Elevated CO2 Reduces Losses of Plant Diversity Caused by Nitrogen Deposition

See allHide authors and affiliations

Science  04 Dec 2009:
Vol. 326, Issue 5958, pp. 1399-1402
DOI: 10.1126/science.1178820

This article has a correction. Please see:


The interactive effects of rising atmospheric carbon dioxide (CO2) concentrations and elevated nitrogen (N) deposition on plant diversity are not well understood. This is of concern because both factors are important components of global environmental change and because each might suppress diversity, with their combined effects possibly additive or synergistic. In a long-term open-air experiment, grassland assemblages planted with 16 species were grown under all combinations of ambient and elevated CO2 and ambient and elevated N. Over 10 years, elevated N reduced species richness by 16% at ambient CO2 but by just 8% at elevated CO2. This resulted from multiple effects of CO2 and N on plant traits and soil resources that altered competitive interactions among species. Elevated CO2 thus ameliorated the negative effects of N enrichment on species richness.

Two global change factors likely to have widespread influence on plant communities are nitrogen (N) deposition and rising atmospheric carbon dioxide (CO2) levels (17). Levels of N deposition and CO2 have risen in recent decades and are expected to increase further (8). Because increased CO2 and N supply often drive plant stoichiometry in opposite directions but productivity in the same direction, and as plant resources are primarily available aboveground versus below ground, there are many possible ways in which competitive or other biotic interactions that influence biodiversity might be affected (27, 911). Although increasing N supply frequently results in declining species diversity (1, 6, 7, 9), there has been less research about (4, 5, 10), and no consensus regarding, how rising CO2 levels will influence species diversity. Even less is known about the influence of rising CO2 on the effects of N deposition on diversity (4, 5).

Experimental and observational studies in terrestrial ecosystems have typically shown that increases in N availability increase productivity and decrease plant diversity, and this has been explained by a variety of mechanisms (1, 6, 7, 9, 1216). Investigations in different study systems have provided evidence that a decline in diversity under elevated N can result from resource preemption (i.e., from either belowground resource or light competition) and associated competitive exclusion, shifts in competitive intensity aboveground versus belowground, alterations of soil acidity, a switch from one limiting resource to another, and/or a shift in niche dimensionality (1, 6, 7, 9, 1216). Therefore, although the suppression of diversity by increasing N availability is almost ubiquitous, no single mechanism is universally responsible.

In contrast, evidence and theory about CO2 effects on species richness are less well developed. Much like enriched N, elevated CO2 could result in decreasing species richness as it also commonly increases productivity (2, 1719), potentially leading to competitive exclusion following resource preemption. Alternatively, as rising CO2 levels change plant stoichiometry (17, 20), potentially resulting in greater relative limitations by other dominant resources such as N (18, 19, 21), elevated CO2 could reduce competitive exclusion and lead to increased species richness. Evidence of CO2 effects on species richness is scarce (4, 5, 10, 22) and shows mixed results, with positive, neutral, and negative responses seen in the few published reports.

Equally important to impacts of multiple global change agents is whether their effects are interactive (24, 18, 19), as it will bode poorly for future biodiversity conservation if rising CO2 exacerbates the considerable negative impacts of N deposition on community-scale species richness (1, 6, 7, 13, 14). However, a plethora of possible mechanisms suggests that synergistic, additive, or antagonistic interactive outcomes of joint CO2 and N effects are plausible (23).

To address the issues raised above, species richness was measured in 48 experimental grassland plots (each 2 m by 2 m) planted in 1997 with 16 perennial species and treated since 1998 with all combinations of ambient and elevated atmospheric CO2 (ambient and +180 μmol mol−1 delivered by means of a free-air CO2 enrichment technique) and ambient and enriched N (ambient and +4 g N m−2 year−1 delivered as ammonium nitrate in three equal doses each year) (11, 19, 23). This experiment, called BioCON, is conducted in an ecosystem co-limited by CO2 and N (11, 19) and dominated by belowground interactions (24, 25). Although wet and dry N deposition to terrestrial ecosystems is primarily of atmospheric origin, the effects are largely mediated through belowground processes, because uptake by soil microbes and plant roots generally begins the incorporation of this N into the plant biogeochemical cycle. Species richness (the number of species observed in a plot), belowground and aboveground biomass, root C/N ratio, soil solution N concentration, percent soil water content, and percent light transmission were measured in each plot in all years from 1998 to 2007 (23) and used to evaluate treatment effects on species richness and the underlying mechanisms (Tables 1 and 2).

From 1998 to 2007, there were significant main effects of N treatment (P < 0.001) and year (P < 0.0001) on species richness, and a significant interaction between CO2 and N treatments (P = 0.02) (Tables 1 and 3). On average, enriched N supply reduced species richness by 16% under ambient CO2, but only by 8% under elevated CO2 (Table 3 and Fig. 1). The N effect was consistently smaller under elevated than under ambient CO2 from the second to tenth year of the experiment (Fig. 1). From the CO2 effect perspective, elevated CO2 had minimal impact (−2%) on observed species richness at ambient N, whereas at enriched N, elevated CO2 modestly increased species richness by 7% (Table 3). The CO2 × N interaction was more pronounced once the experimental plots were well established. For example, during the most recent 7 years, enriched N supply reduced species richness by 15% under ambient CO2, but only by 5% under elevated CO2.

Table 1

Repeated-measures analysis of variance of CO2 and N effects on species richness. Effects of year, CO2, and N, and all interactions, on species richness (the number of species observed in a plot during sampling) are shown.

View this table:
Table 2

Species richness of the plant community as a function of total root biomass, root C/N ratio, soil solution N concentration, percent soil water content, and percent light transmission. Data are mean values for species richness, biomass, C/N ratio, soil solution N concentration, percent soil water content, and percent light transmission for each of 48 plots measured over 10 years (26). Model was selected based on Akaike’s Information Criteria from a suite of models involving these variables and all possible interactions.

View this table:
Table 3

Percent response (average, 1998 to 2007) of community and ecosystem variables to elevated CO2 at both ambient and enriched N, as well as to enriched N under both ambient and elevated CO2 levels. Variables include species richness, root biomass, root C/N ratio, soil solution N, percent soil water content, percent light transmission (all at plot scale), and species richness and relative abundance of each functional group (defined as the fraction of total aboveground biomass). Also shown is whole-model R2 and significance levels for analysis of variance for each variable in relation to CO2, N, and their interaction.

View this table:
Fig. 1

(A) Species richness (±SE among plots) of experimental plots (n = 48), under four combinations of CO2 and N (n = 12 plots for each), averaged from 1998 to 2007. (B) The percent effect of enriched N treatment on species richness under ambient CO2 (filled circle) and elevated CO2 (open circle) conditions from 1998 to 2007. (C) Mean species richness (±SE) and soil solution N (mg/kg, 0 to 20 cm depth) (±SE) from 1998 to 2007 for ambient and elevated CO2-treated plots under ambient and enriched N conditions. Relevant statistics for all panels are shown in Tables 1 to 3 and the text.

What accounts for the more consistently negative effect of added N than of elevated CO2 on species richness, and what caused the observed CO2 × N interaction? To address these questions, it is useful to focus on elements of plot-scale structure or function that might be influenced by CO2 and N and contribute to effects on species richness, asking (i) which plot-scale attributes were related to species richness, (ii) how did CO2 and N treatments influence those attributes, and (iii) were such responses consistent with observed treatment effects on species richness? Relevant measures (6, 7, 12, 13, 15, 2426) include (i) total root biomass, an indication of productivity and potentially of capacity to preempt (competitively obtain) soil resources; (ii) soil solution N, an indication both of resource supply and of resource preemption; (iii) root C/N, which is an indication of species differences in root chemical stoichiometry and hence of both relative physiological limitation by N and of treatment-induced differences in soil N availability; (iv) percent soil water content, an indication both of resource supply and of resource preemption; and (v) percent light transmission, which can indicate variation in the level of asymmetric competition for this aboveground resource.

These five measures were not closely related among plots (23) and thus, in theory, each could serve to independently drive species richness. In bivariate relations, species richness was negatively related (P < 0.001) to increased soil solution N and root biomass, and positively related to root C/N ratios (P < 0.001) and percent soil water content (P < 0.05) (Fig. 2, A to D), but unrelated (P > 0.50) to percent light transmission. Moreover, all five of these attributes were significant predictors of species richness in multiple regression models (Table 2 and table S1), with residual plots from the full model (Fig. 2, E to H) similar to those for the bivariate relationships for the four belowground attributes (Fig. 2, A to D). This similarity indicates that the associations seen in bivariate relations are also significant (and in the same direction) once the effects on species richness of other important driving variables are accounted for. The question then is whether these attributes responded to CO2 and N treatments in ways that would “drive” species richness in the observed patterns.

Fig. 2

The relation of mean species richness per plot versus root biomass (0 to 20 cm, g m−2), root C/N ratio, soil solution N (mg/kg, 0 to 20 cm depth), and percent soil water content (0 to 20 cm depth) for all plots (n = 48), each averaged over 10 years [(A to D), all relations significant P < 0.05], as well as the partial residual relationship (E to H) for each of these from the full multiple regression model. The arrows show directional effects of enriched N (+N) or elevated CO2 (+CO2) on both species richness and the ecosystem attribute shown in each panel. Confidence intervals (95%) for the model fit are shown with the dotted lines. Multiple regression model details are provided in Tables 1 and 2.

Total root biomass, root C/N, soil solution N, and percent soil water content were influenced by CO2, N, and their combination in ways that were consistent with CO2 and N effects on species richness, and with soil and root factor relations with species richness (Tables 1 to 3 and Figs. 1 and 2), whereas percent light transmission was not (23). For example, enriched N increased soil solution N concentration and root biomass, and decreased root C/N and percent soil water content (Table 3), all of which are consistent with decreased species richness given relations shown in Figs. 1 and 2. Treatments that made plants N-rich or productive, and soils N-rich or dry, also reduced diversity. Moreover, joint effects of CO2 and N on biomass, C/N, and soil solution N mirrored their joint effects on species richness (i.e., enriched N effects were smaller at elevated than in ambient CO2 treatments, significantly so for soil solution N; Table 3). Hence, the main and interactive effects on species richness of CO2 and N in this experiment were apparently the result of impacts of CO2 and N on several belowground drivers of species richness, perhaps, most importantly, soil solution N (Figs. 1 and 2 and Table 1), which itself showed a significant CO2 × N interaction (Table 3).

The changes in species richness under CO2 and N can also be viewed through the lens of individual species and functional group responses (4, 5, 10), given that all plots were initially planted with four species from four functional groups (11). The relative abundance of C3 grasses increased markedly with N enrichment (by 69%) and was associated with a modest increase (+8%) in C3 grass species richness (Table 3). In contrast, with enriched N, two of the three other functional groups had large decreases in relative abundance, and all three had lower species richness (Table 3), with this suppression more modest at elevated than at ambient CO2, especially for the C4 grasses (P = 0.0006 for the CO2 × N interaction). Additionally, changes in species richness in response to CO2 and N and their interaction were little influenced by changes in frequency (fraction of plots in which they are present) of either the rarest or most frequent species; instead, a set of species of intermediate frequency and abundance played the major role in this regard (23).

These results suggest that changes in ecosystem attributes and in functional group relative abundances together explain the main and interactive effects on CO2 and N on species richness in this system. The decreasing species richness under enriched N likely resulted from competitive exclusion of other functional groups by increasingly abundant C3 grasses, and was associated with greater root biomass, lower root C/N, greater soil solution N, and lower percent soil water content under enriched N. At this site, increased C3 grass biomass under N enrichment can reduce soil water availability, leading to increased mortality of other species (26), and the BioCON C4 grasses show reduced relative abundance in simulated drought treatments (23). These enriched N effects were somewhat muted under elevated compared to ambient CO2, because enriched N-induced increases in root biomass, decreases in root C/N, and especially increases in soil solution N were smaller in the enriched CO2 treatment (Table 3). Given the relations of species richness to each of these potential drivers (Fig. 2), the smaller responses of these attributes to enriched N (in elevated than in ambient CO2) likely contributed to the smaller N-induced declines in species richness (Tables 1 to 3 and Figs. 1 and 2) in elevated than in ambient CO2.

In summary, elevated CO2 had modest effects on species richness (compared to N enrichment), in part because CO2 effects on key drivers of species richness were smaller and sometimes offsetting. For instance, CO2-induced increases (Table 3) in root C/N and percent soil water content, which were linked to increases in species richness, counteract biomass productivity effects (which decrease species richness) (Fig. 2). In contrast, enriched N has a consistent negative effect on species richness because its effects on productivity, soil solution N, soil moisture, and root C:N ratio all individually suppress species richness (Fig. 2). Moreover, because the effects of CO2 and N on the drivers of species richness differ depending on the particular combinations of levels of CO2 and N, the joint effects of enriched CO2 and enriched N were nonadditive, i.e., could not be predicted from knowledge of each alone. These results for a temperate perennial grassland contrast with results of a study of multiple global change effects in annual Mediterranean grassland, where such impacts were predictable from knowledge of each alone (4). Competition for soil N, for which soil solution N may be a surrogate, and for soil water may be particularly important in BioCON, as (i) there were CO2 × N interactions on soil solution N as well as on species richness, (ii) enriched N and CO2 treatments drove percent soil water content in opposite directions, and (iii) competition for N and water have been shown to influence the outcomes of competition of these species and functional groups at Cedar Creek (7, 2325). Indeed, when averaged over the entire experiment for the four contrasting CO2 and N levels, the correspondence between species richness and soil solution N was pronounced (Fig. 1)—with treatments diverging along a single species richness and soil solution N axis.

Results of this study have important implications for natural ecosystems under global change, because they demonstrated that within 2 years and persisting for 10, altered CO2 and N regimes had significant, interactive, impacts on species diversity. From a biodiversity conservation perspective, there was no evidence to support the worst-case scenario in which rising CO2 and N deposition each suppresses diversity and jointly do so additively or synergistically. Instead, their joint interaction ameliorated the diversity loss due to N enrichment that occurs under ambient CO2. However, in viewing the possible implications of these results at broad scales and in other ecosystems, it is uncertain whether rising CO2 and N deposition will generally cause changes in plant biomass, plant or soil stoichiometry, soil chemistry, or soil moisture or other drivers of biotic interactions in ways that lead to the same nonlinear interactive effect on species richness. Regardless, the sensitivity of species richness to factors that themselves were sensitive to CO2 and N suggests that predicting responses of species richness at local community scales may be challenging, as responses to multiple global change drivers are perhaps not generally predictable from the responses to each alone. Given that humankind is enriching the biosphere in both CO2 and N (8) and that species diversity is a key ecological attribute providing ecosystem services, such uncertainty further contributes to our concern about systemic impacts of global environmental change on Earth’s ecological sustainability.

Supporting Online Material

Materials and Methods

Table S1


References and Notes

  1. Materials, methods, and supplemental information are available as supporting material on Science Online.
  2. Funding was provided by the U.S. Department of Energy, National Institute for Climate Change Research; and the NSF, Long Term Ecological Research (LTER), Biocomplexity, and Long Term Research in Environmental Biology (LTREB) programs. Thanks to S. Hobbie, K. Worm, J. Trost, and D. Bahauddin in particular, and to many others for their contributions to the BioCON project; and to R. Dybzinski, K. Suding, I. Woodward, and B. Medlyn for critiquing an earlier version of the manuscript.
View Abstract

Navigate This Article