Introduction

The increasing global atmospheric carbon dioxide concentration is expected to increase plant productivity. However, the strength of the response depends on the interaction with other limiting factors, of which nitrogen has been identified as one of the most important. This makes predictions of future plant responses difficult, as CO2 and nitrogen will interact at several temporal and spatial scales (Reich et al. 2006). Therefore, it is of interest to find ways of expressing the interaction between CO2 and nitrogen that simplify the description of this interaction. Most studies show that increasing CO2 concentration increases nitrogen use efficiency (e.g., Drake et al. 1997; Curtis and Wang 1998; Taub and Wang 2008). However, the responses are not universal and seem to depend on species and growth conditions and involve both morphological and physiological changes (Field et al. 1992; Lou et al. 1994). Nitrogen use efficiency involves also the length of the retention time of nitrogen in the plant, making it difficult to apply to dynamic conditions. On the other hand, nitrogen productivity, the growth rate of plant biomass per unit nitrogen in the plant, is directly designed to handle plant growth under dynamic nitrogen conditions. For example, Lou et al. (1994) found in their literature review that the nitrogen productivity in elevated CO2 could be from 5% less to 80% higher than at ambient CO2 in different experiments with a decrease in only 1 out of 12 observations. Nitrogen productivity here integrates carbon and nitrogen uptake and allocation to plant organs, and thus the impact of organs’ nitrogen concentrations. Using nitrogen productivity, it is therefore possible to directly express the growth response to nitrogen uptake without having to follow the more complicated route through photosynthesis and respiration (Ågren 1996). This circumvents the problem of differences between photosynthetic and growth responses to CO2 (Ainsworth and Long 2005). Temperature is another important control of plant growth and nitrogen productivity has previously been shown to be insensitive to temperature (Ladanai and Ågren 2004). For these reasons, it is of interest to use nitrogen productivity as an integrative concept of plant growth and to have quantitative estimates of how nitrogen productivity changes in response to increasing CO2. An important aspect of the plant response to elevated CO2 is a potential shift in biomass allocation, which would affect nitrogen productivity and ecosystem carbon cycling, if there were a shift between tissues of different nitrogen concentration and degradability. Increased CO2 could also lead to higher allocation to fine roots to increase nutrient uptake, to balance the increased carbon availability (Finzi et al. 2007) although Hilbert et al. (1991) suggest that this response should be small. Ågren and Franklin (2003) demonstrated also that changes in nitrogen productivity would only have minor effects on root: shoot allocation. In this study, we examined how nitrogen productivity responds to an increase in CO2 concentration for nine deciduous and three conifer tree species. In addition, we analyzed biomass allocation and leaf nitrogen concentration to disentangle their effect on nitrogen productivity.

Materials and methods

Experimental procedure

The experiment was conducted in the phytotron at the Justus Liebig University of Gießen in Rauischholzhausen (a detailed description is given in Kattge 2002). In brief, one individual of two-year old saplings of 12 tree species, selected to represent commercially and environmentally important species, were grown in vermiculite-filled pots in the phytotron in atmospheres of approximately 360 (ambient), 1000 (elevated), ppm CO2 during 91–117 days (Table 1). The photosynthetically active radiation was 500 µmol quanta m−2s−1, with a 18/6 h day/night period at 24/15 °C and relative humidity of 70/85% during two experimental periods. Four different nutrient treatments (N0, N1, N5, and N25) were combined with each CO2 level, where N0 meant that no nitrogen was added, and N1, N5, and N25 contained 1, 5, and 25 mmol L−1 nitrogen in a nutrient solution. Other nutrients were added in excess relative to nitrogen, following recommendations by Ingestad and Lund (1986). Plants were watered daily in surplus with the respective nutrient solution. Three plants per species were harvested at the start and one for each combination of CO2 level and nutrient treatment at the end of the experiment. All plants developed new leaves during the experiment. Plants were separated into leaves, stem, roots, and fine roots (diameter <2 mm), and dry masses and nitrogen concentrations determined for each fraction. As the experimental conditions have been identical in the two experimental periods, all results are pooled from them.

Table 1 List of species, length of experimental period, and nitrogen productivities (average over N treatments) (g g−1d−1)
Full size table

Estimating nitrogen productivity

To estimate the nitrogen productivity requires the temporal course of nitrogen uptake. From measurements of nitrogen content and dry weight on two observation occasions, nitrogen productivity was estimated as follows for each N and CO2 level:

The amount of N in a plant was assumed to increase exponentially with time, t:

$$ N(t) = N_{0} e^{rt} $$
(1)

from which the parameter r was calculated. This is the simplest assumption and requires estimation of only one parameter. A more realistic assumption where the uptake rate declines with the size of the plant would have required an unavailable time series of plant nitrogen uptake. However, in this experimental setup it is a reasonable assumption that nitrogen uptake increases in proportion to plant size, which creates an exponential rate of uptake (Ågren 1985).

The growth of plant dry weight, W, was calculated from nitrogen productivity, P N (Ågren 1985):

$$ \frac{{{\text{d}}W}}{{{\text{d}}t}} = P_{N} N(t) = P_{N} N_{0} e^{rt} $$
(2)

giving

$$ W(t) - W_{0} = \frac{{P_{N} }}{r}N_{0} (e^{rt} - 1) $$
(3)

or

$$ P_{N} = r\frac{{W(t) - W_{0} }}{{N(t) - N_{0} }}. $$
(4)

Results

Nitrogen productivity

Nitrogen productivity ranged between 0.20 and 3.79 g dw g−1 N day−1 in ambient CO2 and between 0.22 and 4.35 (Pseudotsuga menziesii at N5) g dw g−1 N day−1 in the elevated CO2 treatment (Table 1; Fig. 1, Supplementary information). Averaged over all species and all N levels, nitrogen productivity was 28% higher at the elevated CO2 level. The increases in P N were significant (p < 0.1) for four species (Table 1) but with clear differences between species; P N increased with increased CO2 at all N levels for Picea abies and Eucalyptus pauciflora, but in no case for Pinus sylvestris. An ANOVA analysis considering plant species, N supply and CO2, showed significant effects on P N for species and N supply (p < 0.05), with a tendency for CO2 (p = 0.17, Table 2). N levels also had an influence on whether P N would increase with CO2; at N0 and N1 only five and six species increased P N , whereas at N5 and N25 eight species had increased P N . All species except two (holm oak in the first exposure period and beech) increased their nitrogen productivity and pedunculated oak nearly doubled its productivity. The data suggest that nitrogen productivity of saplings of deciduous trees responded more strongly to CO2 (average increase 34%) than conifers (average increase 8%).

Fig. 1
figure 1

Nitrogen productivities under ambient and elevated CO2. Circles deciduous species. Squares conifer species. Nitrogen treatments: black N0, red N1, green N5, yellow N25. The dotted line is the 1:1 line

Full size image
Table 2 Three-way ANOVA of treatment effects on nitrogen productivity
Full size table

Allocation and leaf nitrogen concentration

There were only small effects of the CO2 treatment on allocation (Fig. 2) and no evidence of changes during the experiment. For example, under ambient CO2 the allocation to leaves was 38% of total plant biomass, compared with 39% in the elevated treatment. Similarly, fine roots constituted 15 and 17% in the ambient and elevated CO2 environments, respectively. The CO2 treatments changed plant nitrogen concentrations (Fig. 3) although none of the changes were statistically significant. In the two highest N treatments (N5 and N25), contrary to expectations, the elevated CO2 level actually increased plant nitrogen concentrations, whereas in the two low N treatments (N0 and N1) it lowered the concentrations. However, at the leaf level, there was a clear decrease in nitrogen concentration, from 28 to 23 mg g−1 over all N treatments and species.

Fig. 2
figure 2

Relative allocation of plant biomass under ambient and elevated CO2 and different N supply rates averaged over species. t0 is the allocation at the start of the experiment

Full size image
Fig. 3
figure 3

Whole-plant nitrogen concentration (mean and SE) under ambient and elevated CO2 and different N doses (N0, N1, N5, N25) averaged over species. t0 is the concentration at the start of the experiment. The whiskers show the SE of the concentrations. None of the differences in concentrations between ambient and elevated CO2 are statistically significant

Full size image

Discussion

We have demonstrated here that tree saplings response to elevated CO2 can be described as an increase in nitrogen productivity although the evidences are not unequivocal as different statistics give different results, which is not surprising, given the factorial range of the experiment (2 CO2 levels × 4 nutrient treatments × 12 species) and the limited number of plant individuals. The choice to not use replicates within species was a result of logistic restrictions and a desire to include as many species as possible in the study. To fully capture the growth response, we would also need information on the mean residence time of nitrogen in the plants, but measuring that was outside the scope of this study. Nevertheless, the estimated nitrogen productivities are in line with values reported previously (Ågren and Bosatta 1998) and the responses to CO2 are in line with other studies (Table 1). The increase caused by increased CO2 was higher than the 12% calculated in the DUKE FACE experiment (Finzi et al. 2002). However, in the latter experiment there was a steady decline from 20 to 4% in the response over the four years of the experiment. Our values are also higher than the 14% increase in relative growth rate, which is proportional to nitrogen productivity, under doubled CO2 reported by Kirschbaum and Lambie (2015) for 78 experiments. The larger response in our study could be a result of a larger increase in CO2 (from 380 to 1000 ppm) than in the cited studies and should probably be seen as an estimate of the maximal response. A response, with both increases and decreases in nitrogen and phosphorus productivity in response to increased CO2, has been found by Huang et al. (2015). A critical aspect of our way of estimating nitrogen productivity is the assumption that the added N is taken up in an exponential fashion. However, to calculate the nitrogen productivity it is necessary to know the temporal pattern of nitrogen uptake and the assumption of an exponential uptake is the simplest, and deviations from this pattern are likely to be different between N levels, which should explain part of the differences between responses between N treatments. It has been shown that the temporal pattern of N uptake can greatly modify the growth pattern (Wikström 1994; Ågren and Bosatta 1998). However, as the level of N supply had limited effect on the estimated N productivity, this was probably not a major problem here. The constancy of allocation within the experiment (Fig. 2) is also an indication that growth should have followed an exponential pattern.

The observation that deciduous trees responded more strongly than conifers agrees with observations by Niinemets et al. (2011). The 20% decrease in leaf nitrogen concentration is similar to the value found by Ainsworth and Long (2005) in their meta-analysis of 12 FACE experiments.

Conclusions

Given the factorial range of the experiment and its limited number of plant individuals, our analysis confirms that nitrogen productivity is a robust concept to integrate the multiple components of growth responses to increasing atmospheric CO2 concentrations. However, more studies are required to firmly establish the magnitude of the response. At the same time, the concept of nitrogen productivity provides a convenient tool for use at larger scales, as it bypasses some of the complex interactions between nitrogen and CO2. The nitrogen productivity concept is also convenient to apply at the forest stand scale (Ågren 1983) and over longer time scales, bypassing some of the difficulties with spatial and temporal scaling (Way et al. 2015).

Author contribution statement

JK performed the experiment. GÅ made the analyses. GÅ and JK wrote the manuscript.