Carbon dioxide enrichment alters plant community structure and accelerates shrub growth in the shortgrass steppe

Abstract

A hypothesis has been advanced that the incursion of woody plants into world grasslands over the past two centuries has been driven in part by increasing carbon dioxide concentration, [CO(2)], in Earth's atmosphere. Unlike the warm season forage grasses they are displacing, woody plants have a photosynthetic metabolism and carbon allocation patterns that are responsive to CO(2), and many have tap roots that are more effective than grasses for reaching deep soil water stores that can be enhanced under elevated CO(2). However, this commonly cited hypothesis has little direct support from manipulative experimentation and competes with more traditional theories of shrub encroachment involving climate change, management, and fire. Here, we show that, although doubling [CO(2)] over the Colorado shortgrass steppe had little impact on plant species diversity, it resulted in an increasingly dissimilar plant community over the 5-year experiment compared with plots maintained at present-day [CO(2)]. Growth at the doubled [CO(2)] resulted in an approximately 40-fold increase in aboveground biomass and a 20-fold increase in plant cover of Artemisia frigida Willd, a common subshrub of some North American and Asian grasslands. This CO(2)-induced enhancement of plant growth, among the highest yet reported, provides evidence from a native grassland suggesting that rising atmospheric [CO(2)] may be contributing to the shrubland expansions of the past 200 years. Encroachment of shrubs into grasslands is an important problem facing rangeland managers and ranchers; this process replaces grasses, the preferred forage of domestic livestock, with species that are unsuitable for domestic livestock grazing.

Fig. 1.

Plant species richness (A and B); evenness (J′) (C and D); and diversity (H′) (E and F) determined from ambient (360 ppm CO₂) and elevated (720 ppm CO₂) OTC plots, plus unchambered plots (A, C, and E), and as affected by years (B, D, and F) on the shortgrass steppe of eastern Colorado. Unchambered plots were included in the experiment as a control to determine whether the chamber had any effect on plant community dynamics, which, for the most part, did not. ANOVA detected significant treatment effects (P < 0.01) for richness only and year (P < 0.01) effects for all indices. Data are averaged across years (A, C, and E) and across treatments (B, D, and F). Mean differences among treatments and years were determined using a least squares difference test (P < 0.05) and are indicated by different letters.

Fig. 2.

Whittaker's community association index for contrasts of ambient vs. elevated chambered plots for the pre-CO₂ fumigation year (1996) and the following 5 years of CO₂ enrichment (1997–2001). An index value of 0 indicates both treatments having all species in common and in the same proportions, and a value of 1 indicates no species in common. Regression analyses indicated significant relationships between year and the index calculated for all 34 species (P = 0.04); bars indicate SEM.

Fig. 3.

Within-year comparisons of aboveground biomass of functional groups C₃ grasses, C₄ grasses, herbaceous forbs, and subshrub (A. frigida) as affected by growth in ambient (360 ppm CO₂) and elevated (720 ppm CO₂) OTC plots and nonchambered control plots conducted across 6 years, a baseline year (1996) before treatments, and 5 years of treatments (1997–2001). Means comparison tests were determined among treatments within each year based on a least squares difference test (P < 0.05); differences are indicated by different letters, and standard error bars are given.

Fig. 4.

Evaluations for changes in vegetative cover over 5 years of CO₂ enrichment among four functional groups: C₃ grasses (A), C₄ grasses (B), herbaceous forbs (C), and one subshrub species [A. frigida (D)]. For each functional group, the annual mean percentage of vegetative cover was regressed on year to evaluate differences in temporal trends for plants grown under elevated (720 ppm CO₂) and ambient (360 ppm CO₂) [CO₂] and the unchambered control conditions in native shortgrass steppe from 1997 to 2001. Only significant (P < 0.10) regressions are plotted, which include ambient (P = 0.03) and elevated (P = 0.08) C₄ grasses, unchambered forbs (P = 0.04), and all subshrub treatments (P < 0.0001 for elevated and ambient plots; P = 0.05 for unchambered plots). For the subshrub, slope of the elevated CO₂ line was significantly greater than slopes of the ambient CO₂ (P = 0.006) and unchambered (P = 0.001) lines. Each data point represents annual means from measurements taken six or seven times within each growing season; bars indicate SE.