Atmospheric CO₂ concentrations have increased by more than a third and are expected to rise because of fossil fuel and changes in land usage (Houghton, 2003). Globally as well as locally, this has led to what is currently referred to CO₂ fertilization. The impact of CO₂ enrichment studies has become better understood with the advances in the available technology dealing with field as well as greenhouse experimentations. The variations within the published data are the consequences of differences in experimental protocols adopted, plant species used, ecosystems covered, biotic and abiotic stresses applied and CO₂ concentrations tested. Added to these in-consistencies, are the complexity of scales, both temporal and spatial, and interpretations of research outcomes. Consequently, data comparisons are becoming challenging and may lead to various and sometimes contradictory, interpretations and uses. The increase in plant production was reported to be negative (Newman et al., 1999), neutral (Ghannoum et al., 1997) or positive (Kinsman et al., 1997). Carbon dioxide enrichment trials also have a range of setups such as Free Air Carbon Dioxide protocol, also known as FACE sys- tem, (Idso and Idso, 1994), open top chambers (Kimball, 1992), controlled growth chambers (Cave et al., 1981) and even locally made chambers erected within greenhouses (Ksiksi and Youssef, 2010).

The variety of studies relating to CO₂ enrichment and plant responses has led to major challenges facing collective interpretation of results from CO₂ trials. Grass species such as Agrostis curtisii have been reported to have no response to CO₂ enrichment (Norton et al., 1999), while other grass species (Lolium perenne) were reported to have benefited by about 20% from CO₂ enrichment. A forb like Hemizonia congesta has been reported to benefit from CO₂ by about 69.6% (Edwards et al., 2001). Different photosynthetic pathways (eg. C3 and C4) have also been assessed with varied responses (Ebersberger et al., 2003; Mateos-Naranjo et al., 2008).

Ecosystem variations are an added complexity when trying to compare results from the variety of the published body of information on CO₂ en-richment. Studies have been reported about temperate (Kammann et al., 2005), humid (Ebersberger et al., 2003), semi-arid (Xu and Zhou, 2008) and Mediterranean (Roumet et al., 2000) ecosystems, with a wide range of re- sponses. The CO₂ concentrations that have been studied were between 440 and 900 ppm.

In this study, we conducted an analysis of 90 articles published between 1994 and 2010. Analyses included simple descriptive information on proportions of responses as well species functional groups in addition to correlation and regression analyses.  The focus was on overall morphological responses of each plant species. The overall aim was to make sense of what was found in order to contribute toward future research directions as well as modeling exercises relating to the field of CO₂ enrichment and plant responses.

Data sources

A total of 90 published papers was included in this database. The years of publications ranged between 1994 and 2010. Each of the articles included in the database has more than one  entry  because of different species, functional groups and/or photosynthetic pathways reported. Therefore, a total of 395 entries was analyzed in the present attempt to understand the practical implications of (CO₂ enrichment studies. Data analyzed in this report were extracted from the published data (tables and graphs) within each of the database articles. Each data entry contained the following variables: photosynthetic pathway (C3, C4 and unknown), plant functional groups (forbs, grasses, legumes, mixed and woody), biotic/abiotic stress (CO₂ (defoliation, nutrient, salinity, temperature and water) and the studied ecosystem (alpine, dry, humid, Mediterranean, spring, temperate and semi-arid). Studies dealing with Cadmium stress were not included in the summary. The magnitude of response was another important variable which was entered as an average plant species response for a specific database entry. Consequently, an average response was calculated for each functional group and photosynthetic pathway based on specific species magnitude of responses. Specific ecosystem differences were not included as the sample data covering many of the ecosystems are not enough to run a robust analysis.

Standard deviation of each group is also included in the graphs (vertical bars). Some low SD values are an indication of the limited number of articles within the studied category, in addition to low variations among the different entries. Magnitude of response above 200% was not included in the analysis. Simply because of the high variability and most of these data points were identified as outliers using Statistical Package for the Social Sciences or SPSS (Norusis, 2010). It is also worth noting that the listed article published by (Wand et al., 1999) is not included in the analysis. It was an article dealing with meta-analysis of published data.

Descriptive and analytical approach

SPSS was also used to perform a correlation analysis using Pearson correlation coefficient between the magnitude of responses and the level of CO₂ under which the study was conducted.  Linear regression analyses were performed between CO₂ fertilizing levels and average magnitude of responses within each photosynthetic pathway (that is, C3, C4 and unknown). It was decided to report the exact P values to allow the reader to make their own judgment on the relevance of the statistical significance of the regression tests. The hypothesis was meant to address the question if the increase in the CO₂ content was correlated with the magnitude of plant growth, within each photosynthetic pathway (C3, C4 and unknown). The analyses did not show any significant correlation for all 3 pathways at P≥0.05.

The database included 90 references with a total of 395 entries. As each article has multiple entries for different species, functional groups and/or photosynthetic path-ways. Averages are reported in this section, including standard deviations. Please refer to the experimental section below for more details on the articles and resulting database included in this analysis.

Out of the total entries, the results revealed 238 and 40 entries for C3 and C4 pathways; respectively (Figure 1). All unidentified photosynthetic pathways were grouped into a class of unknowns with a total of 117 entries grouped under this rubrique. Figure 2 summarizes the magnitude of response of the different photo- synthetic pathways. Of the total entries into the database. C4 plants had an average magnitude of response of 27.1±0.4%.  C3 plant species had an average growth response of 22.7 ±0.4% while 8.2±0.8% was the response attributed to unknown pathways of the species included in the database. For comparison purposes, the average response across pathways was 19.3 ±0.5%.

Figure 3 presents the average response for each of the studied plant functional groups. Studies with woody species reported the highest magnitude of plant responses (33.5±0.4%). Legumes were reported to have had   an    average    magnitude of   plant   response of 31.4±0.5%. Forb and grass species were reported to have a growth response of 23.8±0.7% and 16.9±0.5%; respectively. Mixed species functional group had the lowest reported response of -1.25±0.3%.

Figure 4 shows the summary of the plant species responses to CO₂ enrichment for various stress factors. Among the experimental factors or stresses studied in the reviewed articles, CO₂ alone was reported to have the highest magnitude of response of about 20.4±0.6%. Temperature stress, coupled with CO₂ enrichment, had the lowest response of about   5.2   ±   0.1%.    Salinity, nutrient, defoliation and water stresses had average responses of 15.7±0.2%, 12.3±0.3%, 10.80 ±0.2% and 7.6±0.2%; respectively.

Figure 5 presents the average plant responses within the various terrestrial ecosystems for studies dealing with CO₂ fertilization.  The highest average response was reported for alpine ecosystems (187 ± 0.2%) while the second highest response was for dry ecosystems (126.3 ± 0.2%). The average magnitude of plant responses within humid and mediterranean ecosystems was 93±0.04% and 71.5±0.1%; respectively. Negative reponses, however, were reported for semi-arid and temperate ecosystems as -32.2 ±0.02% and -141.2 ±1.6%; respectively. Article number 38 (THÜRIG et al., 2003) was not included in the report as it was discovered  that it dealt with a spring ecosystem, not considered a terrestrial ecosystem, in Switzerland.

Multiple factors interact under CO₂ enrichment, or fertilization, and simple predictions are to be adopted to foresee future projections at regional and global scales. Sophisticated modeling and experimentation techniques have come a long way in studying CO₂ fertilization (Norby and Luo, 2004). Moreover, interactions between CO₂ and biotic/abiotic stresses are to be simplified in order to predict possible outcomes at the species, population and ecosystem levels. Especially those extreme variations in the published data are a concern during application and modeling. Plant species groups responded differently to CO₂ enrichment (Reich et al., 2001). C4 plant species have been reported to benefit from CO₂ fertilization (Ghannoum et al., 1997). While other reports stated reduction in biomass production for C4 plants (Reich et al., 2001). Classifications based on functional groups may be useful but not enough to assess plant and ecosystem responses to CO₂ enrichment (Reich et al., 2001). The present 90-article assessment revealed a high bias for C3 plant species, against those with C4 photosynthetic pathways. This high-lights the need for more studies on C4 plants which constitute an important majority of primary production within terrestrial ecosystems. On average, C4 plants were reported to have growth responses of about 27.1 ±0.4%. At the functional group level, woody species had the highest magnitude of plant responses (33.5 ±0.4%), while salinity as a stress tested under CO₂ enrichment conditions – led to an average response of 15.7±0.2%. The regression analyses revealed a negative predictive power (P=0.111) between CO₂ levels and magnitude of C3 plant responses. The regression results were not significant for C4 plants and for all other plant species grouped as unknown. The above attempt places much of what was studied in relation to plant responses to CO₂ enrichment (1994 to 2010) into an understandable level which can contribute in directing future research endeavours in the field of CO₂ enrichment and plant responses. It is also believed that much of what has been reported here could be incorporated into simplistic predictive modeling of CO₂ enrichment impact on terrestrial ecosystems, functional plant groups and floral species. This does not lessen the importance of deeper and more sophisticated analyses, such as those using meta analyses, to summarize the body of published data and make it useful and practical. But whichever analysis we adopt, periodic assessments, every 12 to 15 years, are to be conducted in order to keep up with the body of research into CO2 fertilization and terrestrial plant responses.

In short, there are unbalances in the published data on CO₂ responses of C3 vs C4 species. Stresses such as salinity, nutrient, defoliation and water have not been appropriately studied too. 

The authors have not declared any conflict of interests.

This research was supported by grants provided by the UAE University and the National Water Center (Projects No. 31S034 and 31R044).. The author would like to express his sincere appreciation to all colleagues at the Biology Department for being encouraging and supportive. Special thanks are extended to Ms Chedia Guinaoui, of IRA Tunisia, for compiling the database and the associated articles. Special thanks are also extended to Dr Katera Moore for her review of the manuscript. Her feedback made the readability much more acceptable.