas may become much more important for forest dynamics than in western Europe and North America (Sasek & Strain 1991; Forseth & Innis 2004). We have now unpublished evidence (F. Grob & Ch. Körner) that the second most important European liana, the deciduous Clematis vitalba, responds very similar to Hedera, which corroborates that lianas are among the most strongly affected plant types as atmospheric CO2 concentrations increase.

An important observation is that similar to Hättenschwiler & Körner (2003) we found no further effect of elevated CO2 at our test concentration above 500 µl l−1 in the understorey. Granados & Körner (2002) studied tropical lianas at four different CO2 concentrations and also observed the strongest CO2 effect below 500 µl l−1. In their case, growth even declined at 700 µl l−1 compared with 560 µl l−1 and came close to the 420 µl l−1 rate. It thus seems that the current increase of atmospheric CO2 concentration is most influential in such shade-growing plants.

In summary, elevated CO2 strongly affects the extension growth of juveniles of Hedera helix in the understory, while growth of plants that have already reached the forest canopy benefits to a smaller degree from CO2 enrichment. Thus, in the future we expect both a larger proportion of trees to host Hedera helix and individual tree crowns to be more heavily packed.


We thank Olivier Bignucolo and Erwin Amstutz for crane operations. Financial support came from the Swiss National Science Foundation (grant 3100–067775·02 to CK), the Swiss Federal Office of the Environment (BAFO) and the University of Basel. Dr Günter Hoch (Basel) helped with NSC analyses.