Phosphorus supply drives nonlinear responses of cottonwood (Populus deltoides) to increases in CO2 concentration from glacial to future concentrations

Authors

  • James D. Lewis,

    1. Centre for Plants and the Environment, University of Western Sydney, Richmond, NSW 2753, Australia
    2. Louis Calder Center, Biological Field Station, and Department of Biological Sciences, Fordham University, P.O. Box 887, Armonk, NY 10504, USA
    Search for more papers by this author
  • Joy K. Ward,

    1. Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, KS 66049, USA
    Search for more papers by this author
  • David T. Tissue

    1. Centre for Plants and the Environment, University of Western Sydney, Richmond, NSW 2753, Australia
    2. Department of Biological Sciences, Texas Tech University, Lubbock, TX 79409, USA
    Search for more papers by this author

Author for correspondence:
James D. Lewis
Tel: +1 914 273 3078
Email: jdlewis@fordham.edu

Summary

  • Despite the importance of nutrient availability in determining plant responses to climate change, few studies have addressed the interactive effects of phosphorus (P) supply and rising atmospheric CO2 concentration ([CO2]) from glacial to modern and future concentrations on tree seedling growth.
  • The objective of our study was to examine interactive effects across a range of P supply (six concentrations from 0.004 to 0.5 mM) and [CO2] (200 (glacial), 350 (modern) and 700 (future) ppm) on growth, dry mass allocation, and light-saturated photosynthesis (Asat) in Populus deltoides (cottonwood) seedlings grown in well-watered conditions.
  • Increasing [CO2] from glacial to modern concentrations increased growth by 25% across P treatments, reflecting reduced [CO2] limitations to photosynthesis and increased Asat. Conversely, the growth response to future [CO2] was very sensitive to P supply. Future [CO2] increased growth by 80% in the highest P supply but only by 7% in the lowest P supply, reflecting P limitations to Asat, leaf area and leaf area ratio (LAR), compared with modern [CO2].
  • Our results suggest that future [CO2] will minimally increase cottonwood growth in low-P soils, but in high-P soils may stimulate production to a greater extent than predicted based on responses to past increases in [CO2]. Our results indicate that the capacity for [CO2] stimulation of cottonwood growth does not decline as [CO2] rises from glacial to future concentrations.

Introduction

Atmospheric concentrations of CO2 ([CO2]) have increased by 40% over the past 150 yr since the beginning of the industrial revolution, yet few studies have examined plant responses to preindustrial [CO2] (Overdieck et al., 1988; Polley et al., 2008; Anderson et al., 2010; Ghannoum et al., 2010). The rapid change in atmospheric [CO2] from a range of c. 180–280 ppm over the last 650 000 yr to 389 ppm at present suggests that many plants may be adapted to lower [CO2] than they currently experience (Sage & Cowling, 1999; Körner, 2006). Elucidating the mechanisms that regulated the responses of long-lived plants to past increases in atmospheric [CO2] will improve our ability to predict future responses to rising [CO2]. Understanding tree responses may be particularly important because forests are a major component of the terrestrial biome (Waring & Schlesinger, 1985; Field et al., 1998), and their responses to rising atmospheric [CO2] play a prominent role in the global carbon cycle (Melillo et al., 1993; Schimel et al., 2001; Norby et al., 2005).

The role of nutrient availability is a key unresolved issue regarding plant responses to rising [CO2] (Körner, 2006; Liberloo et al., 2009; McCarthy et al., 2010). Plant responses to elevated [CO2] often vary with the availability of inorganic nutrients, but generally decline in parallel with declining nutrient availability (Curtis & Wang, 1998; Saxe et al., 1998; de Graaf et al., 2006). In most elevated [CO2] studies, research on the effects of nutrient availability focus on nitrogen (N) (Johnson et al., 1998; Peterson et al., 1999; Lewis et al., 2004; Reich et al., 2006). However, responses to N limitation may not be indicative of the effects of phosphorus (P) (Conroy et al., 1992; Thomas et al., 1994), which differs from N in its photosynthetic and metabolic functions (Cernusak et al., 2010). P limitation is common in many terrestrial ecosystems (Reich & Schoettle, 1988; Raaimakers et al., 1995; Lambers et al., 2006; Nord & Lynch, 2009), and the influence of P availability on plant growth is likely to increase in many regions as N deposition eases N limitation (Wassen et al., 2005).

Long-term nutrient limitation may affect plant responses to [CO2] through effects on physiology, morphology and ontogeny. Previous studies suggest that long-term P limitation regulates growth responses to elevated [CO2] primarily through effects on biomass allocation rather than through direct effects on photosynthetic capacity (Conroy et al., 1988; Field et al., 1992; Lewis & Strain, 1996). It is, however, unclear to what extent the effects of P on allocation reflect morphological plasticity or the degree to which these effects are independent of ontogenetic shifts. For example, P limitation may alter developmental rates (Nord & Lynch, 2008), and subsequent, associated changes in the proportions of carbohydrate source and sink tissue may alter responses to rising [CO2] (Reekie et al., 1998; Centritto et al., 1999; Bruhn et al., 2000; Lewis et al., 2002).

Rising [CO2] may moderate or compound the effects of P limitation on ontogeny. In contrast to P limitation, elevated [CO2] accelerates development in many plant species (Johnston & Reekie, 2008; but see Lewis et al., 2003; Springer & Ward, 2007). However, rising [CO2] may induce or compound the effects of P deficiency by stimulating CO2 uptake rates (Conroy et al., 1988; Morin et al., 1992; Duchein et al., 1993; Lewis et al., 1994a). Enhanced CO2 uptake capacity may initially increase growth rates, but sustaining these increased rates requires increased P uptake. By contrast, P deficiency may have smaller effects on growth or be induced at lower P availabilities under glacial [CO2] as a consequence of reductions in carbohydrate production rates and, hence, demand for P (Campbell & Sage, 2002, 2006).

In this study, we examined the effects of photosynthesis, morphology and development on the response of cottonwood (Populus deltoides) growth to glacial (200 ppm), modern (350 ppm) and future (700 ppm) atmospheric [CO2] across a range of P supplies. Cottonwood plays a key role in ecological and economic responses to climate change because it is widely distributed and its intrinsic high growth rates make it suitable for carbon sequestration initiatives and for use as a biofuel (Luxmoore et al., 2008). We hypothesized that increasing P supply would increase the relative growth response to increasing [CO2], and that P supply would affect growth responses to increasing [CO2] through effects on photosynthesis and dry mass allocation. If P supply mediates allocation responses to increasing [CO2] through effects on growth rates, then responses should be directly proportional to total dry mass. Alternatively, if P supply mediates responses to increasing [CO2] by altering patterns of allocation independent of effects on growth rates, then responses should be independent of total dry mass.

Materials and Methods

Growth conditions

Populus deltoides (Bartram ex Marsh.) plants were obtained as rooted nursery cuttings (Tom’s Tree Place, Lubbock, TX, USA) and kept in a cold room (4°C) until planting. Before planting, roots were rinsed with distilled water, secondary branches were removed, and initial fresh weight was determined. Individual cuttings were then planted in 3.5-l pots filled with a 2 : 1 (v/v) sand:gravel mixture and placed in controlled-environment growth chambers in the Duke University Phytotron (Durham, NC, USA). The [CO2] in each chamber was monitored and automatically controlled to generate glacial (200 ppm), modern (350 ppm) or future (700 ppm) [CO2]. Air temperature in the chambers was maintained at 29 : 19°C (light:dark) with a photosynthetic photon flux density (PPFD) of c. 1000 μmol m−2 s−1 generated by high-intensity discharge (HID) lamps during a 14-h photo- and thermo-period. Relative humidity was c. 70% during the light period and 95% during the dark period. All seedlings were watered twice a day with distilled water for the first 2 wk. Thereafter, seedlings were assigned a P treatment and watered to through-flow daily in the morning with half-strength Hoagland’s solution modified to generate six P treatments (0.004, 0.012, 0.02, 0.06, 0.1 and 0.5 mM P as KH2PO4). These P supply rates were chosen to represent a broad and ecologically relevant range in P availability (e.g. Conroy et al., 1988; Lewis et al., 1994a,b). Nutrient solutions were supplemented with the appropriate amount of K as KCl, such that all plants received the same amount of K at each watering. Seedlings were watered to through-flow with distilled water each afternoon to reduce salt accumulation. Eight plants were randomly assigned to each experimental combination (3 [CO2] × 6 P) for a total of 144 plants.

Net CO2 assimilation rate measurements

Light-saturated net CO2 assimilation rates at the growth [CO2] (Asat) were measured using an LI-6400 portable photosynthesis system (Li-Cor Inc., Lincoln, NE, USA) on recently expanded, fully mature leaves near the top of the crown. Measurements were taken on four to five plants per treatment combination (CO× P) 10 wk after [CO2] treatments were initiated. Saturating irradiance (2000 μmol m−2 s−1 PPFD) was provided by a light source containing red and blue light-emitting diodes (Li-Cor Model 6400-02B) mounted above the leaf cuvette. Leaves were maintained at their growth temperature (29°C), and the leaf-to-air vapor pressure deficit (VPDl) was 1.0–2.0 kPa, reflecting ambient water vapor conditions in the chambers. Because of equipment failure during measurements, Asat data from the 0.012 and 0.06 mM P treatments in modern [CO2] were excluded from the analyses.

Growth, dry mass allocation, and P measurements

Whole plants were harvested at the end of the experiment (i.e. 10 wk after [CO2] treatments were applied) and separated into their component parts (leaves, stems and roots). Leaf area was measured using a LI-3100 (Li-Cor). Tissues were dried to constant mass in an oven at 70°C and then weighed. Total dry mass production of each plant during the experimental period was determined by subtracting the initial dry mass (consisting of the initial stem and root, and estimated at the time of planting based on a fresh mass to dry mass regression generated using 20 plants) from the final dry mass. The leaf area ratio (LAR) was calculated using total leaf area and total plant dry mass. Four to six plants were harvested for each [CO2] × P treatment combination. Following growth measurements, tissues were ground for phosphorus analysis. Phosphorus concentrations were measured by A&L Laboratories (Lubbock, TX, USA).

Statistical analyses

This study was a two-factor experiment testing the interactive effects of growth [CO2] (200, 350 and 700 ppm) and P supply (0.004, 0.012, 0.02, 0.06, 0.1, and 0.5 mM) on growth, dry mass allocation, and photosynthesis. All statistical analyses were conducted using R version 2.8.1 (R Develop-ment Core Team, 2008). In all analyses, test results were considered significant if  0.05. Two-way analysis of covariance (ANCOVA) was used to test for the main and interactive effects of [CO2] and P supply on growth, photosynthetic and tissue P variables. [CO2] treatment was treated as a fixed, categorical factor, while P supply was treated as a fixed, quantitative factor. Because responses to increasing P supply were expected to be logarithmic, P supply concentrations were log-transformed before analyses. Initial total dry mass was used as a covariate to account for differences in plant size at the start of the study. Data were tested for homogeneity of slopes before ANCOVA analyses. Data were tested for normality by analyzing normal probability plots and for homogeneity of variances by plotting residuals vs predicted values. These tests indicated that the assumptions of normality and homogeneity of variances were met for all dependent variables and no transformations were necessary. For significant [CO2] × P treatment interactions, slopes were compared using t-tests. For significant [CO2] treatment effects, pairwise comparisons of means were conducted using Tukey’s HSD test.

To examine the effects of [CO2] on allometric relationships between plant parts and total plant dry mass, the following equation was used:

image(Eqn 1)

(x, total plant dry mass; y, tissue characteristics; m, the slope of the relationship (Gebauer et al., 1996).) Treatment effects on m were tested using ANCOVA. For significant [CO2] × total plant dry mass interactions, slopes were compared using t-tests. For significant [CO2] effects, pairwise comparisons of means were conducted using Tukey’s HSD test.

Results

Net CO2 assimilation rates

Light-saturated net CO2 assimilation rates at the growth [CO2] (Asat) increased with increasing P supply in cottonwood seedlings grown in modern (= 0.006) and future (< 0.001) [CO2], but did not vary with P supply in seedlings grown in glacial [CO2] (Fig. 1; = 0.092). Asat increased to a significantly greater extent with increasing P supply in future [CO2] compared with modern [CO2] (= 0.035). Although Asat significantly varied with P supply in the modern [CO2] treatment but not in the glacial [CO2] treatment, the relationship between Asat and P supply did not significantly vary between these two treatments (= 0.328). As a result, across P treatments, Asat was 45% higher on average in plants grown in modern [CO2] compared with plants grown in glacial [CO2] (= 0.031). Comparing plants grown in future [CO2] with plants grown in glacial [CO2], Asat was c. 15% higher on average in the lowest P treatments and c. 130% higher on average in the highest P treatment. The response of Asat to P supply was disproportionately larger in future [CO2] than in modern [CO2] when compared with glacial [CO2] (Fig. 1 inset).

Figure 1.

 Light-saturated net CO2 assimilation rates at the growth CO2 concentration ([CO2]) (Asat) and total dry mass production of Populus deltoides seedlings under glacial (open circles and solid line), modern (solid circles and broken line) and future [CO2] (open squares and dotted line) across phosphorus (P) supplies. Inset graphs show the effect of [CO2] on the slope of the response to P supply. Error bars, 1 SE. The x-axis is plotted on a log scale.

Growth

P supply affected relative responses to [CO2] for dry mass production, total leaf area and leaf area ratio (< 0.001). Although dry mass production increased with increasing P supply in all three [CO2] treatments (Fig. 1; < 0.012 in all cases), after taking into account differences in initial dry mass, the response to increasing P supply was larger in future [CO2] compared with glacial and modern [CO2] (< 0.001 in both cases). Future [CO2] increased total dry mass by 80% in the highest P supply, but only by 7% at the lowest P supply compared with plants in modern [CO2]. The effect of P supply on dry mass production was similar between glacial and modern [CO2] (= 0.865). Plants grown in modern [CO2] produced 25% more total dry mass, on average, across P treatments than plants grown in glacial [CO2] (= 0.034). Comparing plants grown in future and glacial [CO2], total dry mass production was c. 27% higher on average in the lowest P treatments and c. 105% higher on average in the highest P treatment (Fig. 2). The response of total dry mass production to P supply was disproportionately larger in future [CO2] than in modern [CO2] when compared with glacial [CO2] (Fig. 1 inset).

Figure 2.

 Effect of CO2 concentration ([CO2]) on total dry mass production of Populus deltoides seedlings relative to total dry mass production at glacial [CO2]. For clarity, only the lowest (open triangles) and highest (closed triangles) phosphorus (P) supplies are shown.

Total leaf area increased with increasing P supply across [CO2] treatments, after taking into account differences in initial dry mass, and the response to increasing [CO2] increased with increasing P supply (Fig. 3; = 0.014). In contrast to the relative responses of dry mass production and leaf area to the [CO2] and P treatments, LAR (the leaf area per unit plant biomass) increased to a greater extent with increasing P supply in the glacial and modern [CO2] treatments compared with the future [CO2] treatment (Fig. 3;  0.001 in both cases). The slope of the relationship between LAR and P supply did not significantly differ between the glacial and modern [CO2] treatments (= 0.700). As a result, LAR was significantly greater in the glacial [CO2] treatment than in the modern [CO2] treatment across P treatments (= 0.011).

Figure 3.

 (a) Total leaf area and (b) leaf area ratio (LAR; leaf area per unit plant dry mass) of Populus deltoides seedlings under glacial (open circles and solid line), modern (closed circles and broken line) and future (open squares and dotted line) CO2 concentration ([CO2]) across phosphorus (P) supplies. Error bars, 1 SE. The x-axis is plotted on a log scale.

Dry mass allocation

Patterns of relative allocation of dry mass were significantly affected by [CO2] treatment and P supply (Fig. 4), but the effect of [CO2] treatment on relative allocation to leaves, stems and roots did not significantly vary with P supply ( 0.236 in all cases). More dry mass was allocated to leaves and roots, and less to stems, in plants grown in future [CO2] compared with plants grown in glacial and modern [CO2] ( 0.033 in all cases). Plants grown in glacial [CO2] allocated more dry mass to stems compared with plants grown in modern [CO2] (= 0.042), but plants grown in glacial and modern [CO2] were similar in dry mass allocation to leaves and roots ( 0.593 in both cases). As is generally observed in P fertilization experiments, increasing P supply was associated with increased dry mass allocation to leaves (< 0.001) and reduced allocation to roots (< 0.001). Dry mass allocation to stems appeared to increase with increasing P supply, particularly in the glacial and modern [CO2] treatments (= 0.051).

Figure 4.

 Relative dry mass allocation to leaves (a), stems (b) and roots (c) of Populus deltoides seedlings under glacial (open circles and solid line), modern (solid circles and broken line), and future (open squares and dotted line) CO2 concentration ([CO2]) across phosphorus (P) supplies. Error bars, 1 SE. The x-axis is plotted on a log scale.

Allometric relationships

Effects of growth [CO2] on relative allocation of dry mass in part reflected effects on allometric relationships between the dry mass of individual tissues and total dry mass. Increasing [CO2] did not alter the slope of the relationships between root, stem or leaf dry mass and total dry mass (Fig. 5; = 0.032). However, even after taking into account treatment differences in total plant dry mass, increasing [CO2] was associated with increased root and leaf dry mass and decreased stem dry mass (P < 0.001 in all cases). Therefore, when comparing plants of similar sizes, increasing [CO2] was associated with increased leaf and root dry mass and decreased stem dry mass. Accordingly, differences among [CO2] treatments in dry mass allocation to leaves, stems and roots were not attributable simply to differences among treatments in total plant size.

Figure 5.

 Allometric relationships between total dry mass and leaf (a), stem (b), and root (c) dry mass for plants grown at glacial (open circles and solid line), modern (solid circles and broken line) and future (open squares and dotted line) CO2 concentration ([CO2]). The slopes (m) and coefficients of determination (r2) of the associated allometric equations are shown for each regression.

Tissue P concentrations

In general, increasing [CO2] reduced P concentrations in leaves, stems and roots (Table 1). There was a significant interaction between [CO2] and P supply such that the greatest reduction in P concentration in leaves, stems and roots caused by future [CO2] was at the highest P treatment (0.5 mM). This resulted in large reductions in leaf P (57%), stem P (33%) and root P (40%) when comparing future [CO2] plants with glacial [CO2] plants in the highest P treatment.

Table 1.   Tissue phosphorus (P) concentrations of Populus deltoides seedlings exposed to glacial (200 ppm), modern (350 ppm) or future (700 ppm) CO2 concentration ([CO2]) and one of six P supplies
VariableCO2 (ppm)P1 (0.004 mM)P2 (0.012 mM)P3 (0.02 mM)P4 (0.06 mM)P5 (0.1 mM)P6 (0.5 mM)
  1. Data are presented as mean (± 1 SE). The effect of increasing P supply on leaf, stem and root P concentrations significantly decreased with increasing [CO2] (< 0.001 in all cases).

Leaf P (mg g−1)2000.62 ± 0.030.64 ± 0.020.60 ± 0.000.80 ± 0.050.83 ± 0.062.54 ± 0.17
3500.54 ± 0.020.56 ± 0.040.62 ± 0.020.68 ± 0.090.76 ± 0.042.14 ± 0.10
7000.42 ± 0.030.43 ± 0.020.40 ± 0.000.40 ± 0.000.47 ± 0.021.10 ± 0.03
Stem P (mg g−1)2000.28 ± 0.020.32 ± 0.020.30 ± 0.000.35 ± 0.020.36 ± 0.021.14 ± 0.05
3500.28 ± 0.020.34 ± 0.020.36 ± 0.020.30 ± 0.030.38 ± 0.021.16 ± 0.07
7000.27 ± 0.020.32 ± 0.040.28 ± 0.020.30 ± 0.000.37 ± 0.020.76 ± 0.05
Root P (mg g−1)2000.43 ± 0.030.40 ± 0.030.48 ± 0.030.57 ± 0.090.65 ± 0.031.54 ± 0.09
3500.50 ± 0.050.46 ± 0.040.52 ± 0.060.52 ± 0.020.62 ± 0.041.26 ± 0.09
7000.47 ± 0.040.44 ± 0.050.50 ± 0.070.56 ± 0.020.52 ± 0.020.96 ± 0.08

Discussion

Increasing [CO2] was associated with a transition from CO2 as the primary limiting resource for cottonwood seedling growth at glacial [CO2] to P as the primary limiting resource at future [CO2]. Increasing [CO2] from glacial (200 ppm) to modern (350 ppm) [CO2] increased growth by c. 25% across the full range of P supplies. By contrast, the relative response to future [CO2] (700 ppm) varied with P supply, such that seedlings exhibited a minimal response in low P and the largest response in high P, which is consistent with patterns observed in other tree species, such as Pinus spp. (Conroy et al., 1988; Lewis & Strain, 1996). In the highest P treatment, future [CO2] increased growth by c. 80% compared with plants in the modern [CO2] treatment. Fast-growing plants such as cottonwood are expected to be among the most responsive to future [CO2] because growth rates in these plants are often constrained by the availability of CO2 for photosynthesis. However, low P already limits tree growth in many ecosystems (Reich & Schoettle, 1988; Raaimakers et al., 1995; Lambers et al., 2006), and our results suggest that low P will increasingly affect tree growth as [CO2] continues to rise.

Consistent with our hypotheses, increasing the growth [CO2] increased Asat and dry mass production in cottonwood seedlings not limited by P. Under well-watered, high-light and relatively nonlimiting nutrient conditions, future [CO2] increased Asat by 44% on average and nearly doubled dry mass production in cottonwood compared with plants in modern [CO2]. These relative responses are similar to or larger than the relative responses of other Populus species to elevated [CO2] in the POP/EUROFACE (Liberloo et al., 2007, 2009) and AspenFACE experiments (Isebrands et al., 2001), as well as the growth responses of other tree species grown at elevated [CO2] (reviewed by Ceulemans & Mousseau, 1994; Saxe et al., 1998; Ainsworth & Long, 2005; Norby et al., 2005). The relatively large stimulation we observed under nonlimiting light, water and nutrient conditions may reflect the relatively large difference in [CO2] between the modern and future [CO2] treatments (350 ppm vs 700 ppm) in our study, as well as gender (Ward et al., 2002), genotypic (Ward & Kelly, 2004), or species-specific differences compared with other studies. Cottonwood is among the fastest-growing tree species in North America, and is particularly adapted to riparian corridors and other habitats with high resource availabilities (Cooper & Van Haverbeke, 1990), potentially enhancing responses to rising [CO2].

The large enhancement of Asat and dry mass production at future [CO2] under relatively nonlimiting nutrient conditions contrasted markedly with responses to future [CO2] at the lowest P supply. Consistent with our hypotheses, decreasing P supply decreased relative responses to future [CO2] such that, in the lowest P treatment, plants in future [CO2] were only 7% larger on average, and had slightly lower Asat, than plants in modern [CO2]. Decreasing P supply has been shown to reduce or eliminate responses to elevated [CO2] of Asat (Conroy et al., 1986; Lewis et al., 1994a; Thomas et al., 1994) and dry mass production (Conroy et al., 1986, 1992; Lewis & Strain, 1996) in other tree species. Although our study is among the first to examine the effects of P supply on cottonwood responses to future [CO2], as well as glacial [CO2], decreasing N supply has been shown to reduce responses of cottonwood and other Populus species to elevated [CO2] (e.g. Kubiske et al., 1998; Zak et al., 2000; Sigurdsson et al., 2001; but see Kruse et al., 2003). Although N fertilization had a limited effect on the long-term responses of three Populus species to elevated [CO2] in POP/EUROFACE, the limited effect may reflect initial high nutrient availability in the soils at the start of that study (Liberloo et al., 2006, 2009; Lagomarsino et al., 2008).

In contrast to our hypotheses, however, P supply did not affect the relative response of Asat or dry mass production to an increase in [CO2] from glacial to modern concentrations. As a result, the transition from glacial to modern [CO2] was associated with a 25% increase in growth across P supplies. This response is equivalent to the 25% increase in growth on average observed across a range of species grown in preindustrial and modern [CO2] (Sage & Coleman, 2001), although the responses of individual genotypes and species may differ substantially from this average (Rogers et al., 1998; Cunniff et al., 2008; Ghannoum et al., 2010). The 25% increase in growth with a 150 ppm increase in [CO2] from glacial to modern concentrations is proportionately less than the c. 80% increase in growth we observed with a 350 ppm increase in [CO2] from modern to future concentrations under relatively nonlimiting resource conditions; however, it is proportionately more than the c. 7% increase we observed in the most-limiting P treatment.

Nonlinear responses to glacial and future [CO2]

The nonlinear response in cottonwood seedlings to increasing [CO2] at high P supplies differs from the findings of other studies, where the [CO2] stimulation of growth is often greater between glacial and modern [CO2] than between modern and future [CO2] (e.g. Baker et al., 1990; Dippery et al., 1995; Ward et al., 1999; Gill et al., 2002). Growth responses of C3 plants to rising [CO2] are expected to be largest in the transition from glacial to modern [CO2] because the relative effects of short-term increases in [CO2] on photosynthesis are greatest at low [CO2] and subsequently decline with rising [CO2] (Farquhar et al., 1980; Farquhar & von Caemmerer, 1982). Further, photosynthetic enhancement by elevated [CO2] may decline during long-term exposure, particularly in plants limited by nutrient availability (Tissue & Oechel, 1987; Gill et al., 2002; Reich et al., 2006; Ainsworth & Rogers, 2007). However, morphological changes, such as changes in leaf area, may occur during long-term exposure to elevated [CO2] that alter growth responses from those predicted based solely on photosynthetic responses to short-term increases in [CO2] (e.g. Zak et al., 2000; Ghannoum et al., 2010).

In the present study, the nonlinear growth responses to rising [CO2] reflected differences in morphological (leaf area) and physiological (Asat) responses to both the [CO2] and P treatments. Although growth responses to all three [CO2] treatments were affected by changes in Asat, only the response to future [CO2] was affected by changes in leaf area, which was similar between the glacial and modern [CO2] treatments. Further, increasing P supply only affected the growth response to future [CO2] by increasing the stimulatory effect of future [CO2] on Asat and mitigating the negative effect of future [CO2] on leaf area. These results suggest that nonlinear growth responses to glacial, modern and future [CO2] may occur across a range of P supplies because plant responses to future [CO2] may reflect a different combination of morphological and physiological traits than those that explain past responses to rising [CO2]. Further, contrasting nonlinear growth responses to rising [CO2] may occur as a result of differential effects of limiting and relatively nonlimiting P supplies on Asat and leaf area.

Plant dry mass allocation

Morphological changes with rising [CO2], such as increased dry mass allocation to leaves, may reflect differences in developmental rates (Gebauer et al., 1996). As a result, differences in plant size between glacial and future [CO2] may account for observed shifts in dry mass allocation rather than result from these shifts (Centritto et al., 1999; Bruhn et al., 2000; Lewis et al., 2002). Our allometric analyses, however, indicated that the effect of increasing [CO2] on relative allocation to leaves, stems and roots was not simply an effect of differences in developmental rates, as changes in allocation to these tissues were observed even after accounting for differences in total plant dry mass. Hence, our results suggest that increasing [CO2] had a direct effect on dry mass allocation to leaves, stems and roots, and that treatment differences in leaf area generated nonlinear growth responses.

Treatment differences in the relationship between total dry mass and dry mass allocation are consistent with general findings obtained across forest ecosystems that carbon allocation is not closely related to total biomass and that resource availability affects allocation (Litton et al., 2007). The increased allocation to roots with rising [CO2] and the shift in dry mass allocation from roots to leaves with decreasing P supply are consistent with responses predicted if dry mass is allocated to maximize resource uptake (Dewar et al., 2009; McCarthy et al., 2010). Accordingly, these shifts in dry mass allocation, coupled with reduced Asat and leaf area at the lowest P supplies in future [CO2], suggest that rising [CO2] may induce or exacerbate P limitation of cottonwood seedlings by easing carbon limitation. Further, the contrasting nonlinear responses to rising [CO2] in the lowest and highest P supplies suggest that P becomes increasingly limiting to growth as rising [CO2] increases the amount of carbon available for photosynthesis. The pattern of increasing P limitation with rising [CO2] is similar to the effects observed for N limitation with rising [CO2] in a variety of ecosystems (e.g. Gill et al., 2002; Reich et al., 2006; McCarthy et al., 2010). These results suggest that increasing demand for P to support increased growth rates may be another key constraint on plant responses to future [CO2].

Our P supplies were chosen to represent a broad and ecologically relevant range in P availability, and our results are consistent with patterns typically observed in studies on plant responses to P limitation (e.g. Lewis & Strain, 1996; de Groot et al., 2003; Nord & Lynch, 2008). For example, across [CO2] treatments, increasing P supply increased leaf P, LAR and leaf allocation while reducing root allocation. However, it is difficult to directly relate our P supply rates to P availabilities in natural soils because P in natural soils may be unavailable to plants as a result of sorption by soil particles and occlusion by iron and aluminum (Lambers et al., 2006). The magnitude of P limitation in our study was also enhanced because cottonwood seedlings were not mycorrhizal. Cottonwood plants readily form mycorrhizal associations (Khasa et al., 2002), and the lack of mycorrhizas may make cottonwood more susceptible to P limitation. Further, elevated [CO2] often enhances mycorrhizal formation (e.g. Lewis et al., 1994b), which may alter the effect of P supply on plant responses to increasing [CO2] (Lewis & Strain, 1996). More generally, rising [CO2] may enhance nutrient uptake in plants in the field with unrestricted rooting volumes by enhancing root proliferation and soil penetration, and these effects may vary among individual plants (Anderson et al., 2010). As a result, although our results are consistent with long-term studies of tree responses to elevated [CO2] under field conditions (Ainsworth & Rogers, 2007), and with studies that suggest that rising [CO2] may increase nutrient demand to support increased photosynthetic (Lewis et al., 2004) and growth rates (Reich et al., 2006), our results cannot be directly extrapolated to specific field conditions.

Conclusions

Our results suggest that low P may increasingly affect cottonwood growth as rising [CO2] reduces the CO2 limitation of photosynthesis. Also, as a result of the effects of elevated [CO2] and P on allometry, Asat and leaf area, growth responses to future [CO2] may reflect the effects of a different combination of morphological and physiological traits than those that drove past responses to rising [CO2]. As a result, when not limited by P availability, cottonwood may be more responsive to future [CO2] than predicted by responses to past increases in [CO2]. However, the increasing sensitivity of cottonwood growth to P supply with rising [CO2] suggests that P demand by cottonwood may increase to support positive growth responses to future [CO2], and that the role of cottonwood in ecosystem responses to climate change may increasingly be affected by P supply.

Acknowledgements

This study was supported by a National Science Foundation grant (IBN-0130885) to J.D.L. and D.T.T., a National Science Foundation CAREER award (IOS-0746822) to J.K.W., internal grants by Fordham University (J.D.L.) and Texas Tech University (D.T.T.), and a University of Western Sydney International Science Research Schemes Initiative (71846) to J.D.L. This is contribution number 247 from the Louis Calder Center and Biological Station, Fordham University.

Ancillary