31
Plant Ecology 135: 31–41, 1998.
c 1998 Kluwer Academic Publishers. Printed in Belgium.
Natural pasture community response to enriched carbon dioxide atmosphere
Liette Vasseur1;2 & Catherine Potvin2
1
Department of Biology, Saint-Mary’s University, Halifax, Nova-Scotia B3H 3C3 Canada; 2 Department of
Biology, McGill University, 1205 Dr. Penfield, Montréal, Québec H3A 1B1 Canada
Received 3 December 1996; accepted 29 October 1997
Key words: Canonical correspondence analysis, CO2 , Multivariate analysis, Open-top chambers, Plant community,
Succession
Abstract
We examined the response of a pasture community in southern Quebec (Canada) to long-term exposure of enriched
atmospheric CO2 conditions. The study was conducted using open-top growth chambers directly placed on top
of the natural pasture community. To investigate the change in the overall species composition in time and
space, we used canonical correspondence analysis, a direct ordination method. Over the three years, the overall
community responded significantly to enriched CO2 . The analyses show that, after three years, CO2 was the
most important environmental variable affecting the species composition. Initially the presence of the wall of the
chambers influenced the composition but CO2 became more important by the third year. Soil and air temperatures
only slightly influenced the community composition. The first two axes of the canonical correspondence analysis
explained a large proportion of the variation in the three years and these trends appeared to increase with time.
Species such as Agropyron repens appeared to be positively influenced by the presence of the wall (slightly warmer
conditions). However, the analyses suggest that Phleum pratense and Trifolium repens, for example, were favored
by the increase in atmospheric CO2 . The variation in species composition in enriched versus ambient CO2 chambers
suggests that the effect of the environmental factors, particularly CO2 , were important in affecting the rate and
pattern of succession. Furthermore, the temporal increase in importance of the variable CO2 in the present analyses
indicates that there might be a time-lag in response to atmospheric enrichment.
Introduction
It is well known that because of human activities,
the atmospheric concentration of carbon dioxide is
increasing and may double within the next century
(Schneider 1989; Houghton et al. 1990). Besides
potential effects on climatic conditions, specifically
temperature and precipitation, this atmospheric enrichment is known to directly affect plants. Positive effects
of elevated CO2 concentrations on plant growth have
been reported in other experiments (Kimball 1983;
Bazzaz 1990). However, most experiments have been
conducted in controlled environments (greenhouse or
growth chambers) and at the individual species level
(Lagauderie et al. 1988; Overdieck 1986; Hocking &
Meyer 1991; Korner 1993a).
Recently with the development of new in situ
technologies, open-top chambers and Free-Air
CO2 Enrichment (FACE) systems, experiments on
plant communities are becoming possible. Several of
these in situ experiments have examined the responses
of short stature communities such as alpine tundra
(Korner et al. 1996) or salt marsh (Drake et al. 1996).
Of most interest to us, a number of experiments have
been conducted in grassland ecosystem. The longest
running grassland CO2 experiment was established
by Owensby and co-workers (1996) in the tallgrass
prairies of Kansas. Two experiments in Switzerland
examined the responses of either calcareous grassland
(Luscher et al. 1996) or grassland species (Leadley
& Korner 1996) to elevated CO2 . Finally, the Jasper
Ridge experiment (Field et al. 1996) was conducted in
a serpentine Mediterranean grassland. The vegetation
32
at our field site bears more resemblance with the Kansas tallgrass prairies that the other sites. Our results
therefore will be compared with those of Owensby et
al. (1996).
The main objective of community ecology is to document and understand spatial and temporal patterns of
species occurrence and abundance. In this study, we
examine the impact of CO2 enrichment on a grassland community. The present paper is part of a larger
study that examines community response of a natural
pasture to long-term CO2 exposure (Stewart & Potvin
1996; Potvin & Vasseur 1997; Taylor & Potvin submitted). In the first paper of this series (Potvin and Vasseur
1997), we show that, after 3 years of monitoring, species richness is significantly higher in the elevated CO2
open-top chambers than in the ambient environment.
We suggest that it is largely due to the persistence in the
high CO2 environment of dicot species such as, Taraxacum officinale and Plantago major. CO2 enrichment
apparently may slow down the natural succession pattern of our pasture (Potvin & Vasseur 1997). A limitation of our work is that inferences about the community
were based on either general indices, such as species
richness, or individual species responses. Therefore,
they do not address the impact of enriched CO2 on the
community as a whole.
To complement our first study (Potvin & Vasseur
1997), the present paper examines the multivariate
nature of community-environment relationships. Here,
we focus on two questions: (1) how does the complexity of the community change through time and
(2) which environmental variables (in this case, CO2
level, presence of the wall, soil and air temperature)
best correlate with community changes. Direct ordination methods actually represent some of the best tools
to understand species-environment relationships (ter
Braak 1995). Consequently, we use canonical correspondence analyses, to examine the changes in plant
community in relation to CO2 enrichment. Specifically, we determine the linear combinations of environmental variables best describing the observed changes
in the community.
the site had been grazed by cows. To characterize
the experimental site, in 1991, a preliminary vegetation survey was carried out. Some 11 C3 species
were identified, the most abundant being Poa pratensis,
Phleum pratense, Agropyron repens, Plantago major,
Polygonum spp, Taraxacum officinale and Trifolium
repens. In order to examine the effects of increased
CO2 on the pasture, we built four open top chambers, 3 m in diameter and 2.4 m high. We also set-up
two controlled field plots of the same diameter but
without wall (labeled hereof C1 and C2). The design
of the open top chambers was similar to that described
by Rogers et al. (1983) and Reeves et al. (1994) and
is detailed in Potvin and Vasseur (1997). Two of the
open-top chambers were kept at the ambient CO2 concentration of 350 60 l l,1 CO2 (labeled hereof
A1 and A2). The two other chambers were enriched to
650 50 l l,1 CO2 (labeled hereof H1 and H2) during
the entire course of the growing season. Throughout
each growing season, from 1992 to 1994, average daily
temperatures were recorded in each environment using
air and soil temperature probes. Rainfall was collected
after each rain event in each chamber by 2 rainguages.
Vegetation surveys
Vegetation mapping started soon after snowmelt in
May 1992 and continued thereafter every three to four
weeks during three growing seasons. In the present
analyses, we used surveys of May, July and September
for 1992 to 1994. Three quadrats of 0.3 m 0.3 m
were permanently installed in each chamber. A total of
18 quadrats was therefore monitored. To account for
the light and rain shadow effects of the chambers’ wall,
the quadrats were located in a triangular array. At each
mapping, the cover surface occupied by each species
was drawn on a map. Surface area was later estimated
using an AgVisionTM Scientific image analyzer (Version 1.2 Decagon Devices Inc., Pullman WA, USA).
Species identification and distribution at the cover level
were verified by photographs.
Data analyses
Materials and methods
Study site
The study site was located in Farnham, an agricultural region in southern Quebec, Canada (45 170 N,
72 590 W). Before the beginning of the experiment,
To analyze cover data, we used two different but related
multivariate methods. First, a correspondence analysis
(CA) was performed for May 1992 to examine the initial difference in community composition in the three
environments considered: field controls, ambient and
high CO2 open top chambers. Canonical correspondence analyses (CCA) were used to analyze the relation-
33
ships between community composition and environmental factors in each of the three years of monitoring.
In the analyses, we pooled the three quadrats of each
chamber to estimate cover area per species. Specific
cover area per chamber was used as our variable in the
analysis. Because the cover data were normally distributed, untransformed data were used. In the CCA, the
environmental factors studied were concentrations of
CO2 , average air and soil temperatures, and presence
of wall. The month variable was included as a covariable to account for seasonal fluctuations in species
importance.
Both CA and CCA were performed using the computer program CANOCO (version 3.12, ter Braak
1991) which selects the linear combination of environmental variables giving the smallest total residual
sum of squares (ter Braak 1995). Monte Carlo simulations, using 7500 random permutations were used to
test the null hypothesis that the variation found in the
plant community was unrelated to variation measured
in the environmental data (ter Braak 1995).
Results
This first vegetation survey was done, in May 1992,
a week before the beginning of the experiment. These
results were used as a baseline to examine the initial community structure. To assess initial spatial heterogeneity, we performed a correspondence analysis
(CA) on cover data. CA eigenvalues indicated how
much variation in community structure was due to the
importance values of each species. The eigenvalues of
the first and second axes had values of 0.36 and 0.24,
indicating little heterogeneity in species composition.
The ambient (A1) and high (H) chambers were dominated by Polygonum sp. and Taraxacum officinale while
in the controls (C) Phleum pratense and Poa pratensis were dominant (Figure 1). The ambient chamber
A2 presented a different community structure, mostly
dominated by Agropyron repens.
Canonical correspondence analysis (CCA) was
used to follow up changes in the relationship between
community composition and environmental factors
over the course of three growing seasons. In CCA,
eigenvalues indicate how much variation in community
structure is explained by both species composition and
species-environment interaction. From 1992 to 1994,
the eigenvalues of axis 1 increased in importance from
0.289 to 0.413 (Table 1). On the second axis, however,
values decreased from 0.164, in 1992, to 0.094, in
1994. Species-environment correlations were used to
measure the strength of the relationship between species composition and environmental factors. For axis
1, these correlations became gradually more important
over time in explaining the variation in community
pattern (Table 1). A reverse pattern was observed
for axis 2. Similar trends in eigenvalues and speciesenvironment from 1992 to 1994 suggest that changes
in community structure were mainly explained by the
first axis. Monte Carlo permutation tests for repeated
measures indicated that axis representations were significant at p = 0:01 indicating that observed patterns
did not arise by chance.
In 1992, of all the environmental factors, wall was
most correlated to axis 1, followed by air and soil temperatures (Table 1). Axis 2 correlated mostly with CO2
concentration and to a lesser extent with wall. In 1993,
CO2 concentration showed a strong negative correlation with axis 1 while both CO2 and wall had similar
correlation with axis 2. In 1994, wall and CO2 were
the most important environmental variables correlated
to axis 1 but their effects were almost in opposite directions suggesting a possible counteracting effect. As
in 1993, the CO2 inter-set correlation coefficient was
negative. None of the predictor variables showed any
important correlation with axis 2 suggesting that other
factors not accounted for here were coming into play.
Thus, with time, the loading of CO2 in the canonical
correspondence analysis, moved from axis 2 to axis 1,
increased in value indicating that its role in shaping
the pattern of the community became gradually more
important. Conversely, the wall effect on axis 1 slightly
decreased in importance and temperature effect gradually disappeared (Table 1). Therefore, by the end of the
three years, these two variables played a lesser role in
explaining the CCA axes than CO2 .
The cumulative percentage of the variance
explained by species composition increased constantly
through time, and this was true for both axes (Table 1).
In 1992, species were clearly partitioned by the CCA.
Agropyron repens, Polygonum spp. and Capsella
bursa-pastoris had high scores on axis 1 (Figure 2).
These species were apparently favored by warm temperatures and by the presence of a wall. Conversely,
Phleum pratense, Poa pratensis and Trifolium repens
showed negative scores on axis 1. On axis 2, Phleum
pratense and Plantago major were the only species
showing relatively high scores. Consequently they
would be expected to show the largest response to elevated CO2 . Conversely, Trifolium repens, Acer negundo,
Capsella bursa-pastoris, and Leontodon automnale
34
Figure 1. CA ordination diagram of the initial species composition in the chambers of the pasture community of Farnham (spring 1992). The
chambers are identified by the letters A for ambient, H for high CO2 and C for control, the number is the chamber number and the letter s is
for the spring survey. The species names are identified as follow: ar Agropyron repens, pm Plantago major, po Polygonum sp., poa
Poa pratensis, pp Phleum pratense, to Taraxacum officinale and tr Trifolium repens. (Variation explained by axis 1 36% and axis 2
24%).
=
=
=
had very low values on axis 2. In 1993, species also partitioned clearly along axis 1. Quadrats on the positive
side of axis 1 were dominated by Agropyron repens and
Plantago major while Phleum pratense, Poa pratensis
and Taraxacum officinale dominated quadrats on the
negative side of axis 1 (Figure 3). The strong negative
correlation of CO2 with axis 1 suggests that, in 1993,
=
=
=
=
=
=
species clustered in two groups regarding that environmental factor. After 3 years of monitoring, in 1994, one
species, Agropyron repens, stood isolated on the positive side of axis 1 (Figure 4). That species was apparently dominant in the ambient CO2 chambers. Phleum
pratense, Taraxacum officinale, Trifolium repens and
35
Table 1. Eigenvalues and inter-set correlations of standardized environmental variables with the first two CCA
axes, for the three years of the experiment, using plant cover surface areas.
Year
Axis
Eigenvalue
Species-environment correlations
Cumulative percent of variance (%)
– of species data
– of species-environment relation
Inter-set correlations
– Treatment
– Wall
– Air temperature
– Soil temperature
1992
1
0.289
0.925
26.3
56.5
,0.060
0.724
0.235
0.205
to some extend Poa pratensis were the most abundant
species in communities in the other quadrats.
As for species variance, we observed a progression
in the cumulative proportion of the variance explained
by the species-environment interaction for both axes
(Table 1). Looking jointly at mappings of May, July
and September 1992, the three different environments
were clearly classified by CCA into distinct groups
along the two axes (Figure 2). The ambient environments showed positive values on axis 1 and did not
move on axis 2. On the other hand, the quadrats from
the high CO2 open top chambers had negative values
on axis 1 and positive values on axis 2. Finally, the
control quadrats had negative values on both axes. For
1993, the biplot of CCA showed two distinct clusters
of quadrats with the ambient chambers on the positive
side of the X-axis and the control and high CO2 chambers on the negative side (Figure 3). This polarization
of ambient and high CO2 quadrats along axis 1 was
coherent with the strong negative inter-set correlation
for the CO2 environmental factor (,0.429, Table 1).
The control and high CO2 environments further separated along axis 2 with high CO2 quadrats having
positive values along that axis. In 1994, as in 1993,
quadrats clustered on either side of axis 1 with positive
values for the ambient environment (Figure 4). The
other group of quadrats included those of the control
and high CO2 environments. This pattern of separation
was consistent with the negative correlation found for
the CO2 factor (,0.337, Table 1). Quadrats from both
the high CO2 and control environments extended along
axis 2 without easy clustering.
2
0.164
0.838
41.2
88.5
0.834
0.447
0.104
0.141
1993
1
0.319
0.950
29.2
62.5
,0.536
0.396
0.138
0.056
2
0.132
0.813
42.1
88.5
0.460
0.545
0.114
0.168
1994
1
0.413
0.898
32.1
75.0
,0.348
0.382
0.077
0.041
2
0.111
0.453
40.7
92.1
0.061
,0.038
,0.096
,0.032
Discussion
This study documents community trajectories following in situ CO2 enrichment of a natural pasture. It was
designed to monitor community changes through time
and to assess the importance of various environmental factors in causing such changes. Correspondence
analysis (CA) indicated some initial weak spatial heterogeneity among the chambers in 1992. Except for
chamber A2, species composition in the ambient and
high CO2 environments was relatively similar. Since
our goal was to monitor and compare changes in community composition under ambient and elevated CO2 ,
it was essential to start with comparable communities. It is noteworthy to note that Polygonum sp. was
initially more frequent in open-top chambers than in
controls probably mainly due to the disturbance of the
site during the installation of the chambers.
Environmental factors
The independent CCAs performed for each year
showed that CO2 and wall effects had the largest contribution of all experimental factors. Community structure varied in time and was more and more affected by
CO2 . This observation is interesting given the contention in the literature that CO2 might have short-term
effects that would disappear through time (Reynolds
et al. 1996). This contention came from the study
of individual species grown in pots that often show
more CO2 enrichment effects early than late in the life
cycle (Saebo & Mortensen 1995, Potvin & Tousignant
1996). This damping off of CO2 fertilizing effect has
been hypothesized to be due to root binding in pots
(Bazzaz 1990). The fact that soil volume was not lim-
36
Figure 2. Biplot based on the canonical correspondence analysis, done for 1992, of the species composition of the chambers exposed to ambient
(A1 and A2), high CO2 (H1 and H2) or control (C1 and C2) conditions with respect to the environmental conditions: CO2 treatment (treat), wall
effect (wall), soil (soiltemp) and air (airtemp) temperatures (each represented by an arrow). The three surveys are represented by s for spring,
m for mid-summer and f for fall and they are labeled with the chambers (e.g. A1s represents the survey done in chamber A1 in the spring). The
species names are identified as follow: ar Agropyron repens, pm Plantago major, po Polygonum sp., poa Poa pratensis, pp Phleum
pratense, to Taraxacum officinale, tr Trifolium repens, cb Capsella bursa-pastoris, la Leontodon automnale, and an Acer negundo.
(Variation explained by: axis 1 28.9% and axis 2 16.4%, scaling 1).
=
=
=
=
=
=
=
ited in the field might explain the difference in results.
The increased importance through time of CO2 as an
environmental factor suggests that its impact on natural systems will be gradual and might not be detected in
the short-term. This calls for more long-term studies.
=
=
=
=
=
=
The CCA indicated that the strength of the response
to high CO2 was related to a shift in community composition. We hypothesized that long-lived ecosystems,
like forests, would respond more slowly than ecosys-
37
Figure 3. Biplot based on the canonical correspondence analysis, done for 1993, of the species composition of the chambers exposed to ambient
(A1 and A2), high CO2 (H1 and H2) or control (C1 and C2) conditions with respect to the environmental conditions: CO2 treatment (treat), wall
effect (wall), soil (soiltemp) and air (airtemp) temperatures (each represented by an arrow). The three surveys are represented by s for spring, m
for mid-summer and f for fall and they are labeled with the chambers (e.g. A1s represents the survey done in chamber A1 in the spring). Refer
to Figure 2 for the names of the species. (Variation explained by: axis 1 31.9% and axis 2 13.2%, scaling 1).
=
tems having a more rapid species turn-over, such as
grasslands.
Not surprisingly, the presence of a wall was the
second most important environmental factor. Open top
chambers are known to be affecting light availability,
water availability, wind pattern, air and soil temperatures (Drake & Leadley 1991). At the onset of this
=
=
experiment, we expected that air and soil temperatures
would have been significant environmental factors and
they were included in the CCAs. While in the first year,
air and soil temperatures had significant effects on plant
communities, these effects were much smaller than the
wall effect itself. In later years, the effect of temperature became negligible. The wall effect was appar-
38
Figure 4. Biplot based on the canonical correspondence analysis, done for 1994, of the species composition of the chambers exposed to ambient
(A1 and A2), high CO2 (H1 and H2) or control (C1 and C2) conditions with respect to the environmental conditions: CO2 treatment (treat), wall
effect (wall), soil (soiltemp) and air (airtemp) temperatures (each represented by an arrow). The three surveys are represented by s for spring, m
for mid-summer and f for fall and they are labeled with the chambers (e.g. A1s represents the survey done in chamber A1 in the spring). Refer
to Figure 2 for the names of the species. (Variation explained by: axis 1 41.3% and axis 2 11.1%, scaling 1).
=
ently related to either light, wind or water availability
not necessarily to temperature. The presence of a wall
may also limit phenomena as herbivory, seed dispersal, species migration and establishment. Other studies
(Owensby et al. 1996) have reported that the restriction
of herbivores in the chambers and their limited effects
on carbon acquisition and community structure may
=
=
bias the actual changes observed under elevated CO2
levels. In our study, field observations and data collection indicate that herbivores were equally present in
open top chambers and control plots (Wilsey et al., in
prep.).
The absence of strong temperature effect itself is
most interesting because temperature has always been
39
considered as a critical environmental factor. Predictive scheme for community composition, such as the
Holdridge classification, uses temperature as one of
three determining environmental characteristics. In the
present study, we know that air and soil temperatures
were warmer in the open top chambers than in the field
plot. Temperature increment due to the chambers was
on average 1–3 C and peaked at noon in warm days
with differentials as high as 3–5 C. However, despite
this important warming effect, temperature was far
less important than CO2 level at explaining changes in
community structure.
Shifts in community composition
In all the plots, early successional species characteristic of recent disturbance such as Polygonum spp. were
rapidly replaced by late successional species. But since
all this replacement occurred in the first experimental year, no relation can be made to the increase of
CO2 . Parallel experiments have shown that the early
successional species were responsive to disturbance
(Taylor & Potvin, in press). In the next two experimental years, late successional species colonized the
pasture in all the different treatments. However, a most
rapid change in community structure occurred in ambient CO2 chambers. In fact, the biplot of CCA suggests
that the quadrats of the ambient open top chambers
rapidly became depleted in species. Conversely, the
quadrats of high CO2 chambers and control plots still
appeared diversified in 1994. These findings support
our earlier observations on changes in species richness
and dominance that high CO2 concentration apparently
slowed down the natural succession patterns of our pasture community (Potvin & Vasseur 1997).
However, the multivariate approach used in this
study sheds new light on the processes that may be
occurring. The CCA biplots for 1994 provide us with
clearer understanding of the community response to
elevated CO2 . Three species were found to respond
negatively to high CO2 concentration: Agropyron
repens, Phleum pratense and Taraxacum officinale.
On the other hand, Plantago major, Poa pratense
and Trifolium repens responded positively. Except for
Plantago, the classification regarding CO2 sensitivity
was identical in 1993. The positive response of Poa
pratensis uncovered by our CCA was, a priori, surprising in view of earlier results published by Owensby
et al. (1996). They reported that the abundance of
P. pratensis declined under elevated CO2 while that
of C3 forbs increased. The authors however explained
these results as a response to the lack of grazing which
favors C4 grasses. The different response to elevated
CO2 observed in the present study can, in part, be
explained by the absence of C4 grasses in that community. In the analysis, we fail to see a sharp distinction between monocots and dicots. The observed
changes in succession patterns related to a differential
species sensitivity to CO2 . This observation appears to
be supported by results of Solomon (1986). In his study,
he suggested that mid-successional red oak drastically
responded to CO2 , influencing successional patterns
and community structure.
Claims have been made repeatedly in the literature that functional grouping should be used to predict
community responses. The argument is that grouping brings species diversity at a manageable level and
allows one to identify trends otherwise difficult to
detect (Korner 1993b). The absence of clear functional or taxonomic grouping in response to CO2 would
diminish the predictability of community studies. Our
earlier study had looked at species singly and attempted
to identify key species responsible for shifts in community composition under enriched CO2 (Potvin &
Vasseur 1997). Considering these results, we had predicted that dicots might benefit from high CO2 concentration more than monocots. The present data failed to
support that prediction. However, an effective grouping may exist regarding successional status. Two of
the species showing negative responses to high CO2 ,
Agropyron and Phleum are considered as capable to
survive in late succession (Taylor & Aarssen 1988,
1989). These two species are usually better competitors against Poa and Trifolium (Taylor & Aarssen 1989).
Conversely, Trifolium, Poa, Taraxacum and Plantago,
all showing a positive response to CO2 , may be looked
at species present in early succession and their competitive ability to be less than the other late successional
species. The grouping of early and late successional
species might provide an appropriate frame-work to
understand the impact of elevated CO2 on grassland.
In situ monitoring of community changes and in depth
examination are needed to explain the competition pattern between early and late successional species under
elevated CO2 and validate our assertion.
40
Acknowledgements
We would like to thank Leanne Jablonski, Marilou
Beaudet, Lisa Vasil, Jean-Francois Hardy, Kirsteen
Green, and Julia Stewart for their help in the field.
This study was supported by a NSERC postdoctoral
fellowship to L.V. and an operating grant to C.P.
References
Bazzaz, F. A. 1990. The response of natural ecosystems to the rising
global CO2 levels. Ann. Rev. Ecol. Syst. 21: 167–196.
Diaz, S. 1995. Elevated CO2 responsiveness, interactions at the
community level and plant functional types. J. Biogeogr. 22:
289–295.
Drake, B. G. & Leadley, P. W. 1991. Canopy photosynthesis of crops
and native plant communities exposed to long-term elevated
CO2 . Plant, Cell Environ. 14: 853–860.
Drake, B. G., Peresta, G., Beugeling, E. & Matamala R. 1996. Longterm CO2 exposure in a Chesapeake Bay Wetland: Ecosystem
gas exchange, primary production and tissue nitrogen. Pp. 197–
214. In: Koch, G. W. and H. A. Mooney. Carbon dioxide and
terrestrial ecosystems. Academic Press, San Diego.
Field, C. B., Chapin III, F. S., Chiarillo, N. R., Holland, E. A. &
Mooney, H. A. 1996. The Jasper Ridge CO2 experiment: design
and motivation. Pp. 121–146. In Koch, G. W. and H. A. Mooney.
Carbon dioxide and terrestrial ecosystems. Academic Press, San
Diego.
Hocking, P. J. & Meyer, C. P. 1991. Effects of CO2 enrichment
and nitrogen stress on growth, and partitioning of dry matter
and nitrogen in wheat and maize. Aust. J. Plant Physiol. 18:
339–356.
Houghton, J. T., Jenkins, G. J. & Ephraums, J. J. (eds.). 1990.
Climate change – The IPCC Scientific Assessment. Cambridge
University Press, Cambridge. 365 pp.
Kimball, B. A. 1983. Carbon dioxide and agricultural yield: an
assemblage and analysis of 770 prior observations. WCL Report
14, U.S. Water Cons. Lab., Phoenix, AZ.
Körner, C. 1993a. CO2 fertilization: the great uncertainty in future
vegetation development. Pp. 53–70. In: Solomon, A. M. and
H. H. Shugart (eds.) Vegetation dynamics and global change.
Chapman and Hall, New York.
Körner, C. 1993b. Scaling from species to vegetation: the usefulness
of functional groups. Pp. 117–140. In: E. D. Schulze and H. A.
Mooney (eds.) Biodiversity and ecosystem function. Springer,
Berlin.
Körner, C., Diemer, M., Schapp, M. & Zimmerman, L. 1996.
Response of alpine vegetation to elevated CO2. Pp. 177–196.
In: Koch, G. W. and Mooney, H. A. (eds.) Carbon dioxide and
terrestrial ecosystems. Academic Press, San Diego.
Larigauderie, A., Hilbert, D. W. & Oechel, W. C., 1988. Effect of
CO2 enrichment and nitrogen availability on resource acquisition and resource allocation processes in a grass, Bromus mollis.
Oecologia 77: 554–549.
Larigauderie, A., Reynolds, J. F. & Strain, B. R. 1994. Root response
to CO2 enrichment and nitrogen supply in loblolly pine. Plant
Soil 165: 21–32.
Leadley, P. & Korner, C. 1996. Effects of elevated CO2 on plant
species dominance in a highly diverse calcareous grassland.
Pp. 159–176. In: Korner, C. and Bazzaz (eds.) F. A. Carbon
dioxide and terrestrial ecosystems. Academic Press, San Diego.
Luscher, A., Hebeisen, T., Zanetti, S., Hatwig, U. A., Blum,
H., Hendrey, G. & Norberger, G. 1996. Differences between
legumes and non legumes of permanent grassland in their
responses to free-air Carbon dioxide enrichment: its effect on
composition in multispecies mixture. Pp. 287–300. In: Korner,
C. and Bazzaz F. A. (eds.) Carbon dioxide and terrestrial ecosystems. Academic Press, San Diego.
Newton, P. C. D., Clark, H., Bell, C. C., Glascow, E. M., Tate, K. R.,
Ross, D. J., Yates, G. W. & Saggar, S. 1995. Plant growth and
soil processes in temperate grassland communities at elevated
CO2 . J. Biogeogr. 22: 235–240.
Oechel, W. C., Hastings, S. J., Vourlitis, G. L., Jenkins, M. A. &
Hinkson, C. L. 1995. Direct effects of elevated CO2 in chaparral
and mediterranean-type ecosystems. Pp. 58–75. In: Moreno, J.
and W. Oechel (eds.). Global change and mediterranean-type
ecosystems.
Overdieck, D. 1986. Long-term effects on an increased CO2 concentration on terrestrial plants in model ecosystems. Morphology
and reproduction of Trifolium repens L. and Lolium perenne L.
Int. J. Biometeor. 30: 323–332.
Owensby, C. E., Ham, J. M., Knapp, A., Rice, C. W., Coyne, P.
I. & Aven, L.M. 1996. Ecosystem-level responses to tallgrass
prairie to elevated CO2 . Pp. 147–162. In Koch, G. W. and H. A.
Mooney. Carbon dioxide and terrestrial ecosystems. Academic
Press, San Diego.
Potvin, C. & Tousignant, D. 1996. Evolutionary consequences of
simulated global change: genetic adaptation or adaptive phenotypic plasticity. Oecologia 108: 683–693.
Potvin, C. & Vasseur, L. 1997. Long-term CO2 enrichment of a pasture community: species richness, dominance and succession.
Ecology 78: 666–677.
Reeves, D. W., Rogers, H. H., Prior, S. A., Wood, C. W. & Runion,
G. B. 1994. Elevated atmospheric carbon dioxide effects on
sorghum and soybean nutrient status. J. Plant Nutr. 17: 1939–
1954.
Reynolds, J. F., Kemp, P. R., Acock, B., Chen, J.-L. & Moorhead,
D. L. 1996. Progress, limitations, and challenges in modeling
the effects of elevated CO2 on plants and ecosystems. Pp. 347–
380. In: Korner, C. & Bazzaz, F. A. (eds.). Carbon Dioxide and
Terrestrial Ecosystems. Academic Press, San Diego.
Rogers, H. H., Bingham, G. E., Cure, J. D., Smith, J. M. & Surano, K.
A.1983. Responses of selected plant species to elevated carbon
dioxide in the field. J. Environ. Qual. 12: 569–574.
Saebo, A. & Mortensen, L. M., 1995. Growth and regrowth of
Phleum pratense, Lolium perenne, Trifolium repens and Trifolium pratense at normal and elevated CO2 concentration. Agric.,
Ecos. Environ. 55: 29–35.
Schneider, S. H. 1989. The greenhouse effect: science and policy.
Science 243: 771–781.
Solomon, A. M. 1986. Transient response of forests to CO2 -induced
climate change: simulation experiments in eastern North America. Oecologia 68: 567–579.
Stewart, J. & Potvin, C. 1996. Effect of elevated CO2 on artificial
grassland community: competition, invasion and neighborhood
growth. Funct. Ecol. 10: 157–166.
Taylor, D. R. & Aarssen, L. W. 1988. An interpretation of phenotypic
plasticity in Agropyron repens (Graminae). Am. J. Bot. 75: 401–
413.
41
Taylor, D. R. & Aarssen, L. W. 1989. On the density dependence of
replacement-series competition experiments. J. Ecol. 77: 975–
988.
Taylor, K. & Potvin, C. 1998. Competition and disturbance under
enriched CO2 . Can. J. Bot. (in press).
ter Braak, C. J. F. 1991. CANOCO - Version 3.2. Agricultural mathematics Group, 6700 AC Wageningen.
ter Braak, C. J. F. 1995. Ordination. Pp. 91–173. In. Jongman, R.
H. G., ter Braak, C. J. F. & van Tongeren, O. F. R. (eds.).
Data analysis in community and landscape ecology. Cambridge
University Press, Cambridge.
Woodward, F. I., Thompson G. B. & McKee, I. F. 1991. The effects of
elevated concentrations of carbon dioxide on individual plants,
populations, communities and ecosystems. Ann. Bot. 67(suppl.
1): 23–28.