The CO2 fertilising effect – does it occur in the real world?


The International Free Air CO2 Enrichment (FACE) Workshop: Short- and long-term effects of elevated atmospheric CO2 on managed ecosystems, Ascona, Switzerland, March 2004

It would seem simple. There are only two immediate primary responses of plants exposed to elevated levels of atmospheric CO2 concentration above the ambient (which currently averages approx. 375 ppm by volume, 33% up from the preindustrial 280 ppmv). First, in C3 species, competition between CO2 and O2 at the active site of the photosynthetic enzyme ‘rubisco’ is shifted in favour of reaction with CO2 thereby increasing gross photosynthetic CO2 fixation and decreasing CO2 loss via photorespiration. Second, in most species, both C3 and C4, stomatal aperture narrows thereby reducing stomatal conductance and, combined with the photosynthetic response, leads to increased water use efficiency in C-acquisition. That's it. No other primary responses have been identified, although I do wonder about whether there are subtle developmental effects associated with interactions between endogenous ethylene production and action and atmospheric CO2 concentration. But nothing in nature is simple and, with there being two known primary responses, the long-term repercussions for ecosystems may be twice as difficult to quantify as the linked issue of impact of the increasing greenhouse gas concentration on global climate for which there is only one primary response – namely, more of the infrared back radiation emitted from the Earth's surface is absorbed in the lower atmosphere.

‘Should this be seen as a mechanism of plants “resisting” a positive response to elevated CO2, i.e. showing resilience to change?’

For both the ‘greenhouse’ and ‘CO2 fertilising’ effects, debate has persisted over at least half a century as to whether these primary effects are leading, respectively, to global warming, and to increased vegetation productivity and C-stocks in the terrestrial biosphere. In both debates the power of constraints and feedbacks (both negative and positive) developing through time, and operating on various timescales and spatial scales, in the complex, adaptive climatic and ecological systems, have been invoked by some to argue for resilience to change. In both cases the debate continues despite continuing accumulation of observational evidence. For the CO2 fertilising effect, both new evidence and continuing debate was seen at the recent Free Air CO2 Enrichment (FACE) workshop in Switzerland ( How resilient are plant processes, crop yields, ecosystems and the terrestrial C-cycle to modification by elevated atmospheric CO2 concentration in the long term?

Doubts about long-term field-expression of the CO2 fertilising effect arise partly because the majority of such research has been in chambers, glasshouses, open-topped chambers, and controlled environments of various types, these often being short-term experiments. However, the longest enrichment experiment by far has been in open topped chambers on a salt marsh vegetation on Chesapeake Bay. Bert Drake (Smithsonian Environmental Research Center, Edgewater, MD, USA) reported at the meeting that after 17 yr the elevated CO2 concentration had increased the marsh shoot density by > 100% compared with ambient air control chambers. The development of the FACE technology (Box 1) in the mid-1980s by George Hendrey at the Brookhaven National Laboratory (Upton, NY, USA) has provided the opportunity to test responses in the field.

Table Box 1 .  FACE methodology
• FACE methodology (Lewin et al., 1992; Hendrey et al., 1999; Miglietta et al., 2001; Okada et al., 2001) involves a ring of separately controlled CO2 release points above the ground in circles from 1 m to 30 m in diameter.
• The point-releases can be computer-controlled to be always upwind of the central experimental zone (or ‘sweet-spot’), with the rate of release varied more or less with windspeed.
• There have been 13 large diameter-ring (> 8 m) FACE systems in the world, 10 still operating.
• There are approx. 20 ‘miniFACE’ ring systems 1–2 m in diameter, for which CO2 is usually released all around the ring continuously by day. The small ones do not have scope for a wide guard-zone around the sweet-zone and are unsuited to tall vegetation.

FACE versus Chamber – are the minor differences real?

Kimball et al. (2002) conducted a quantitative comparison of the conclusions about elevated CO2 effects on 11 crops (including grass, cereal, C4 sorghum, tuber and woody crops) from the four FACE experiments then available, compared with results from the large number of prior chamber experiments (including open-topped field chambers) over many years. It was comforting to those using both kinds of facility that FACE experiments had, with two exceptions and within the ranges of variability of reported results, confirmed under longer-term field conditions all the prior quantitative chamber-findings on crops grown and measured in elevated CO2 concentration compared with ambient CO2 concentration (persistently increased light saturated photosynthesis, decreased stomatal conductance, decreased water use, increased shoot biomass growth, increased root growth, decreased specific leaf area, decreased leaf nitrogen concentration, increased soluble carbohydrate content, little effect on phenology, and increased agricultural yield though for small grain cereal yield the increases were at the lower end of the range found in enclosed environments). The two exceptions were for reduction of stomatal conductance and enhancement of root growth relative to shoot growth, both of which were more strongly expressed in the FACE experiments than in the chamber experiments.

Lisa Ainsworth reported results of a statistical meta-analysis of results now available from experiments conducted over several years in 12 large-scale FACE facilities on four continents (Long et al., 2004). This again confirmed, with greater statistical rigour and for a much wider range of species including crops, pasture species and trees, most of the conclusions of the evaluation by Kimball et al. (2002) for a CO2 concentration of 550 ppmv. In addition, she noted that, in the open field, elevated CO2 increased apparent quantum yield of light-limited photosynthesis by 13% (a figure close to the theoretical short-term response expectation), that growth under water or N stress exacerbated the response of stomatal conductance to elevated CO2 concentration, and agricultural yield increased by 17% (average of C3 and C4) a figure similar to the average of 15% (scaled to 550 ppmv CO2) reported by Kimball (1986) for prior chamber studies. However, again the responses of rice and wheat yields were found to be lower than in chamber studies. Growth rate of above-ground biomass was also increased on average across all C3 and C4 species by 17%. For trees it increased by 28%, though this high result is influenced by the strong positive response of fast growing poplar saplings. Dicots were more responsive than grasses, and legumes more responsive than nonlegume forbs. Interestingly, the decrease in N-content per unit leaf area that has generally been observed in elevated CO2 chamber-studies was less pronounced in FACE experiments, −4% on average, a decline consistent wholly with the reduction in Rubisco content. To establish whether the apparent, relatively minor, differences in results between the FACE and enclosure experiments are real, coordinated FACE and enclosure experiments are needed as Alistair Rogers observed.

Positive and negative feedback

A fast-acting negative feedback, which has often been thought might lead to lower fractional response of growth than of photosynthesis rate in the short term (days to weeks), is down-regulation of photosynthesis under continuous exposure to elevated CO2 associated with reduced leaf N-content. Ainsworth's meta-analysis confirmed that this does usually occur in the field, with the maximum carboxylation capacity (Vc,max) decreasing on average by 13% under continuous exposure to elevated CO2. Down-regulation of Vc,max was more strongly expressed in grasses, shrubs and crops than in legumes and trees. Should this be seen as a mechanism of plants ‘resisting’ a positive response to elevated CO2, i.e. showing resilience to change? Probably not. Stephen Long (University of Illinois, Urbana, IL, USA) presented an elegant exposition of how photosynthetic down-regulation involves an optimisation of the deployment of N from photosynthetic machinery to growth organs such that a balance between C-source and C-sinks is maintained in the plant under elevated CO2 concentration – a response that generally increases the nitrogen use efficiency (Wolfe et al., 1998).

At an ecosystem scale over years to decades, another type of adaptation to continuous elevated CO2 concentration could be changes in plant community structure. One might reasonably hypothesise that species that are most responsive in growth to elevated CO2 concentration would become more dominant over time thereby leading to a positive feedback. Mike Jones (Trinity College, Dublin, Ireland) described the Megarich study in which monoliths of six grasslands across Europe were exposed to FACE over 3–6 yr. Generally, under competition, occurrence of dicots was enhanced and monocots relatively suppressed by continuous elevated CO2. And there was a significantly increased fraction of legumes in the swards (Teyssonneyre et al., 2002). While the Megarich study did not include determination of N-fixation, the increased preponderance of legumes in the swards is supportive of the notion that, in the long run, elevated CO2 concentration may cause N-fixation to entrain more atmospheric N2 into the ecosystem, leading ultimately to fuller expression of the increased growth and standing biomass potential that the elevated CO2 provides (Gifford, 1992). It might take several decades for such a positive feedback to build up in an ecosystem to the extent that it could be measured as increased N-stocks in the field. To date no FACE experiment has been for long enough. If such N-accumulation were eventually to occur much of it would be expected in the soil and, associated with it, more soil C.

Does elevated CO2 concentration lead to more C accumulation in the soil?

Chris van Kessel (University of California, Davis, CA, USA) addressed this question by studying soil C accumulation in the intensely N-fertilised Swiss grassland FACE system. He concluded that over 10 yr elevated CO2 concentration had no effect on soil C-stocks, no effect on soil microbial biomass including Rhizobium after an initial surge, and no effect on above ground litter decomposition. From this he posited the ‘resilience hypothesis’ that initial responses of soil C-cycle and N-cycle processes are short lived and that they relax back to their original stocks and rates. One mechanism for this may be the ‘priming’ of oxidation of some older more stable forms of soil organic matter by the input of more new easily oxidised organic matter as proposed by Marcel Hoosbeek (Wageningen University, Netherlands; Hoosbeek et al. (2004)). However, the artificial N-input to the Swiss FACE study was extraordinarily high (either 140 or 560 kg ha−1 yr−1 over the 10 yr). From an ancillary study at the same Swiss FACE site towards the end of the treatment decade, Paul Hill (University of Wales, Bangor, UK) observed that the greater potential for sequestration of C below ground was by the swards that had the lower N-supply. This partly agrees with a microcosm study in a controlled environment over 4 yr in which a native C3-grass was grown in a very low-N soil (total initial N of 0.02%) under elevated CO2 concentration with only 22–198 kg ha−1 yr−1 N supplied dilutely in the irrigation water. Over 4 yr the soil had gained 15–57% (respectively) more C with elevated CO2 concentration than without (Lutze & Gifford, 1998). Thus it is possible that under both extremely high and extremely low N-nutrition, elevated CO2 has no effect on soil C concentration while with intermediate N-nutrition elevated CO2 increases soil C stocks. If so, that would parallel the tendency for plant N concentration to be unaffected by elevated CO2 concentration at extremely low and high N-status, but diminished by elevated CO2 concentration in the intermediate range of N-nutrition (Gifford et al., 2000). Resolution of this issue is one for which long-term investigations are required. The workshop returned again and again to the need for long-term experiments in the field.

The profits and pitfalls of FACE

Every experimental system in vegetation studies has its advantages and drawbacks. The great advantage of the FACE approach is that it is technically possible – if funded appropriately – to apply long-term CO2 treatment to patches of existing ecosystem, even tall ones, over the long-term. Also leaf temperature can respond to the reduced transpiration naturally in the open air. George Hendrey urged researchers to be more aware of several inherent limitations with the FACE approach. He emphasised particularly the rapid (down to minutes or seconds) and sometimes large fluctuations in concentration of CO2 at each point in a FACE-ring owing to the inherent time delays of enrichment associated with sample-line length, with distance from release point to sweet-zone, with wind speed and direction changes, and with the eddy-structure of the atmosphere on the scale of FACE rings. CO2 concentration at any one place can undergo large fluctuations within seconds to minutes under FACE, a feature that is not mirrored, in terms of either amplitude or frequency spectrum, in the control treatment. Hendrey's analysis (Hendrey et al., 1997) of the impacts of such fluctuations combined direct measurements of the fluorescence responses of wheat leaves exposed to such CO2 fluctuations, which are embedded in the unweighted mean CO2 concentration, concluded that photosynthesis rate can be decreased by 17% or more for the mean concentration reported when that mean is of large CO2 fluctuations on the order of half the mean, and the deviations from the mean occur over a minute or longer. This derives from the fluctuating concentration driving the internal leaf concentration into the saturated part of the photosynthetic response curve. The larger the concentrations swing above the saturating concentration the worse the underestimate becomes of the response at the calculated mean CO2 enrichment.

A poster by Joe Holtum and Klaus Winter showed experimental data supporting Hendrey's conclusion. They showed (Holtum & Winter, 2003) that for two tree species the photosynthetic enhancement by CO2 concentration elevated to 600 ppmv was diminished by one third when that concentration was an average of subminute fluctuations between 433 and 766 ppmv. They also reported that the 26% growth response of rice seedlings to a stable 600 ppmv CO2 was eliminated when that average comprised 30 sec fluctuations having just a 150 ppmv amplitude. Thus extant FACE technology might be systematically understating the effect of globally elevated CO2 on ecosystem productivity. However, it is not only FACE facilities that can suffer such fluctuations. Open topped chambers and poorly designed or managed enrichment systems in CO2-enriched growth-chambers can also produce large ‘hunting’ effects that the investigators may be unaware of.

Thus CO2 concentration fluctuations in CO2 enriched but not ambient treatments may be a more general problem for elevated CO2 plant research than even Hendrey and Holtum realised. In chambers, however, it should not be such an insurmountable problem as in FACE. Perhaps a ‘second-best’ way forward is to routinely characterise the fluctuations and to model the effective concentration that the plants perceive. That would require, however, clear understanding of all the mechanisms involved. There might be other mechanisms. For example, regular fluctuation of CO2 concentration on a 10–30 min timescale might resonate with the inherent relaxation time of stomatal opening or closing and sometimes drive the pores fully open or fully closed artificially.

A second major potential problem for FACE technology is ethylene contamination of the CO2. Carbon dioxide sources vary enormously in their level of trace ethylene. Supplier scrubbing methods may be of variable efficacy. In our hands even when the supplier's quality control laboratories indicate virtually undetectable levels, our own routine ethylene scrubbing columns (containing proprietary potassium permanganate-based oxidation granules) can change colour at considerably different speeds from batch to batch of CO2 gas delivered. Ethylene scrubbing has been a substantial cost for growth chambers studies in my laboratory since identifying the problem with our supplies (Morison & Gifford, 1984). For FACE, the huge quantities of gas used might preclude routine on-site scrubbing. Ethylene, being a natural plant hormone, has growth inhibitory and specific developmental effects on some, but not all, species in the part per billion range. Apparently this is a problem that no FACE, and not all chamber, investigators have addressed in the past. As with the fluctuating CO2 concentration issue, the implication is that the methodology may understate the productivity-enhancing effect of elevated CO2. However, in some chambers having low air replacement rates, there is the added problem that ethylene naturally produced by the plants themselves can build up to inhibitory levels (Klassen & Bugbee, 2002). As CO2 and ethylene interact physiologically (at the higher CO2 levels involved in fruit ripening research, at least) this may also produce subtle confounding interactions in some chamber studies too.


In summary, as with global warming, there are substantial issues yet to be addressed with the CO2 fertilising effect, but the evidence for its existence in the real world continues to consolidate. Long-term FACE studies are showing that the CO2 fertilising effect on vegetation productivity may not, after all, be an artefact of ‘plant physiologists and their greenhouses’.