Elevated CO2 enrichment induces a differential biomass response in a mixed species temperate forest plantation


Author for correspondence:

Andrew R. Smith

Tel: +44 1248 383052

Email: a.r.smith@ceh.ac.uk


  • In a free-air carbon dioxide (CO2) enrichment study (BangorFACE), Alnus glutinosa, Betula pendula and Fagus sylvatica were planted in areas of one-, two- and three-species mixtures (= 4). The trees were exposed to ambient or elevated CO2 (580 μmol mol−1) for 4 yr, and aboveground growth characteristics were measured.

  • In monoculture, the mean effect of CO2 enrichment on aboveground woody biomass was + 29, + 22 and + 16% for A. glutinosa, F. sylvatica and B. pendula, respectively. When the same species were grown in polyculture, the response to CO2 switched to + 10, + 7 and 0% for A. glutinosa, B. pendula and F. sylvatica, respectively.

  • In ambient atmosphere, our species grown in polyculture increased aboveground woody biomass from 12.9 ± 1.4 to 18.9 ± 1.0 kg m−2, whereas, in an elevated CO2 atmosphere, aboveground woody biomass increased from 15.2 ± 0.6 to 20.2 ± 0.6 kg m−2. The overyielding effect of polyculture was smaller (+ 7%) in elevated CO2 than in an ambient atmosphere (+ 18%).

  • Our results show that the aboveground response to elevated CO2 is affected significantly by intra- and interspecific competition, and that the elevated CO2 response may be reduced in forest communities comprising tree species with contrasting functional traits.


Forests occupy one-third of the land surface of the Earth, and account for almost one-half of the carbon (C) stored in the terrestrial biosphere (Schlesinger & Lichter, 2001). In a summary of studies conducted to investigate the effects of increased atmospheric CO2 on forest C cycles, Norby et al. (2005) calculated that an enrichment of 200 ppm CO2 above the current ambient CO2 level caused a 23% median increase in forest net primary productivity (NPP). However, interactions with other environmental factors may dampen such a response at larger temporal or spatial scales (Leuzinger et al., 2011). Nevertheless, increasing atmospheric CO2 concentrations may fundamentally alter forest ecosystem functioning by altering species growth, resource use and community interactions (Eamus & Jarvis, 1989). As forests are inextricably linked to the global C cycle, elevated CO2-driven environmental change may impact upon global C storage in phytomass, complex biogeochemical feedback mechanisms and, ultimately, long-term C sequestration in soils.

Empirical studies on woody plants exposed to elevated atmospheric CO2 have demonstrated that growth and aboveground biomass production increase, but that there is a considerable variation in response (Curtis & Wang, 1998). The observed variation in responses to elevated CO2 has been attributed to a large number of confounding factors, such as the length of study, interactions with other environmental stresses, plant functional group, species morphological physiology (Poorter, 1993), symbiotic associations (Godbold et al., 1997) and community dynamics (Kozovits et al., 2005). Recent research efforts have focused on the study of the whole ecosystem responses in near-natural conditions, chiefly achieved by employing free-air carbon dioxide (CO2) enrichment (FACE) facilities (Hattenschwiler et al., 2002; Karnosky et al., 2003; Körner et al., 2005; Hoosbeek et al., 2011). Körner (2006) has suggested that elevated CO2 studies should be divided into the following two types: high abundance of major resources other than C – ‘decoupled systems’ (type I); and near to steady-state nutrient cycling and full canopy development – ‘coupled systems’ (type II). Type I systems include the present study, aspen FACTS II FACE (Karnosky et al., 2003) and EuroFACE (Calfapietra et al., 2003) experiments. The remaining three (type II) experiments have used CO2 enrichment in stands with an already closed canopy. The Oak Ridge (Norby et al., 2002) and DukeFACE (Oren et al., 2001) experiments both started enrichment c. 10–20 yr after planting, whereas, at the Basel Web-FACE (Körner et al., 2005), enrichment was conducted in a mature deciduous forest comprising four species > 100 yr old. Using data from four of these studies (DukeFACE, FACTS II FACE, Oak Ridge and EuroFACE), Norby et al. (2005) demonstrated that an enrichment of 200 ppm CO2 above the current ambient CO2 level caused a 23% median increase in forest NPP. This conclusion was largely based on the initial response of forest ecosystems to elevated CO2. Subsequent investigations have shown that this response may not be maintained over a longer time horizon (Norby et al., 2010), as the response to elevated CO2 has been found to decline (Norby et al., 2010) or be maintained (Drake et al., 2011; Zak et al., 2011) after 10–11 yr of exposure. In both of these examples, the response to elevated CO2 was probably mediated by nitrogen (N) availability. The decline in response to elevated CO2 was attributed to N limitation (Norby et al., 2010), whereas no change in response was a result of greater N cycling (Zak et al., 2011). The comparison of these two studies clearly demonstrates that nutrient availability, in particular N, is a strong factor mediating the response of woody plants to elevated CO2.

Much of the research investigating species diversity, ecosystem functioning and productivity has been focused in grasslands (Hooper et al., 2005). Many experiments have shown a positive relationship between productivity and increased biodiversity (Tilman et al., 1996, 1997). Fornara & Tilman (2009) suggested that the increased productivity of N-limited, species-rich plant communities is dependent on the seasonal accumulation of root N pools by N-fixing plants. The importance of the incorporation of N-fixing plants in the facilitation of greater plant community productivity was also supported by Hooper & Dukes (2004), but it was argued that N fixation is not the only mechanism explaining the overyielding of species-rich communities. Elevated CO2 has been found to stimulate symbiotic N fixation in several studies (e.g. Hungate et al., 1999; Schortemeyer et al., 2002), and the incorporation of N-fixing plants to facilitate the N dynamics of co-occurring species with elevated CO2 was explored by Lee et al. (2003), who found that, in nine different grassland species, assemblages incorporating N-fixing Lupinus did not facilitate a larger community growth response to elevated CO2.

In forests, controversy surrounding the benefits of mixed species stand productivity dates back to the 18th century (Hartig, 1791), with the silvicultural practice of mixed species forests being subject to much conjecture. Only recently have rigorous scientific studies been initiated to elucidate the precise mechanisms mediating the productivity differences of trees grown in polyculture (Pretzsch, 2005). For example, in southern Germany, mixed stands of Fagus sylvatica and Picea abies produced up to 59% more aboveground biomass than adjacent pure stands (Pretzsch & Schütze, 2009). By contrast, Jacob et al. (2010) found decreases in aboveground biomass of F. sylvatica with increasing species richness in comparison with F. sylvatica in monoculture. Early on, most research on forest diversity focused on one or two tree species, but recent studies have included more species in an attempt to verify the applicability of grassland findings to forest stands (DeClerck et al., 2006; Vila et al., 2007; Paquette & Messier, 2010). In large-scale investigations, support has been found for the assertion that increased tree diversity leads to increased biomass production (Vila et al., 2007; Paquette & Messier, 2010). The studies of both Vila et al. (2007) and Paquette & Messier (2010) used databases originating from national forest inventories, whilst taking into account the effects of environment. Paquette & Messier (2010) used 12 000 permanent forest plots in boreal and temperate forest in Canada, and showed a strong positive and significant effect of tree biodiversity on aboveground productivity. The study of Vila et al. (2007) used over 8000 permanent forest plots in Mediterranean forests in Catalonia, and showed a mean 30% higher wood production in mixed forest compared with monospecific stands, and a production increase from 23% in two-species stands to 59% in five-species stands. In a meta-analysis of 54 forest studies investigating diversity–productivity relationships, Zhang et al. (2012) showed a 24% higher productivity in polycultures than in monocultures, with most of the variation accounted for by evenness, the heterogeneity of shade tolerance, species richness and stand age, in decreasing order of importance. Recently, high plant diversity has been shown to be required to maintain ecosystem function and services through time (Isbell & Wilsey, 2011); however, the role of tree diversity in ecosystem productivity, resistance and resilience is still poorly investigated (DeClerck et al., 2006). In the case of resistance to drought, DeClerck et al. (2006) found that the relative percentage of different species was more important than the species richness per se. Differing species resistance to drought can change the competitive relationship between the species, and may thus result in altered species composition. Reich et al. (2001) showed that the enhancement of biomass accumulation in response to elevated levels of CO2 was smaller in species-poor than in species-rich assemblages of herbaceous plants. However, although it has long been known that tree seedlings of co-occurring species show differing response to CO2 (Bazzaz & Miao, 1993), the influence of elevated CO2 on tree competition and the influence of tree biodiversity on community response to CO2 have not been investigated.

The objectives of this work were to investigate the effects of elevated CO2 (580 μmol mol−1) on the species and community response of monocultures and polycultures of tree mixtures under field conditions. Using a FACE system, we investigated the aboveground response of monocultures and a three-species polyculture of Alnus glutinosa, Betula pendula and F. sylvatica to elevated CO2 over 4 yr. We tested the hypothesis that interspecific competition modifies the response of tree species to elevated CO2.

Materials and Methods

Site description

The BangorFACE experimental site was established in March 2004 at the Bangor University research farm (53°14′N, 4°01′W) on two former agricultural fields with a total area of 2.36 ha. Both fields were originally pastures: one field has been used for small-scale forestry experiments for the last 20 yr; the other field was ploughed and planted with oil seed rape in 2003. Climate at the site is classified as Hyperoceanic with a mean annual temperature in 2005–2008 of 11.5°C and an annual rainfall of 1034 mm. Soil parent material is postglacial alluvial deposits from the Aber river which comprises Snowdonian rhyolitic tuffs and lavas, microdiorites and dolerite in the stone fractions and Lower Palaeozoic shale in the finer fractions. Soil is a fine loamy brown earth over gravel (Rheidol series) and is classified as Fluventic Dystrochrept (Teklehaimanot et al., 2002). Soil texture is 63% sand, 28% silt and 9% clay, N content in the top 30 cm is 2.6% with a C : N ratio of 10.5. The topography consists of a shallow slope of c. 1–2° on a deltaic fan. The site aspect is northwesterly, with an altitude of 13–18 m asl. The depth of the water table ranges between 1 and 6 m.

Eight octagonal plots, four ambient and four CO2 enriched, were established at the site, creating a 2 × 4 factorial block design across the two fields. We used three tree species (Alnus glutinosa (L.) Gaertner, Betula pendula Roth. and Fagus sylvatica L.) selected for their contrasting shade tolerance, successional chronology and to represent a range of taxonomic, physiological and ecological types. A replacement series design (with the inter-tree spacing constant between treatments) was selected because of the experiments objective of being realistic in reflecting the practical realities of how forests comprising monocultures or mixtures of potential canopy tree species could be established (Jolliffe, 2000). The site was planted with 60-cm saplings of each species with an inter-tree spacing of 0.8 m, giving a density of 15 000 trees ha−1. A systematic hexagonal planting design (Aguiar et al., 2001) was used to maximize the mixing effect, so that, in the three-species polyculture subplots, each tree was surrounded by nearest neighbours of two conspecific individuals and one and three individuals of the other two species, respectively, resulting in each tree having six equidistant neighbours. Each plot was divided into seven planting compartments and planted in a pattern creating areas of one-, two- and three-species mixtures (Fig. 1). The present study makes use of observations originating from three single-species subplots containing nine trees of B. pendula, A. glutinosa and F. sylvatica, and a fourth subplot which contained a species-balanced polyculture of all three species. The planting pattern of each pair of control and elevated CO2 plots was rotated by 90° to avoid potential artefacts introduced by microclimate, soil and uneven growth rates of the different species. Each plot was surrounded by a 10-m border of B. pendula, A. glutinosa and F. sylvatica planted at the same density. The remaining field was planted at a 1-m spacing (10 000 trees ha−1) with a mixture of birch (B. pendula), alder (A. glutinosa), beech (F. sylvatica L.), ash (Fraxinus excelsior L.), sycamore (Acer pseudoplatanus L.), chestnut (Castanea sativa Mill.) and oak (Quercus robur L.). To protect the saplings, the entire plantation was fenced.

Figure 1.

Layout of ambient and elevated CO2 plots; a, Alnus glutinosa; b, Betula pendula; F, Fagus sylvatica. Each plot contains 27 trees per species. The monoculture species area is indicated by a solid lined oval and the three-species polyculture plot is indicated by a dot-dashed line oval.

Eight steel towers were erected around each plot to delineate the experimental area and to provide supporting infrastructure for the CO2 enrichment system in the treatment plots. Ambient CO2 control plots were identical to the treatment plots, but for the absence of CO2 injection piping, to ensure that any infrastructure-introduced artefacts were applied to both the treatment and control. CO2 enrichment was carried out using high-velocity pure CO2 injection (Okada et al., 2001). In the first two growing seasons, CO2 was delivered from a horizontal pipe held at canopy level. In growing seasons 3 and 4, an additional pipe suspended 2 m below the canopy pipe was added to provide adequate enrichment throughout the canopy. The control of CO2 delivery was achieved using equipment and software modified from EuroFACE (Miglietta et al., 2001). The target concentration in the elevated CO2 plots was ambient plus 200 ppm. The elevated CO2 concentrations, measured at 1-min intervals, were within 30% deviation from the pre-set target concentration of 580 ppm CO2 for 75–79% of the time during the photosynthetically active (daylight hours between budburst until leaf abscission) period of 2005–2008. Vertical profiles of CO2 concentration, measured at 50-cm intervals through the canopy, showed a maximum difference of + 7% from the reference value obtained at the top of the canopy. The effect of CO2 fumigation on the diameter and height of the trees grown within the plots was not modified by the distance from the CO2 delivery pipe (Supporting Information Fig. S1). The CO2 used for enrichment originated from natural gas and had a δ13C of −39‰.

Biometric measurements

Tree height and stem diameter at 22.5 cm were measured after tree establishment in March 2005 and then in February of each following year during CO2 enrichment (2006–2009). Tree measurements were taken during the winter dormant phase to prevent growth-introduced variation. Tree height was determined using a telescopic pole, and two measurements of diameter were taken perpendicular to each other using digital vernier callipers. To account for the elliptical stem shape, a geometric mean was calculated. As the initial tree height was < 137 cm, it was only possible to measure the diameter at breast height (DBH) in subsequent years as the stand developed.

Allometric relationships, stem volume index

Two trees of each species were selected for destructive harvest from the downwind buffer zone of each treatment and control plot. The selection of trees for each species was based on average height and diameter data collected during the previous season. Tree height and stem diameter at 22.5 cm were measured and the trees were excavated to a root diameter of 3–4 mm, and then separated into leaves, branches stems and roots. Roots were washed free of adhering soil and stems were cut into 15–20-cm sections, oven dried at 80°C for 72 h and weighed. As a consequence, a power regression of stem diameter and woody biomass was used to explain the allometric relationship for each species studied, as height was not found to contribute significantly to any of the allometric models tested (Eqn 1). Equation 2 shows the biomass allometric equation in its linear form, where D is the stem diameter at 22.5 cm and with the power regression scaling coefficients a (amplitude) and b (exponent).

display math(Eqn 1)
display math(Eqn 2)

The stem volume index (basal diameter2 × height) was calculated and correlated against the allometrically determined biomass to test the accuracy of the predicted biomass values.


To determine the effect of growing species in polyculture, the total measured aboveground woody biomass values in the three-species polyculture subplots were compared with those of a theoretical mixture calculated from the biomass of each species growing in the monoculture subplots. Equation 3 shows the theoretical mixture biomass calculation based on the stem number contribution of each species to the polyculture, where BSpecies is the biomass component contributing to the mixture. The theoretical basis of this calculation is directly analogous to the Relative Yield of Mixtures index used to quantify the effects of competition (Wilson, 1988). The use of Eqn 3 in this experiment is comparable with the Relative Yield Total (Weigelt & Jolliffe, 2003).

display math(Eqn 3)

Leaf N contents

Leaf N contents were measured on five fully mature, but otherwise unaltered, leaves collected throughout the canopy of each species subplot (120 leaves in total) in 2006 (Ahmed, 2006), 2007 (Anthony, 2007) and 2008 (Millett et al., 2012).

Leaf area index

From the beginning of leaf senescence, fallen leaf litter was collected weekly using litter baskets with an area of 0.11 m2 until all leaves had abscised (October to December). A litter basket was located in each of the monoculture subplots and the three-species polyculture subplot (four in each experimental plot). Litter was washed in a laboratory, sorted by species and then dried at 80°C for 24 h. The dry weight of each species was determined and recorded for each species subplot. Juvenile F. sylvatica were excluded from the calculations as the beech trees retained the foliage until bud burst the following season. The leaf area index was calculated according to McCarthy et al. (2007). The specific leaf area was calculated from fresh leaves collected during 2006 and dried archived leaves collected in 2007. Measurements of leaf area were made with a LI 3000A portable area meter (LI-COR, Lincoln, NE, USA). Immediately following area measurement, leaves were dried at 80°C for 24 h, and weighed to determine the specific leaf area. The leaf area indices obtained were then scaled to calibrate for the different number of trees per species per ground area in the monoculture and polyculture plots.

Statistical analysis

Regression fitting was conducted using SigmaPlot v11.0 (Systat Software Inc., Chicago, IL, USA). All statistical procedures were undertaken with SPSS 17.0 (SPSS Inc., Chicago, IL, USA) with < 0.05 used as the limit for statistical significance. To avoid pseudoreplication, the mean woody biomass per unit area (g m−2) was calculated from the trees contributing to the single- and mixed-species plots and data were subjected to repeated-measures ANOVA for time series analyses using the plots as replicates (= 4); equality of variance was tested using Mauchly's test of sphericity. A General Linear Model was used to calculate univariate analysis of variance for data determined at the conclusion of the experiment. Data were tested for normality using the Shapiro–Wilk's test and homogeneity of variance was determined using Levene's test. Diameter distributions were compared by fitting a normal distribution into the frequency data and testing for differences in the peak diameter by extra sum-of-squares F test.


Stem diameter and tree height

At the conclusion of the experiment, the treatment effect on diameter was most pronounced in single-species subplots, with the largest effect of + 14% observed in A. glutinosa (ambient, 49.1 mm; elevated CO2, 55.9 mm; = 0.007; Table 1). Elevated CO2 did not change significantly the stem diameter of B. pendula or F. sylvatica.

Table 1. Overall effect of elevated CO2 and probability of significance at the end of the 2008 growing season after 4 yr of fumigation
Planting patternSpeciesDiameterHeight
Effect (%)ProbabilityEffect (%)Probability
  1. The effect of elevated CO2 is expressed as a percentage relative to control plot measurements of tree diameter at 22.5 cm and height of Alnus glutinosa, Betula pendula and Fagus sylvatica. Trees were grown in monocultures and in a three-species polyculture. Statistically significant results are shown in bold and denoted by asterisks (**, < 0.01).

Mono A. glutinosa 14 0.007 ** 30.706
B. pendula 60.14600.935
F. sylvatica 60.60300.965
Poly A. glutinosa 40.61810.837
B. pendula 50.61430.728
F. sylvatica −50.483−120.333

We assessed the treatment effects on the diameter distributions of all species by grouping all measured trees into 10 diameter classes with 10-mm step increments. For A. glutinosa, B. pendula and F. sylvatica, the most frequent diameter classes were 50–60, 40–50 and 20–30 mm, respectively. The diameter class distribution of B. pendula and F. sylvatica grown in monoculture was not altered by elevated CO2 enrichment (Fig. S2). However, in A. glutinosa, there was a shift towards larger diameter boles under elevated CO2, where 39% of trees had a diameter > 50–60 mm, which was in contrast with ambient plots, where only 11% of trees were in this diameter class (= 0.021). In polyculture, the mean of the diameter distribution was not altered by elevated CO2 in any of the species. Tree height was unaffected by elevated CO2 enrichment in either mono- or polyculture at the end of the observations (Table 1).

Allometric equations

Height and diameter data gathered from trees in the vicinity of elevated and ambient CO2 plots were subjected to a stepwise biomass prediction regression. Height was excluded during this analysis, as it did not contribute significantly to the regression model. Ultimately, a simple power regression of diameter predicted biomass with the greatest accuracy. Power function scaling coefficients for the three species utilized in this study are shown in Table 2. There were no changes in allometry as a result of elevated CO2 at this stage of tree development and, subsequently, all species-specific data were pooled to produce three allometric relationships with coefficients of variation ranging from 0.78 to 0.85. Strong correlations between stem volume index and predicted biomass confirmed the accuracy of the predictions for A. glutinosa (R2 = 0.98) and B. pendula (R2 = 0.99), but highlighted a small underestimate of the predicted F. sylvatica biomass in elevated CO2 plots (R2 = 0.88).

Table 2. Allometric relationship power function scaling coefficients for the three species utilized in this study determined by regression analysis
Species a b R 2
Alnus glutinosa 0.52002.0200.85
Betula pendula 0.44142.1630.86
Fagus sylvatica 0.68851.8530.78

Aboveground biomass in monoculture and polyculture

Making use of the allometric equations to calculate tree aboveground woody biomass, we showed that species grown in monoculture responded to elevated CO2 treatment more strongly than those grown in the three-species polyculture. Figure 2 and Table 3 detail the relationship between time and biomass accrual for all species in ambient and elevated atmospheric CO2. Under ambient CO2, both A. glutinosa and B. pendula accumulated aboveground woody biomass more rapidly in the polyculture than in the monocultures. The influence of elevated CO2 on aboveground woody biomass production varied between species and years. Unsurprisingly, in an expanding system, sampling year explained the greatest amount of variation in a repeated-measures ANOVA model, being highly significant for all species in both monoculture and polyculture (Table 4). There were no significant year × treatment interactions for any species in the polyculture or for B. pendula and F. sylvatica in the monocultures. However, there was a significant year × treatment interaction for A. glutinosa (= 0.008). Elevated CO2 treatment produced a significant effect on aboveground woody biomass in A. glutinosa grown in monoculture during 2005 (P = 0.022), 2007 (P = 0.025) and 2008 (P = 0.002, Table 3). In polyculture, no statistically significant effects of elevated CO2 were found.

Table 3. Effect of CO2 enrichment on aboveground woody biomass of Alnus glutinosa, Betula pendula and Fagus sylvatica when grown in monoculture and in a three-species polyculture
  1. Statistically significant results are given in bold and denoted by an asterisk (*, < 0.05).

Mono A. glutinosa 29% * 25% 28% * 32% * 29%
B. pendula 27%13%14% 9%16%
F. sylvatica 28%33%20% 9%22%
Poly A. glutinosa 13%12% 3% 8%10%
B. pendula 4%8% 6% 7%6%
F. sylvatica 2%5% 2%−8%0%
Table 4. F values and probability of significance for sampling year and sampling year × CO2 treatment interactions from a repeated-measures ANOVA of tree diameter, height and aboveground woody biomass for Alnus glutinosa, Betula pendula and Fagus sylvatica grown in both monoculture and polyculture
Planting patternSpeciesSource of variationDiameterHeightBiomass
F Probability F Probability F Probability
  1. Statistically significant results are given in bold and denoted by asterisks (*, < 0.1; **, < 0.05; ***, < 0.001).

Mono A. glutinosa Treatment7.216 0.036 ** 0.6810.4413.920 0.095 *
Year506.525 <0.001 *** 512.615 <0.001 *** 253.786 <0.001 ***
Year × treatment2.6890.0550.6030.6645.546 0.008 **
B. pendula Treatment1.8080.2270.0760.7921.0640.342
Year428.974 <0.001 *** 394.712 <0.001 *** 113.580 <0.001 ***
Year × treatment0.6100.6590.1930.9400.0780.971
F. sylvatica Treatment1.0170.3520.5760.4770.4450.529
Year123.828 <0.001 *** 200.403 <0.001 *** 47.454 <0.001 ***
Year × treatment0.4540.7691.1240.3680.2500.860
Poly A. glutinosa Treatment0.3190.5920.1100.7510.2710.622
Year377.886 <0.001 *** 934.984 <0.001 *** 125.788 <0.001 ***
Year × treatment0.8180.5260.2230.9230.1790.909
B. pendula Treatment0.4400.5320.3680.5660.3550.573
Year223.473 <0.001 *** 351.368 <0.001 *** 64.346 <0.001 ***
Year × treatment0.2450.9100.0880.9850.0830.969
F. sylvatica Treatment0.0030.9580.6950.4360.2700.622
Year205.838 <0.001 *** 116.937 <0.001 *** 101.798 <0.001 ***
Year × treatment0.6510.6320.9500.4531.2400.325
Figure 2.

Mean ± SE aboveground woody biomass for the species grown in monoculture subplots under elevated and ambient CO2 for 4 yr. Aboveground woody biomass was calculated from the allometric relationship determined from whole tree harvesting in 2006. Open symbols, elevated atmospheric CO2; closed symbols, ambient CO2.

The temporal fluctuation in the treatment effect of B. pendula and F. sylvatica grown in monoculture and polyculture became more apparent when the aboveground woody biomass NPP for each year was calculated (Table 5). In the monocultures, A. glutinosa showed a positive treatment effect throughout the 4 yr of enrichment, whereas, in B. pendula, both positive and negative treatment effects were found. In F. sylvatica, aboveground woody biomass NPP was initially stimulated under elevated CO2, but the effect turned strongly negative in 2008. In polyculture, A. glutinosa showed a strong positive treatment effect on aboveground woody biomass for all years except 2007. Similarly, in B. pendula, a positive treatment effect on aboveground woody biomass was shown for all years. By contrast, a negative effect of elevated CO2 was shown on the accumulation of aboveground woody biomass in F. sylvatica in all years except 2006. When pooling the species contributing to the polyculture over all years, there was no effect of elevated CO2 on overyielding in the mixture (= 0.094), nor did we observe any modification of the CO2 fertilization when growing trees in monoculture or polyculture (= 0.192, Fig. 3).

Table 5. Effect of CO2 enrichment on annual production of aboveground woody biomass in Alnus glutinosa, Betula pendula and Fagus sylvatica when grown in monocultures and polyculture with other species
  1. Statistically significant results are given in bold and denoted by an asterisk (*, < 0.05).

Mono A. glutinosa 35%20% 33% * 59% * 37%
B. pendula 32%−7%15%−8%8%
F. sylvatica 30%38%−4%−31%9%
Poly A. glutinosa 27%13%−13%29%14%
B. pendula 6%13%4%7%8%
F. sylvatica −2%9%−20%−38%−13%
Figure 3.

Overyielding (a) and CO2 fertilization (b) effects in pooled data for Alnus glutinosa, Betula pendula and Fagus sylvatica. Overyielding was calculated as the aboveground woody biomass measured in polyculture over that predicted from monocultures. Predicted biomass was calculated by taking one-third of the biomass observed in each species when grown in monoculture. CO2 fertilization was calculated as the biomass in elevated over ambient CO2 treatments. Values are mean ± SE,= 4.

At the conclusion of the experiment with all species pooled, aboveground woody biomass reached 16.5 ± 0.8 kg m−2 in ambient CO2 plots and 19.3 ± 0.4 kg m−2 in elevated CO2 plots, a significant increase of 17% (P = 0.022). The contribution of aboveground woody biomass within the elevated CO2 plots followed the order B. pendula (10.1 ± 0.0 kg m−2), A. glutinosa (8.6 ± 0.6 kg m−2) and F. sylvatica (0.6 ± 0.0 kg m−2). A significant 16% (= 0.046) increase in aboveground woody biomass was observed in B. pendula in response to CO2 treatment. Pooling the values for each species in the monocultures, the aboveground woody biomass was 12.9 ± 1.4 kg m−2 in ambient and 15.2 ± 0.6 kg m−2 in elevated CO2 treatments. Polyculture aboveground woody biomass reached 18.9 ± 1.0 kg m−2 in ambient and 20.2 ± 0.6 kg m−2 in elevated CO2 treatments. This resulted in an increase in aboveground woody biomass under elevated CO2 of 18% in monoculture and 7% in polyculture.

To summarize, pooled aboveground woody biomass was affected significantly by elevated CO2 (P = 0.022). We also observed a significant positive effect of species mixture (P = 0.001), but the interaction was not significant (P = 0.534).

Leaf N content and aboveground NPP

Over the course of the experiment, leaf N contents were not affected significantly by elevated CO2 (Table 6). However, we observed a strong increase in foliar N content in time (< 0.001), combined with significant differences between species (< 0.05) over the period 2006–2008 (Fig. S3). Leaf nitrogen use efficiency (NUE), defined as unit of aboveground NPP per unit of foliar N content (Yasumura et al., 2002), fluctuated in time (Fig. 4) and was increased significantly by elevated CO2 from 44.0 to 53.7 g m−2 mg g−1 averaged for all species and years (= 0.017). As a result of data unavailability, we could only establish the effect of mixture on leaf NUE in 2008. Four years into the experiment, growing species in polyculture as opposed to monoculture increased significantly the overall leaf NUE from 23.4 to 38.6 g m−2 mg g−1 (= 0.022, Fig. 5). However, there was no effect of mixture or elevated CO2 on leaf NUE in individual species in 2008.

Table 6. Leaf nitrogen content (% ± SEM) of Alnus glutinosa, Betula pendula and Fagus sylvatica grown under ambient and elevated CO2
  1. Values in bold denote a significant CO2 effect at < 0.05.

  2. Sources: aAhmed (2006); bAnthony (2007); cMillett et al. (2012).

A. glutinosa 4.1 ± 0.5 3.1 ± 0.2 3.4 ± 0.23.7 ± 0.24.1 ± 0.03.9 ± 0.1
B. pendula 3.0 ± 0.12.7 ± 0.12.6 ± 0.52.5 ± 0.23.7 ± 0.13.8 ± 0.2
F. sylvatica 2.0 ± 0.12.0 ± 0.1 1.6 ± 0.5 3.7 ± 0.1 3.0 ± 0.13.1 ± 0.1
Figure 4.

Leaf nitrogen use efficiency (NUE), defined as aboveground net primary production per unit of leaf N content. Leaf N data for (a) Alnus glutinosa, (b) Betula pendula and (c) Fagus sylvatica are from Ahmed (2006), Anthony (2007) and Millett et al. (2012), respectively. Values are mean ± SE,= 4.

Figure 5.

Leaf nitrogen use efficiency (NUE), defined as aboveground net primary production per unit of leaf N content, in trees grown in monocultures and in a three-species mixture. Leaf N data for (a) Alnus glutinosa, (b) Betula pendula and (c) Fagus sylvatica are from Ahmed (2006), Anthony (2007) and Millett et al. (2012), respectively. Values are mean ± SE,= 4.

Leaf area index

Repeated-measures ANOVA showed a significant year × species interaction for species grown in monoculture (< 0.05) and polyculture (< 0.001; Table 7). The response of the leaf area index to elevated CO2 when species were grown in monoculture was a mean increase of 32% in B. pendula and a mean decrease of 6% in A. glutinosa. During the 4 yr of CO2 enrichment, the leaf area index of B. pendula was in the range 1.1–3.2 m2 m−2 in ambient CO2 and 0.8–4.0 m2 m−2 in elevated CO2 plots, whereas that of A. glutinosa was in the range 1.4–7.6 m2 m−2 in ambient CO2 and 1.4–8.2 m2 m−2 in elevated CO2 plots (Fig. 6). Elevated CO2 initially increased the leaf area index of B. pendula by 37%; however, this effect gradually declined to 24% in 2007, recovering to 32% by the conclusion of the experiment. In both mono- and polyculture, the peak leaf area indices in Aglutinosa and B. pendula were recorded in 2007, which was preceded by a severe drought, summer crown defoliation and leaf re-flushing during August of 2006; a strong decline in leaf area index immediately followed in 2008 in monocultures. During 2008 in polyculture, the leaf area indices were 4.6 and 4.4 times greater than in monoculture in ambient atmosphere for B. pendula and A. glutinosa, respectively, whereas, in monoculture, the leaf area indices were 6.1 and 4.6 times greater than in elevated CO2 for B. pendula and A. glutinosa, respectively.

Table 7. Analysis of the leaf area index of trees grown in monoculture and in a three-species polyculture under ambient and elevated CO2 between 2005 and 2008 using repeated-measures ANOVA
Source of variationMonoculturePolyculture
F valueProbabilityF valueProbability
  1. Statistically significant results are given in bold and denoted by asterisks (*, < 0.05; ***, < 0.001).

Year44.478 <0.001 *** 33.451 <0.001 ***
Year × treatment1.3180.2830.1060.956
Year × species3.715 0.020 * 19.008 <0.001 ***
Year × treatment × species0.4230.7371.1740.333
Figure 6.

Measured leaf area index for Alnus glutinosa and Betula pendula grown under ambient and elevated CO2 in monoculture (a) and in polyculture (b). Values are mean ± SE.


Allometric relationships have commonly been used to estimate the biomass of aboveground compartments. The allometric coefficients generated in this study were broadly similar to previously published coefficients (Hughes, 1971; Bartelink, 1997; Pajtik et al., 2011), with the exception of F. sylvatica. The dimorphic growth characteristics of juvenile F. sylvatica under different light regimes during canopy development may explain the difference observed (Delagrange et al., 2006). The application of species and site-specific allometric relationships is likely to be valid for A. glutinosa and B. pendula. However, the relationship for F. sylvatica appears to be slightly weaker and may benefit from closer examination of the differences in morphology when trees are shade suppressed and growing in full light.

In this study, aboveground woody biomass accumulation in A. glutinosa and B. pendula was greater in polyculture than in monocultures. In species-diverse communities, the complementary use of resources may lead to higher yields than in monocultures (Loreau & Hector, 2001). Differences in the tree species life history character traits, such as crown structure, rooting depth, shade tolerance, phenology and photosynthetic light response, may allow for differential access to resources (Kelty, 1992). If the chosen species occupying the same site differ substantially in these characteristics, they may capture site resources more completely or use resources more efficiently to produce biomass. Species with contrasting trait characteristics can be described as having complementary resource use (Haggar & Ewel, 1997) or good ecological combining ability (Harper, 1977). However, it should be noted that complementarity may not necessarily result in a positive effect on productivity; antagonistic interactions (negative complementarity) between species may also occur as a result of character trait interferences that may lower the productivity of species mixtures over those expected from monocultures (Wardle et al., 1998; Loreau & Hector, 2001; Eisenhauer, 2012). In their study, Paquette & Messier (2010), in an analysis of naturally occurring tree biodiversity, showed a strong positive effect of biodiversity on tree productivity. They further suggested that, in the more productive environment of temperate forest, competitive exclusion is the most probable outcome of species interactions, but, in the more stressful environment of boreal forest, beneficial interactions, such as niche partitioning and facilitation, may be more important.

In our temperate forest mixture, we used two pioneer species and a late successional species that strongly differ in their functional traits. Betula pendula is a light-demanding, early successional pioneer species which casts little shade and rapidly occupies open areas as a result of fast juvenile growth (Fischer et al., 2002). Alnus glutinosa is an N-fixing, water-demanding pioneer species, also with high juvenile growth rates (Braun, 1974). The root system of A. glutinosa is adapted to wet soils, with many vertically growing sinker roots that may reach a depth of 5 m (Claessens et al., 2010). In mixed forests, its limited height growth and shade intolerance prevent it from dominating in late successional forest. Lastly, Fagus sylvatica is shade tolerant and slow growing when juvenile (Ellenberg et al., 1991), can persist in the understorey and often dominates late successional forest. The higher polyculture productivity in our 4-yr-old plantation suggests that the dominant pioneer species A. glutinosa and B. pendula are partitioning canopy space made available by F. sylvatica. However, the flattening of the diameter class distribution in B. pendula, but not in A. glutinosa, suggests that some B. pendula are being excluded. In our study, we did not systematically determine crown architecture, but observed that, in polyculture, both B. pendula and shorter A. glutinosa had deeper crowns. Indeed, we saw higher leaf area index in A. glutinosa and B. pendula in polyculture relative to monocultures, but no difference in stem height, which suggests alteration of crown architecture between monoculture- and polyculture-grown trees. Claessens et al. (2010) suggested that A. glutinosa grown in monoculture produces a straight bole and round crown, whereas, when grown in an admixture with other species, forms a stratified canopy. In the meta-analysis of species richness productivity relationships by Zhang et al. (2012), heterogeneity of shade tolerance was the second most important factor explaining increased productivity in mixtures. In addition to an aboveground partitioning of canopy space, an increase in N availability via the N-fixing A. glutinosa could also be a factor in the higher productivity of the polyculture. In A. glutinosa under ambient CO2, the amount of N content in the leaves did not differ between monoculture or polyculture (Millett et al., 2012); however, in polyculture, leaves of F. sylvatica and B. pendula were less enriched in 15N compared with the leaves of these species growing in monoculture. This difference suggests an incorporation of N fixed by the symbionts of A. glutinosa. In other investigations, the contribution of transferred N to total N was 5–15% (Arnebrant et al., 1993) and 1–3% (Ekblad & HussDanell, 1995), on average, between A. glutinosa and P. contorta and A. incana and P. sylvestris, respectively. Furthermore, leaves of both F. sylvatica and B. pendula with greater numbers of A. glutinosa as direct neighbours were significantly depleted in 15N relative to the leaves of those with fewer A. glutinosa as direct neighbours (Millett et al., 2012), suggesting a competition for N as a possible mechanism for the exclusion of some B. pendula.

In response to elevated CO2, aboveground woody biomass for all three species combined was increased by 22% in monocultures. A response of this magnitude is consistent with a previously reported woody plant response of 28% calculated from meta-analyses of elevated CO2 experiments (Curtis & Wang, 1998; Ainsworth & Long, 2005) and 23% from four forest FACE experiments after 6 yr of enrichment (Norby et al., 2005). Utilizing observations spanning somewhat longer exposure to elevated CO2 (up to 11 yr), Norby et al. (2010) have shown that NPP responsiveness decreases in time. The limitation of NPP stimulation may largely be attributed to progressive N limitation (PNL); however, the observed reduction in NPP stimulation was almost entirely accounted for by changes in fine root production. Given the life history character traits of the species chosen in our experimental plantation, it is possible that the increased accrual of woody biomass observed in polyculture may not decrease as the forest stand develops. The presence of A. glutinosa in the mixture should compensate for increased N uptake, and thus negate or at least delay the onset of PNL. Several studies have shown that the rate of N fixation in the nodules of trees supporting this type of symbiosis increases under elevated CO2, presumably as a result of increased C availability (Hungate et al., 1999; Schortemeyer et al., 2002). Betula pendula and F. sylvatica growing in our plantation have been shown to utilize N fixed by A. glutinosa, suggesting that the presence of an N-fixing species might alleviate N limitation for all species grown in a polyculture.

There were considerable temporal differences in the response to elevated CO2 at our site. In the first growing season before canopy closure, all species responded to elevated CO2 enrichment by increasing total biomass by 27–29%. The stimulation of B. pendula began to decline during the second growing season, whereas the response of F. sylvatica declined during the last two growing seasons – an effect often attributed to acclimation to elevated CO2 (Ainsworth & Long, 2005) or nutrient limitation (Oren et al., 2001). In the present study, leaf N was unaffected by elevated CO2 during all stages of development, and thus it is unlikely that the decreasing overall elevated CO2 effect is caused by N limitation. As a result of the history of land use at the site, we did not expect a lack of N to limit plant growth within the first 4 yr. Indeed, foliar N increased, whereas leaf NUE decreased, with time in all treatments, indicating sufficient N uptake. In all species pooled together, leaf NUE was increased by elevated CO2 and also by growing trees in a mixture. However, we did not observe any differences in leaf NUE in individual species, suggesting that a different mechanism may explain the observed species-specific responses.

As we observed an expanding system with at least two canopy levels, the developmental phase of the stand and the strength of competition in our experiment must also be considered. Each species used in this study differs in their shade tolerance. Ellenberg et al. (1991) characterized F. sylvatica, A. glutinosa and B. pendula as shade tolerant (‘3’, out of ‘9’), intermediate (‘5’) and light-demanding (‘7’), respectively. Low leaf mass per leaf unit area and high rate of C assimilation per unit leaf area of light-demanding species allow rapid occupancy of available space and some canopy light penetration (Niinemets, 2006). Considering only monocultures in 2005, the saplings of each species were initially not influenced by intraspecific competition for light and space, allowing a greater response to elevated CO2. The subsequent decline in response of F. sylvatica to elevated CO2 may be explained by strong intraspecific competition through leaf morphology and crown architecture that minimizes canopy light penetration. By contrast, A. glutinosa sustained the stimulation by elevated CO2, ranging between 25% and 32% throughout the 4-yr experiment. Claessens et al. (2010) described A. glutinosa as fast growing when juvenile, but as a poor competitor that does not produce shade leaves. Respirational losses of crown-shaded leaves may result in a leaf C balance that approaches zero, which can lead to rapid leaf death (Reich et al., 2009). In our ecosystem, fast juvenile growth, coupled with rapid self-pruning, enabled A. glutinosa grown in monoculture to fully utilize elevated levels of atmospheric CO2 to accumulate aboveground woody biomass; however, the aboveground growth response to elevated CO2 was dramatically reduced when species were grown in polyculture. Initial increases in biomass of F. sylvatica were marginal, eventually becoming suppressed in the last growing season. The lack of stimulation of F. sylvatica is most probably a result of faster canopy occupation by A. glutinosa and B. pendula under elevated CO2. Changes in leaf area index may influence canopy light penetration and interspecific competition under elevated CO2. In our study, in monocultures, the leaf area index was unaffected by elevated CO2, but there was a consistently higher trend in B. pendula for the first 3 yr. During the summer of 2006, a severe drought resulted in partial canopy defoliation, which may explain the dramatic leaf area index increase in 2007. Both species possess indeterminate growth characteristics that enabled an additional leaf flush when environmental conditions improved later in the 2006 season. We propose two mechanisms to explain this phenomenon: differences in rooting depth between the two species; and the ability to recover from defoliation related to N storage. Alnus glutinosa has been characterized as possessing extensive root systems, with particularly deep tap roots that enable it to access water below the normal water table (Schmidt-Vogt, 1971; Claessens et al., 2010). This confers a considerable advantage in leaf production during, and following, drought conditions. The second explanation centres on the storage of N in tree perennial organs which can be re-mobilized and support leaf re-growth after defoliation. In combination with a flush of C and organic N compounds released for root uptake as the abscised litter decomposed mid-growing season, this mechanism may have facilitated the development of leaf primordia and a greater leaf area index during the following season (Tromp, 1983). Oksanen et al. (2001) found that elevated CO2 consistently increased leaf area index throughout the growing season in aspen, birch and maple stands, which was attributed to larger leaves. By contrast, Gielen et al. (2001) found that the leaf area index of P. nigra increased by 225% during the first growing season. However, a post-canopy closure analysis using a fish-eye canopy analyser revealed no increase in leaf area index, which is in agreement with data obtained at the Oak Ridge deciduous closed-canopy elevated CO2 experiment (Norby et al., 2003).

Our results clearly show that the aboveground response to elevated CO2 is species dependent, but is also affected by intra- and interspecific competition. Indeed, old growth F. sylvatica have been reported to show only a limited response to CO2 enrichment (Körner et al., 2005). In our study, a small, but statistically nonsignificant, positive effect of elevated CO2 on F. sylvatica in polyculture was shown in 2006, a year in which a severe summer drought in June and July resulted in strong leaf loss in A. glutinosa and B. pendula. During this period, only 44 mm of precipitation fell, compared with 101, 216 and 85 mm in the same period of 2005, 2007 and 2008, respectively. In July 2006, the maximum temperature was 34.5°C, 10°C warmer than in other years. The increase in light penetration to the understorey formed by F. sylvatica, in combination with improved water use efficiency, may have stimulated a response to elevated CO2, at least until A. glutinosa and B. pendula re-grew some of their foliage in late August. The literature suggests that much of the response of trees to elevated CO2 is linked to greater water availability, and that trees may be more drought tolerant under elevated CO2 (Eamus, 1991; Holtum & Winter, 2010; Leuzinger et al., 2011). If elevated CO2 had conferred a greater tolerance to drought in our experiment, we would have expected the highest response to elevated CO2 in 2006; this was clearly not the case for A. glutinosa and B. pendula; however, the severity of the drought, in combination with higher temperatures and photosynthetic oxidative stress, should also be considered.

To date, the majority of tree elevated CO2 experiments have used monospecific tree stands and have reported a mean stimulation of NPP for the duration of the observation (Norby et al., 2010). We have shown that, in a short-term empirical study of juvenile deciduous temperate trees grown in polyculture, the aboveground woody biomass response to elevated CO2 is strongly decreased. This result may have implications for the estimation of the global forest response to elevated CO2, as in natural mixed species forest the response to CO2 may be lower than previous estimates. However, caution must be exercised when extrapolating data from small-scale temperate plantations, particularly when there is the potential for experimental artefacts, arising from CO2 enrichment systems and edge effects influencing the response of saplings planted in complex arrangements at high planting densities. Although providing useful data, experimental plantations do not directly mimic the natural species-diverse, multi-aged and complex structures of the majority of the world's forests that grow in differing biomes, constrained by other physical and environmental drivers. Leuzinger et al. (2011) suggested that an increase in the number of driver variables, such as elevated CO2, drought and N addition, will dampen the ecosystem response to single factors through contrasting driver interactions. Similarly, Langley & Megonigal (2010) showed that, in a grassland system, the addition of N under high CO2 promoted a shift in community composition to C4 species that were less responsive to CO2, thus decreasing overall community response. Further, Langley & Megonigal (2010) suggested that, if the addition of N favours species that respond strongly to CO2, the community response to CO2 should increase. In our experimental mixture, complementary resource acquisition led to greater community productivity, which dampened the aboveground woody biomass response to elevated CO2, even though the most responsive species in the monoculture (A. glutinosa and B. pendula) were promoted within the mixed community. This is most probably a result of changes in source–sink relationships and C allocation to belowground organs. Indeed, tree root systems under elevated atmospheric CO2 have been shown to expand more deeply into the soil (Lukac et al., 2003; Iversen, 2010; Smith et al., 2013). Clearly, we are only beginning to understand how changes in elevated CO2-influenced above- and belowground processes may alter plant community dynamics.

In conclusion, atmospheric CO2 enrichment did not alter species-specific allometric relationships. The estimation of aboveground biomass stocks and productivity revealed a differential response to elevated atmospheric CO2. Aboveground biomass responses to CO2 enrichment were species specific and strongly reduced when species were grown in polyculture. In monoculture, A. glutinosa produced the largest and most consistent response, maintaining growth response until the experiment's conclusion. By contrast, the growth response of B. pendula and F. sylvatica diminished with time. In polyculture, the growth of F. sylvatica was not enhanced by elevated CO2. Our results suggest that the determination of how the aboveground biomass response of deciduous species grown in polyculture differs from that of single-species plantations is imperative to improve our understanding of how future CO2 will impact upon natural forest community dynamics.


The development of the BangorFACE site infrastructure was funded by the Science Research Investment Fund. We thank the Aberystwyth and Bangor Universities Partnership Centre for Integrated Research in the Rural Environment and the Forestry Commission Wales for financially supporting the running costs of the experiment. Andrew Smith was supported by the Sir Williams Roberts PhD Scholarship match funded by the Drapers' Company. Many thanks are due to Michael Bambrick and Gordon Turner for technical assistance throughout the BangorFACE experiment. We thank David Ellsworth and two anonymous reviewers for helpful comments in the revision of the manuscript.