Does elevated CO2 facilitate naturalization of the non-indigenous Prunus laurocerasus in Swiss temperate forests?


  • S. Hättenschwiler,

    Corresponding author
    1. Institute of Botany, University of Basel, Schönbeinstrasse 6, CH-4056 Basel, Switzerland
      †Author to whom correspondence should be addressed. E-mail:
    Search for more papers by this author
  • C. Körner

    1. Institute of Botany, University of Basel, Schönbeinstrasse 6, CH-4056 Basel, Switzerland
    Search for more papers by this author

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


  • 1An increasing abundance of the non-indigenous evergreen woody plant species Prunus laurocerasus has been observed in the understorey of Swiss temperate forests. We addressed the question whether rising atmospheric CO2 concentration contributes to the success of this species in a comparative test with four co-occurring native species (Ilex aquifolium, Hedera helix, Fraxinus excelsior, Carpinus betulus).
  • 2We grew plants from germination to the end of the third growing season in open-top chambers exposed to either ambient or two elevated CO2 concentrations (500 and 660 µmol mol−1) in a deeply shaded forest understorey (1·2–3·2% of full sun).
  • 3Species differed greatly in their response to CO2. Biomass growth in Prunus increased by an average of 56% at the two elevated CO2 concentrations compared to ambient CO2; there was no significant difference between 500 and 660 µmol mol−1. In contrast the native Ilex, with the same functional traits, a similar life history and occurring in the same habitat, showed no significant CO2 response.
  • 4A particularly large and nearly linear CO2 effect on seedling growth was observed in the liana Hedera with 100% more biomass and 137% longer stems at 660 µmol CO2 mol−1 compared to ambient CO2. Seedlings of the deciduous tree species Fraxinus produced 43% more biomass at elevated CO2 (no significant difference between 500 and 660 µmol mol−1), but there was no significant CO2 effect on Carpinus seedlings.
  • 5Our results indicate that elevated CO2 might contribute to the current spread of Prunus in natural forests. The strong CO2 response in Hedera suggests an increasing rate of tree colonization with rising CO2. Increasing dominance of non-indigenous understorey species and accelerated liana colonization of canopy trees could both have far-ranging consequences for forest community dynamics and composition.


As a consequence of expanding global transport and commerce, introductions of non-indigenous species have increased dramatically in the past 500 years (di Castri 1989). A few non-indigenous species eventually become invasive (Mack et al. 2000) and threaten native plant communities and ecosystems through competition, predation or altered ecosystem processes (Vitousek & Walker 1989; D’Antonio & Vitousek 1992; Schmitz et al. 1997; Callaway & Aschehoug 2000). Biotic invasions are a major feature of global change (Vitousek et al. 1996; Mack et al. 2000), other elements of which, such as changes in land use, climate, nitrogen deposition or atmospheric CO2 concentration, may enhance the probability of persistence and naturalization of immigrating species and facilitate and/or accelerate invasion rates (Dukes & Mooney 1999).

For example, winters have become increasingly mild and summers moist in the past 30 years compared to the long-term average. These changes have been associated with a dramatic spread of non-indigenous evergreen woody plant species in the understorey of winter-deciduous temperate forests in Switzerland (Klötzli et al. 1996; Walther 1999). These authors observed an increasing distribution, abundance and vigour of several evergreen broad-leaved (laurophyllous) species (e.g. Prunus laurocerasus, Laurus nobilis, Eleagnus pungens) that had naturalized from gardens and parks, where they have been planted for centuries. Significant changes in species composition, structure and function of winter-deciduous forests at low altitudes are to be expected if the current invasion by evergreen woody plant species continues.

Rising atmospheric CO2 concentration may have an additional favourable effect on seedling growth in the forest understorey, and might contribute to improved establishment of non-indigenous species. In the light-limited forest understorey, the relative CO2 effects may be particularly large because higher CO2 concentrations reduce the photosynthetic light compensation point (Osborne et al. 1997) and increase sunfleck utilization efficiency (Naumburg & Ellsworth 2000) in understorey plants. Both these responses may result in a significantly improved carbon balance and growth of shaded plants. Large stimulations of biomass growth under elevated CO2 have been observed in a variety of species in the natural understoreys of tropical (Würth, Winter & Körner 1998) and temperate forests (Hättenschwiler & Körner 2000). Seedlings of co-occurring tree species, however, showed distinct responses to elevated CO2 (Hättenschwiler 2001) as reported in an earlier growth chamber experiment (Bazzaz & Miao 1993). Such CO2-related interspecific differences in plant growth are common (Körner & Bazzaz 1996; Körner 2000) and likely to change community composition, with consequences for biodiversity and ecosystem processes (Körner 2000).

In the present study we asked how the widespread and increasingly abundant non-indigenous evergreen Prunus laurocerasus responds to elevated CO2 in a temperate forest understorey, and if this response differs from that of native woody plant species.

Materials and methods

study site and plant material

The study was conducted at the previously established Hofstetten experimental forest area, 12 km south-west of Basel (47°28′ N 7°30′ E) at an elevation of 550 m a.s.l. (Hättenschwiler & Körner 2000). The forest is about 120 years old and is composed of various broad-leaved deciduous and coniferous tree species, with Fagus sylvatica and Quercus robur as the dominant canopy species. The tallest trees reach a height of about 35 m with a tree canopy leaf area index of ≈5·0. Long-term mean annual precipitation is 885 mm, and long-term average daily mean temperatures are −2·0 °C in January and 18 °C in July The topsoil has a pH of 5·8, is comparatively nutrient-poor (see Hättenschwiler & Körner 2000 for nutrient analyses), and is underlain by limestone bedrock.

Prunus laurocerasus L., the common non-indigenous evergreen woody species found in Swiss deciduous forests, was the focus of our experiment. Prunus laurocerasus is native to the humid Black Sea coast and South-west Asia, and occurs on relatively moist soils in forest understoreys or as a subdominant tree. It has been planted as an ornamental plant in gardens and parks all over southern and western Europe for ≈400 years, from where it has escaped during the past 20–30 years. It is now common in the understorey of native forests at lower elevations (<750 m a.s.l.) and contributes significantly to the increasing dominance of evergreen understorey vegetation observed in Swiss forests (Walther 1999; Walther & Grundmann 2001). For comparison, four additional species native to Switzerland, representing three different functional types of woody plant, were included in the experiment. Like P. laurocerasus, Ilex aquifolium L. is a broad-leaved evergreen shrub or small tree (up to 6–10 m tall) Hedera helix L. is a broad-leaved evergreen liana; and Carpinus betulus L. and Fraxinus excelsior L. are broad-leaved deciduous canopy trees (study species are denoted by genus hereafter). Ilex occurs in the same habitat, attains about the same size, shares the same morphological type of leaves, and produces similar-sized, fleshy fruits dispersed by birds, as is Prunus, but occurs much further north in oceanic parts of Europe (up to 64° northern latitude). Except for Buxus sempervirens L., which has a more restricted occurrence in some dry forests in the south-west of Switzerland, Ilex is the only native broad-leaved evergreen small tree in Swiss forests. The liana Hedera grows up to the top of the tallest trees (>30 m), has a similar distribution to Ilex, and is a common and abundant species in a variety of forests. Hedera often covers the forest floor, where it can grow for many years until it contacts an appropriate tree to climb. The deciduous tree Carpinus is a common canopy species in oak-dominated forests at lower elevations (<750 m a.s.l.); Fraxinus commonly co-occurs with beech and oak, and is somewhat more shade tolerant than Carpinus. All study species occur naturally at or near our study site.

Seeds of Carpinus and Fraxinus were collected from single trees at the study site in autumn 1997 and stored at 5 °C for 8 weeks to allow the seeds to ripen completely. The seeds were then stratified in a 1 : 1 mix of sand and peat in the Botanical Garden of the University of Basel from January to March 1998. Because Prunus and Ilex seeds germinate only after two winters of stratification, they were collected from shrub-sized individuals in a garden near Basel (Prunus) and in the Botanical Garden of the University of Basel (Ilex) in late autumn 1996. After collection, the fruits were soaked in water (replaced every 3–4 days) for 2 weeks, the softened mesocarp was then rubbed off and the stones were allowed to dry, followed by stratification as described above from December 1996 to March 1998. As soon as the seeds started to germinate in early April 1998 they were inserted directly into the undisturbed forest soil within each of the 36 open-top chambers (OTC). At the same time, germinating Hedera seeds were collected from the forest floor at the study site and transferred to the OTCs.

Forty-eight (0·04 × 0·04 m) seed positions were defined per OTC using a grid, and species were randomly allocated to each position. A total of 15 germinating seeds of each of the two species, Carpinus and Fraxinus, and a total of six germinating seeds of the remaining three species, were inserted into each OTC. More seeds of the deciduous tree species were used as greater initial mortality was expected because of their small seeds. Spontaneously emerging natural seedlings and forbs within the OTCs were removed, but the natural litter layer remained in place.

experimental design

Although the experiment was carried out in the same forest understorey as a previous study with a similar experimental design (Hättenschwiler & Körner 2000), new and undisturbed positions were chosen for all 36 OTCs. The cylindrical OTCs had a diameter and height of 38 cm and were constructed from 2 mm thick UV-transmitting Plexiglas (gs 2458, Röhm GMBH, Germany). The relatively small chamber size had the advantage of well defined microsites in terms of light availability. Each OTC was randomly assigned to either ambient (366 µmol mol−1) or two elevated CO2 concentrations (500 and 660 µmol mol−1) resulting in 12 OTCs per CO2 treatment. Three large blowers produced a steady but gentle flow of air sampled at 1 m above the forest floor. The air was distributed to the chambers by manifolds. Two blower systems received computer-controlled CO2 additions to the airflow to achieve the target CO2 concentrations (see Körner et al. 1996; Hättenschwiler & Körner 2000 for more details on CO2 control and system performance). Carbon dioxide enrichment started on 10 April 1998 and continued during winter, except for 8–12 weeks between mid-December and end of February when snow covered the ground, and ended on 3 October 2000 with the final plant harvest. The actual CO2 concentrations deviated little from target concentrations, with annual averages of 502, 499 and 505 µmol mol−1, and 661, 658 and 663 µmol mol−1 for the intermediate and high CO2 concentrations, respectively, over the three consecutive years.

Continuous microclimate measurements inside and outside the OTCs showed no significant differences resulting from the use of OTCs and no differences among the CO2 treatments (Hättenschwiler & Körner 2000), because of the buffering effect of the tree canopy. Photosynthetically active photon flux densities (PPFD) were measured with a permanent light sensor 15 cm above the ground in each OTC (GaAsP photo diodes, spectral response range 300–680 nm, G1115, Hamamtsu Photonics, Hamamtsu City, Japan). Two additional sensors were placed outside the forest as a reference. The photo diodes were sealed in a waterproof tube covered with a white diffuser cap. All sensors were calibrated with a quantum sensor (LI-189, LI-COR Inc., Lincoln, NE, USA). Readings were logged at 10 s intervals (CR 10 with relay multiplexer, Campbell Scientific Ltd, Loughborough, UK) and means of 5 min measurements were stored. Light data were continuously collected for 10–20 days of each month (March 1999–September 1999, June 2000–August 2000). The microsite-specific light availability for the 36 OTCs is indicated as the daily PPFD measured on overcast summer days during the peak season (June–August), as in a previous study (Hättenschwiler & Körner 2000). On average, the total daily PPFD on overcast summer days ranged from 0·41 to 2·26 mol m−2 day−1 (0·9–5·0% of full sun) in the forest understorey studied. These values compare to 20·5 ± 0·7 mol m−2 day−1 measured on the same overcast summer days, and to 44·8 ± 0·6 mol m−2 day−1 measured on clear summer days outside the forest.

During the winter following the second growing season, intruding mice largely destroyed the seedlings in 10 OTCs. We were fortunate that the mice affected the CO2 treatments more or less equally, reducing the final replication to eight OTCs per CO2 treatment and covering a range in total daily PPFD from 0·53 in the darkest OTC to 1·42 mol m−2 (1·2–3·2% of full sun) in the brightest OTC (Fig. 1). The mean daily PPFD across the eight OTCs within each CO2 treatment was almost identical among CO2 treatments (Fig. 1), with an overall mean daily PPFD of 0·99 mol m−2 day−1 (2·2% of full sun) across the 24 OTCs used for the final analysis. On average, more than 80% of all 5 min means of PPFD were below 15 µmol m−2 s−1 on a typical overcast summer day (98% in the darkest OTC and 67% in the brightest).

Figure 1.

Mean daily sums of photosynthetically active photon flux density (PPFD) within eight open-top chambers (OTCs) maintained at each of three different CO2 concentrations. Data are means from three overcast days in June, July and August 1999; error bars represent standard deviations. Light was measured continuously and stored as 5 min averages. All readings above 1 µmol m−2 s−1 were considered for the calculation of daily sums (accuracy of light sensors ±1 µmol m−2 s−1).

plant sampling and data analysis

Seedlings were checked regularly for survival during the first growing season, and to a lesser extent in the second and third seasons. To minimize shading among seedlings in OTCs, some seedlings were removed (particularly weak seedlings were removed first, then those for maximizing spatial distance among remaining seedlings) at the end of the first and second seasons, leaving three or four individuals per species and per chamber. For the final harvest (3–6 October 2000), after three growing seasons, we dug out the entire soil block covered by the OTC to a depth of 20 cm. Each individual seedling was then washed out of the soil using a hose. Seedlings were divided into roots, stems and leaves, and all plant parts were oven-dried at 80 °C for 36 h and weighed. For the determination of total leaf area and specific leaf area (SLA), we took leaf discs from different leaves (avoiding the middle vein) of each seedling (five or six discs per seedling) using a cork borer of a known diameter. From the measured dry masses of all plant parts, root mass fraction (RMF: root mass/total seedling mass), stem mass fraction (SMF: stem + petiole mass/total seedling mass), leaf mass fraction (LMF: leaf mass/total seedling mass), and leaf area ratio (LAR: total leaf area/total seedling mass) were calculated.

Data of individual seedlings within each OTC were averaged for each species and used as the sample units (n = 8 replicates per CO2 concentration). Analyses of covariance (ancova) were performed to test differences among species and CO2 treatments with microsite-specific light availability as the covariate (Sokal & Rohlf 1981). In these analyses, treatment effects were tested as differences in the elevation of the Y-intercept of linear regressions between growth variables and light availability. The effect of the covariate (light) was tested as the regression slope and whether it differed from zero, and the interactions between treatments and light were tested as the differences among slopes of the regression lines (Sokal & Rohlf 1981). Treatment effects, i.e. the differences in the Y-intercepts of regressions, were tested at the lower end of the light gradient (1·2% of full sun); the upper end (3·2% of full sun); and also at 2·2% of full sun as the intermediate light availability. Differences between individual levels within factors were tested using Fisher's LSD post hoc tests. All biomass-related variables were ln-transformed prior to statistical analyses to meet the requirement of normal distribution. systat ver. 5·2·1 (Systat Inc., Evanston, IL, USA) was used for statistical analyses.


seedling survival and height growth

Average mortalities of 20 and 12% in deciduous and evergreen species, respectively, were observed within the first 4 weeks after seeds were transferred to the OTCs. This initial mortality occurred randomly, with no differences among CO2 treatments and among OTCs differing in light availability. Following these initial losses, survival remained high with only little additional seedling death throughout the entire experiment and no CO2- or light-related differences.

The final stem length after three growing seasons differed among species (Table 1; Fig. 2). Post hoc contrast analyses revealed significant differences between Carpinus as the tallest and Ilex (P = 0·038) and Prunus (P = 0·043) as the smallest species, but no other significant differences among species. Increased atmospheric CO2 concentration had a significant positive effect on final stem length (Table 1; Fig. 2), but the species differed in their response (significant species × CO2 interaction). Seedlings of the two broad-leaved evergreens Ilex and Prunus did not grow taller at elevated CO2, whereas the two broad-leaved deciduous species Fraxinus and Carpinus had significantly longer stems at the intermediate CO2 concentration of 500 µmol mol−1 with no further stimulation at 660 µmol CO2 mol−1. In contrast, stem length in the evergreen liana Hedera showed a strong and linear response to increasing CO2 concentration. Hedera stems were 1·7–2·4 times longer at 500 and 660 µmol CO2 mol−1 compared to seedlings grown at current ambient CO2. The average (mean of the two elevated CO2 concentrations) relative CO2 stimulation in stem length of +103% in Hedera is substantially larger than that in Fraxinus (+17%) or in Carpinus (+25%).

Table 1.  Analyses of covariance of seedling stem length, biomass, biomass fractions (see text) and morphological traits (LAR, SLA; see text) with species and CO2 as fixed factors and microsite-specific light availability as covariate
VariableSpeciesCO2Species × CO2LightLight × speciesLight × CO2
  1. Tests for differences among different levels of fixed factors (Y-intercept) were done at the intermediate light level of 2·2% of full sun (1·01 mol m−2 day−1). F ratios, P values and coefficients of determination (r2) are shown.

Stem length 16<0·001 9·9<0·0012·20·043·40·070·40·841·30·270·58
Total mass 41<0·001 9·9<0·0010·70·666·10·020·40·820·30·770·72
Leaf mass 34<0·001 4·5   0·0140·60·814·60·030·30·910·40·690·64
Stem mass 41<0·00115<0·0011·00·455·90·020·60·640·80·450·73
Root mass 63<0·001 9·5<0·0010·80·625·80·020·60·690·60·530·79
LMF105<0·001 3·2   0·050·90·561·10·300·50·752·40·100·85
SMF 29<0·001 3·3   0·050·60·790·90·340·80·560·90·430·62
RMF 40<0·001 0·1   0·880·70·690·00·981·20·303·50·030·69
LAR 16<0·001 3·7   0·030·60·790·10·760·20·956·90·0020·53
SLA682<0·001 0·6   0·531·40·202·80·100·50·717·00·0020·97
Figure 2.

Final stem length in seedlings of five different woody plant species exposed to three different CO2 concentrations (means ± SE of eight OTCs). Seedlings grew in the deeply shaded forest understorey for a total of three growing seasons. Different letters indicate significant differences among CO2 treatments within species.

Differences in microsite light availability only marginally affected seedling height growth, and the weak positive correlation between stem length and light availability across the narrow light gradient was similar among species and CO2 treatments (no light × species and light × CO2 interactions; Table 1).

biomass production and allocation

Total seedling biomass production over the three growing seasons was significantly different among species and CO2 concentrations (Table 1; Fig. 3). Prunus produced significantly more biomass than the two other evergreen species Ilex (P = 0·033, post hoc contrast) and Hedera (P = 0·031). No other post hoc pairwise comparisons between species were significant. The average relative CO2 effect on seedling biomass across species was +51% at 500 and +48% at 660 µmol CO2 mol−1 compared to ambient CO2, suggesting no further CO2 stimulation beyond 500 µmol CO2 mol−1. Although the species × CO2 interaction was not significant, there was some apparent variation among species in their CO2 response (Fig. 3). Tested separately within each species, a significant CO2 effect on total biomass was found in Hedera (+78%, mean across both CO2 concentrations), Prunus (+56%) and Fraxinus (+43%), but not in Ilex and Carpinus (Fig. 3). As the mean biomass in Carpinus even declined somewhat beyond 500 µmol CO2 mol−1, similarly to Fraxinus, we additionally tested the CO2 effect between ambient CO2 and the intermediate CO2 concentration, and between ambient CO2 and the pooled data across the two high CO2 concentrations. Neither analysis yielded significant CO2 effects (P > 0·09) because of the large variation in the size of Carpinus. Leaf, stem and root biomass differed in their relative contributions to total biomass among species, but showed essentially the same pattern of CO2 effects as did total biomass (Table 1; Fig. 3).

Figure 3.

Total seedling biomass of five different woody plant species exposed to three different CO2 concentrations over three growing seasons (means ± SE of eight OTCs). Levels of significance from analyses of covariance to test for effects of CO2 and light (covariable) within species are shown. Different letters indicate significant differences among individual treatment levels within those species exhibiting a significant CO2 effect.

Seedling biomass generally increased with increasing light availability. However, differences in microsite-specific light availability explained comparatively little variability in biomass (Table 1).

Biomass fractions differed among species (Tables 1 and 2), essentially due to the significantly larger leaf mass fractions in evergreen compared to deciduous species and concomitant differences in stem and root mass fractions. Elevated CO2 significantly affected leaf mass fraction (lower LMF with increasing CO2 across all species), while root mass fraction did not change (Table 1). The significant light × CO2 interaction on RMF, however, indicates that regression lines of RMF as a function of microsite light availability differed in their slope among CO2 concentrations. Averaged across all species RMF at ambient CO2 was 0·34 in the OTCs of <1 mol photons m−2 day−1 and 0·38 in those of >1 mol photons m−2 day−1, as the intermediate level across the light gradient. At the two elevated CO2 concentrations (no difference between 500 and 660 µmol CO2 mol−1), RMF was 0·39 (PPFD <1 mol m−2 day−1) and 0·35 (PPFD >1 mol m−2 day−1). The light × CO2 interaction on RMF was significant only across all species, but not when tested within each species separately.

Table 2.  Biomass fractions (see text), leaf area ratio (LAR) and specific leaf area (SLA) in 3-year-old seedlings grown at different CO2 concentrations of 365, 500 or 660 µmol mol−1 in the forest understorey (means with SE in parentheses)
RMF (g g−1)
3650·31 (0·02)0·30 (0·01)0·34 (0·02)0·47 (0·02)0·38 (0·02)
5000·29 (0·03)0·29 (0·01)0·37 (0·02)0·47 (0·02)0·39 (0·01)
6600·26 (0·01)0·28 (0·02)0·40 (0·04)0·47 (0·02)0·37 (0·02)
SMF (g g−1)
3650·34 (0·02)0·32 (0·02)0·27 (0·02)0·38 (0·01)0·40 (0·02)
5000·37 (0·01)0·31 (0·01)0·26 (0·01)0·39 (0·01)0·41 (0·01)
6600·40 (0·01)0·32 (0·02)0·28 (0·02)0·39 (0·01)0·44 (0·03)
LMF (g g−1)
3650·35 (0·01)0·38 (0·03)0·38 (0·01)0·15 (0·01)0·23 (0·01)
5000·34 (0·02)0·40 (0·02)0·38 (0·02)0·14 (0·01)0·20 (0·01)
6600·35 (0·01)0·41 (0·01)0·32 (0·04)0·15 (0·01)0·19 (0·01)
LAR (cm2 g−1)
36570·0 (2·6)61·9 (4·7)48·2 (2·3)62·0 (6·9)78·7 (5·7)
50064·9 (3·3)67·4 (3·5)46·4 (2·4)60·1 (4·2)73·8 (3·0)
66064·4 (2·2)67·8 (1·0)41·1 (5·4)58·1 (2·8)61·4 (4·3)
SLA (cm2 g−1)
365200 (7)163 (6)126 (3)412 (18)350 (15)
500192 (3)169 (5)122 (2)422 (11)364 (12)
660186 (6)166 (2)129 (4)404 (17)331 (14)

Specific leaf area differed considerably among species (Tables 1 and 2), driven mainly by smaller SLAs in evergreen compared to deciduous species. Because large SLA correlated well with small LMF at the species level, the leaf area ratio (LAR) as the product of SLA and LMF was similar among species (no significant differences in post hoc contrast analyses). While there was no significant effect on SLA, LAR decreased with increasing CO2 except in Ilex (Table 2). However, SLA and LAR had significant light × CO2 interactions. Similarly to LMF, SLA and LAR decreased slightly with light availability at ambient CO2, but increased with increasing light availability at elevated CO2. Again, this pattern was significant only across species, but not when tested within each species separately.

With the exception of a marginally significant reduction of SLA, microsite-specific light availability per se did not influence seedling biomass allocation or morphology.


Seedlings of the non-indigenous evergreen P. laurocerasus growing in the forest understorey produced considerably more biomass at elevated atmospheric CO2 concentration than at ambient CO2. The initial phase of growth after germination and early establishment of seedlings is a critical part in a tree's life history, one that largely determines regeneration success (Clark & Clark 1992; Kobe et al. 1995). Our results suggest that rising atmospheric CO2 facilitates seedling establishment of Prunus in the natural forest understorey, perhaps contributing to the observed increasing abundance of this species in Swiss forests (Walther 1999; Walther & Grundmann 2001), but cannot assess the relative importance of elevated CO2 compared to other factors, such as warmer winters, moister summers or changes in forest management, for the apparent spread of Prunus in different native forests. Interactions among different elements of global change, rather than a single cause, are probably common determinants of the success of non-indigenous plant species (Dukes & Mooney 1999), but separating the effects of these elements is complex and difficult (Huenneke 1997).

There is relatively little unambiguous evidence for CO2 stimulation of invasive plant species in natural ecosystems (Dukes & Mooney 1999) because appropriate field experiments are lacking. In line with our findings, although for a completely different ecosystem, Smith et al. (2000) reported a higher above-ground biomass production and seed rain of the invasive annual grass Bromus madritensis spp. rubens compared to native annuals in response to CO2 enrichment in the Mojave Desert. They concluded that this shift in species composition might accelerate the fire cycle, reduce biodiversity and alter ecosystem function in North American deserts. In another experiment with microcosms exposed to elevated CO2 in OTCs, Dukes (2002) observed substantial CO2 stimulation of biomass growth in the invasive annual forb Centaurea solstitialis growing in monocultures, and similar but non-significant stimulation when Centaurea grew in an artificially assembled community of Californian serpentine grassland species. The greater CO2 responsiveness in Centaurea than in native species indicates that the effects of rising atmospheric CO2 may be involved in the success of this problematic species invading grassland ecosystems of western North and South America (Dukes 2002). Comparisons with native species and their responses to the same environmental change are important when assessing if a non-indigenous species benefits more from the change than do native species of similar life histories or ecological niches. The native Ilex has the same functional traits as Prunus, has a similar life history and occurs in the same habitat. However, Ilex did not grow more under elevated CO2 concentrations, contrasting with Prunus. This indicates a selective growth stimulation of a non-indigenous plant species by increasing atmospheric CO2 compared to a native species occupying the same ecological niche. Consequently, elevated CO2 might enhance the long-term recruitment success of the non-indigenous Prunus, potentially increasing its relative dominance compared to the native Ilex.

Other native species of contrasting functional types, however, showed a similar or even a stronger response to elevated CO2 compared to Prunus. The CO2-induced growth stimulation in the liana Hedera was particularly large, with 100% more biomass and 137% longer stems at 660 µmol CO2 mol−1 compared to ambient CO2. Direct competition between Hedera and Prunus in the forest understorey is supposedly very limited. Improved establishment and faster growth of Hedera seedlings, however, increases the probability of tree host contact, resulting in more Hedera individuals climbing on trees. This may have important implications for forest community structure and ecosystem processes, because lianas influence forest regeneration and canopy composition by their impact on tree growth and mortality (Putz 1984; Schnitzer & Bongers 2002). The CO2 response of Hedera as the most important and common liana species of European oceanic temperate forests agrees with an earlier study of tropical lianas exposed to elevated CO2 in growth chambers (Granados & Körner 2002). They concluded that enhanced vigour of lianas at elevated CO2 could accelerate tropical forest dynamics, leading to greater abundance of early successional tree species and to reduced carbon sequestration in the long term.

Biomass production of the two deciduous tree species increased less at elevated CO2 than in Prunus and Hedera. However, we observed no common CO2 responsiveness within the functional groups of evergreen broad-leaved species and deciduous tree species. It might be argued that the number of species was not sufficiently large to generalize about functional group responses to elevated CO2. In accordance with our study, Hättenschwiler & Körner (2000) previously found highly different growth responses to elevated CO2 among three deciduous tree species (F. sylvatica, Acer pseudoplatanus, Q. robur) and three conifer species (Taxus baccata, Abies alba, Pinus sylvestris). Likewise, several deciduous tree species from the north-eastern USA grown under controlled laboratory conditions responded distinctly to elevated CO2 (Bazzaz, Coleman & Morse 1990; Bazzaz & Miao 1993). Taken together, these findings provide little evidence for a common CO2 response within functional groups of co-occurring tree species. Moreover, species’ ranking according to their CO2 responsiveness at low light (average 1·3% of full sun) was reversed at higher light availability (average 3·4% of full sun) (Hättenschwiler & Körner 2000; Hättenschwiler 2001). The CO2 response of the same species changes with a relatively subtle shift in understorey light availability. This does not support predictable species-specific CO2 responses based on relative shade tolerance of co-occurring tree species, as suggested elsewhere (Kubiske & Pregitzer 1996; Kerstiens 1998). Unfortunately, the light gradient in our study was too narrow to detect distinct growth responses to elevated CO2 with increasing light availability. The significant light × CO2 interactions in RMF and in the morphological traits LAR and SLA, however, hint at the possibility of such light-dependent growth responses to elevated CO2. While the range of microsite-specific light availability between 1·2 and 3·2% of full sun covered in this study is representative for understorey conditions of most old-growth, closed-canopy forests, the species-specific CO2 responses reported here may differ in more open forests (Hättenschwiler & Körner 2000).

In conclusion, we have demonstrated that elevated atmospheric CO2 concentrations stimulate seedling growth of the non-indigenous Prunus in the natural understorey of a native forest. Moreover, Prunus is more responsive to CO2 than is Ilex, the potential native evergreen competitor occupying the same niche; Prunus may thus increase in dominance in the future. Facilitation of the establishment of Prunus by elevated CO2 is an example of how an element of global change may interact with the success of an invasive species, leading to shifts in community composition. The reported increasing vigour of Hedera in response to elevated CO2 may have even greater consequences for community composition and ecosystem structure by changing forest dynamics.


We thank S. Pelaez-Riedl for help with seedling harvest, and L. Zimmermann for the development and maintenance of technical field installations and collection of microclimatic and CO2 data.