Interactions between increasing CO2 concentration and temperature on plant growth


James I. L. Morison Department of Biological Sciences, John Tabor Laboratories, University of Essex, Wivenhoe Park, Colchester CO4 3SQ, UK. Fax: + 44 (0)1206 873416; e-mail:


The global environment is changing with increasing temperature and atmospheric carbon dioxide concentration, [CO2]. Because these two factors are concomitant, and the global [CO2] rise will affect all biomes across the full global range of temperatures, it is essential to review the theory and observations on effects of temperature and [CO2] interactions on plant carbon balance, growth, development, biomass accumulation and yield. Although there are sound theoretical reasons for expecting a larger stimulation of net CO2 assimilation rates by increased [CO2] at higher temperatures, this does not necessarily mean that the pattern of biomass and yield responses to increasing [CO2] and temperature is determined by this response. This paper reviews the interactions between the effects of [CO2] and temperature on plants. There is little unequivocal evidence for large differences in response to [CO2] at different temperatures, as studies are confounded by the different responses of species adapted and acclimated to different temperatures, and the interspecific differences in growth form and development pattern. We conclude by stressing the importance of initiation and expansion of meristems and organs and the balance between assimilate supply and sink activity in determining the growth response to increasing [CO2] and temperature.

Abbreviations & symbols: A, net CO2 assimilation rate per unit area; Ci, Ca, intercellular space and ambient air CO2 concentration, respectively; FACE, free-air CO2 enrichment; GEC, global environmental change; RGR, relative growth rate; RA, Rb, relative increase in net CO2 assimilation rate and biomass (dry weight), respectively, with increase in [CO2]; T, temperature; θ, thermal time.


Both CO2 and temperature are key variables affecting plant growth, development and function, and both have changed in the recent past and are predicted to change in the future. One of the pre-eminent and incontrovertible manifestations of global environmental change (GEC) is the increase in atmospheric CO2 concentration, [CO2], over the last century or more, from approximately 280 up to 360 μmol mol–1 (e.g. Schimel et al. 1996). The present rate of increase (average of 1·5 μmol mol–1 year–1 over the decade 1984–93, Schimel et al. 1996) will continue for probably the next century, resulting in concentrations approaching twice the pre-industrial (560 μmol mol–1) by the middle or second half of the twenty-first century (AD) (IPCC scenarios IS92e and IS92a, respectively, Schimel et al. 1996). Increasing world population and economic activity is likely to lead to even higher concentrations (approximately 700 μmol mol–1) by the end of the century. A predicted consequence of this rise in [CO2], and in other so-called greenhouse gases such as methane, chlorofluorocarbons and nitrous oxide, is warmer temperatures at the earth's surface, as originally suggested by Arrhenius (1896). However, the extent depends on the many factors causing radiative forcing and on the many complex feedbacks between the different elements of the climate system. Recent model projections suggest a global mean surface air temperature increase of 1 to 4·5 °C by 2100 AD (dependent on scenario, Kattenberg et al. 1996). Indeed, the 0·3 to 0·6 °C rise of mean annual surface air temperature over the last century may be the first discernible effect of these recent global atmospheric changes (Nicholls et al. 1996). However, such coarse global mean temperature predictions hide aspects that are critically important to natural and managed ecosystems:

1 Recent predictions that include sulphate aerosol effects suggest substantial regional variation in temperature change, with generally much larger increases (up to 6°C by 2100) in the northern hemisphere higher latitudes and possibly decreases in some mid-latitude zones (Kattenberg et al. 1996).

2 Temperature change will have very different impacts on vegetation and ecosystem productivity, structure and composition dependent on the actual temperature range at that location.

3 Broad predictions of mean annual changes in temperature obscure important details about changes in (a) diurnal and seasonal patterns (b) frequency, timing and duration of extremes (e.g. high or low temperatures, late or early frosts) and (c) climatic variability (see Morison 1996). One example is that recent scenarios predict most warming in mid- and high-northern latitudes in late autumn and winter, and little or none (or even a cooling in mid-latitudes) in summer (Kattenberg et al. 1996), which could affect growing season length. Indeed, there is already evidence of a change in growing season length (Myneni et al. 1997). Another example is the strong evidence that, over land, the increase in night-time minimum temperature has been about twice the increase in the maximum (see Nicholls et al. 1996). Continuing changes in diurnal cycles will have very different consequences for plant growth compared with an even change in temperature throughout the 24 h.

Clearly, there could be major effects of such temperature and [CO2] changes on plants; however, an examination of both of these factors is outside the scope of any brief review. Instead, we examine plant response to the interaction of both temperature and [CO2] for two reasons. Firstly, because change in [CO2] is global, different ecosystems in very different temperature regimes will be affected, irrespective of any past or future changes in temperature. Secondly, biochemical photosynthesis models predict that the response of net CO2 assimilation rate per unit area (A) for plants with C3 metabolism to increased [CO2] is largest at high temperatures and smallest at low (e.g. Long 1991; Gifford 1992; Bowes 1996), providing a ‘fundamental basis for expecting an interactive effect of rising temperature and [CO2] at the very point of entry of carbon in photosynthesis and into ecosystems’ (Long 1991). This biochemical understanding along with the earlier largely erroneous application of the concept of single limiting factors (see Gifford 1992), has led to many suggestions that in cooler conditions (< 15 °C) plant growth will not be stimulated by increased [CO2]. Indeed, growth may even be inhibited and there is some experimental evidence in support of these predictions (e.g. Idso & Kimball 1989). However, short- and long-term responses of photosynthesis and productivity to increased [CO2] are very different (Drake, Gonzàlez-Meler & Long 1997). Furthermore, the diversity of the experimental approaches and results may have obscured some general features of responses to temperature (T) and [CO2] and their interaction with other environmental factors. This paper selectively reviews the evidence on the extent of temperature and [CO2] (T × CO2) interactions in plant growth, function and development. We have restricted our scope to terrestrial plants, as aquatic plants are dealt with elsewhere in this issue (Raven & Falkowski, 1999). A framework for understanding the complexities of whole plant responses to these key environmental factors is suggested.


Some 7 years ago, Lawlor & Mitchell (1991) commented on the lack of field data on plant response to [CO2] and particularly the need for studies where [CO2] is examined in combination with other variables. Since then many studies have examined T × CO2 interactions with a variety of experimental approaches (Allen et al. 1992; Lawlor 1996): (a) controlled environment cabinets with fixed temperatures (e.g. Ahmed, Hall & Madore 1993; Wayne, Reekie & Bazzaz 1998) (b) modified greenhouses, with or without additional lighting (e.g. Gifford & Morison 1993; Lawlor et al. 1993; van de Geijn et al. 1994; Casella, Soussana & Loiseau 1996) (c) solar domes (e.g. Rafarel, Ashenden & Roberts 1995) (d) open top chambers (OTC) around vegetation growing in field soil and branch bags on larger trees being used across different seasons (e.g. Teskey 1995; Tissue, Thomas & Strain 1997) or different sites (e.g. Mulholland et al. 1997), or modified OTCs with temperature control (e.g. Jones, Collins & Ingram 1995; Ziska et al. 1997) (e) temperature gradient tunnels (e.g. Rawson, Gifford & Condon 1995; Hadley et al. 1995; Horie et al. 1995), and (f) the most recent innovation, Free-Air Temperature Increase (FATI, Nijs et al. 1996).

Each of these techniques have their advantages and disadvantages, and their suitability depends on the objectives of the work. If the aim is to determine the effects of GEC on plant growth against a background (or ‘control treatment’) of natural conditions, then those techniques where plants are growing as stands in the field or in ‘natural’ ecosystems are most appropriate. Indirect effects of [CO2], through changing radiation interception, energy balance, or vapour pressure deficit, will be included more realistically in these designs than in experiments with isolated plants, for example, in growth cabinets. Furthermore, it is clear that the square waves of temperature time course used in some controlled environment studies cause different plant responses to [CO2] compared with the diurnal changes as experienced in the field (e.g. Rawson 1992; Moot et al. 1996). However, field experiments have their own limitations, for example, being done in one location under specific combinations of conditions which may not be applicable to other areas or other seasons. They should be complemented by use of ‘artificial’ environments to ‘disentangle’ the usual association of environmental conditions, in order to extend understanding of plant responses. Rather than attempting to define conditions for experiments to give a ‘true’ prediction of the plant responses more dynamic and flexible concepts must be employed.

One substantial problem in designing and interpreting experiments that manipulate T and [CO2] is that these two variables interact through the energy balance of the plant. The temperature of a plant organ is determined by its energetic exchanges with the environment, or ‘coupling’ (Monteith 1981). Plant and air temperature can differ substantially but this is often overlooked when comparing treatments and experiments based on air temperature information alone. It is also overlooked in many more empirical plant growth models, unless energy balance routines are included (e.g. Boote, Pickering & Allen 1997). Firstly, the plant–air temperature differences are similar to, or larger than the predicted changes in global mean surface temperature. In addition, the relationship between organ and air temperature is clearly highly changeable, on time scales of seconds to minutes, and longer when considering seasonal conditions. It has been well documented that because one of the effects of increased [CO2] is a reduction in transpiration, the temperature of the leaf (or other transpiring organ) increases. This has been observed at the scale of the individual leaf (e.g. Morison & Gifford 1984a; Idso, Kimball & Mauney 1987), the whole crop (e.g. Lawlor & Mitchell 1991; Kimball et al. 1995), and even the whole vegetated region (at least in models, e.g. Sellers et al. 1996).

In many cases, the warming effect (even a few tenths of a degree) acting cumulatively over long periods of time may significantly affect the rates of both ontogenetic development and of growth in leaf area and biomass (e.g. Pinter et al. 1996, 1997). Also, such changes in plant temperature may affect the carbon balance, and even determine heat ‘stress’ by increasing temperature above some optimum value for a process. For example, floret sterility in rice increased at high temperature (above approximately 33 °C) with + 300 μmol mol–1 CO2 due to increased canopy and ear temperatures at higher [CO2] (Matsui et al. 1997; see also Ingram et al. 1995 for work on rice cultivar variation in response to high temperature and [CO2]). There may also be effects of such changes in organ temperature, caused by altered transpiration rate, on small invertebrates and micro-organisms on plant surfaces (see Cammell & Knight 1992; for discussion on the indirect effects of environment on crop pests, and Norman 1989; for an example of leaf microclimate effects on insects). Clearly, these are additional examples of T × CO2 interactions which may substantially affect the growth and development of plants but we do not consider them further.

There are substantial problems with the design and execution of experiments looking at T × CO2 interactions. One is simply technical: obviously two factor experiments place heavy demands on space and experimental resources, and frequently the level of true replication (see Hurlbert 1984 for a detailed discussion of pseudoreplication) is probably insufficient to give enough statistical power and sensitivity to detect many of the interactions being investigated. Often, the demands are such that inadequate attention is paid to establishing and measuring other key environmental and plant factors. For example, the amount and timing of nutrient supply to plants has now been recognized as a major cause of differences between experiments (see Stitt & Krapp 1999 in this issue). A second is a more conceptual problem in the aims of experiments. In our opinion there are too many papers in the literature which seem to aim at detecting ‘effects of [CO2]’ on the particular process being studied, rather than at reliable quantification and examination of causes of the effect. Allied to this is the problem of defining a base-line for comparison. What are the ‘normal conditions’ with respect to temperature, light, water, nutrients, and even [CO2] which both fluctuates diurnally, and is increasing annually, etc.? Presumably they are those which give, in the long term, the average productivity or, more related to biological ‘fitness’, the reproductive success.

But how can such conditions be translated into practical experimentation? The lack of detail and clarity in experimental design and results leads to many problems of interpretation in the literature. As an example, it is not clear in many studies whether the impact of increased [CO2] on plant development is ‘direct’ (a consequence of [CO2]per se or of additional carbohydrate, etc.) or ‘indirect’ due to [CO2] altering the stomatal conductance, transpiration and organ temperature. Another example is the effect of increased [CO2] which often causes photosynthetic ‘down-regulation’. However, in some cases part of the observed effect can be attributed to different rates of leaf development, so that the comparison at the same time may be a comparison at different leaf ages (e.g. Delgado et al. 1994; Miller et al. 1997). If, in addition, there is a small, [CO2]-dependent temperature increase, then the identification and interpretation of what is a ‘CO2 effect’ is difficult.

Of course, many other mechanisms for indirect effects of [CO2] can be speculated about, such as changes in water or nutrient status. It might be clearer to think of a more readily identifiable ‘short-term’response to [CO2] that can be characterized for example by changing the [CO2] around plants or leaves grown in different [CO2], and the ‘long-term’effect which is determined by the whole combination of interacting effects triggered by the different growth [CO2], working through energy, carbohydrate, water and nutrient balances, and a variety of ecological processes such as competition and symbiosis.


There are many processes in plant growth at a wide range of scales that are affected by both T and [CO2] and therefore potentially affected by interactions. It is convenient to consider the impacts of T × CO2 interactions on the processes which determine carbon balance in the shorter-term, separately from the longer time scales of development and growth, which together lead to accumulation of biomass, and to yield. However, it should be noted that it is potentially misleading to divide up such intimately linked processes.

Carbon balance aspects

Numerous papers have reviewed the evidence for T × CO2 interactions in plant carbon balance (e.g. Long 1991; Lawlor & Mitchell 1991; Gifford 1992; Kirschbaum 1994; Bowes 1996), so the basic points need little repetition here. There are two main reasons to expect progressively increasing CO2 responsiveness of plant carbon balance at higher temperatures: (1) the decreased ratio of photosynthesis to photorespiration; and (2) the decreased ratio of gross photosynthesis to dark respiration in warmer conditions.

Stated very briefly, A of C3 plants is affected by the ratio of [CO2] and [O2] as these compete at the active site of the enzyme ribulose bis-phosphate carboxylase-oxygenase (Rubisco) for the primary acceptor, ribulose bisphosphate (RuBP). Rubisco catalyses both carboxylation and oxygenation of RuBP. Oxygenation results in photorespiratory loss of CO2, and carboxylation results in carbon fixation through the photosynthetic carbon reduction cycle. Increased [CO2] therefore reduces photorespiratory loss. However, increased temperature increases it, due to the reduced solubility of CO2 compared with O2 and the reduced specificity of Rubisco at higher temperature. Therefore, the largest effect of increased [CO2] on photosynthetic carbon fixation is at high temperatures (Lawlor & Keys 1993). Using a model of C3 photosynthesis Long (1991) has shown that (a) the temperature optimum for light saturated photosynthesis (Asat) increases by 5 °C with an increase of [CO2] from 350 to 650 μmol mol–1; (b) the decline of maximum quantum yield of photosynthesis with temperature is reduced as [CO2] increases, so that at 10 °C the increase with [CO2] is only 4% but it is 25% at 28 °C; (c) the light compensation point (which increases markedly with temperature) is substantially reduced at high temperature by increased [CO2]; and (d) photosynthetic acclimation (caused by reduced Rubisco activity and/or content) in higher [CO2], diminishes photosynthesis at low temperatures (< 15 °C). Kirschbaum (1994), working with essentially the same photosynthetic model, and highlighting temporal and biogeographic differences in ecosystem photosynthetic response to climate change, has emphasized that the CO2 sensitivity of photosynthesis is lowest at low temperatures and high [CO2]. He reported that the relative stimulation, RA (= [AhighAlow]/Alow), of Asat by 700 μmol mol–1 CO2 was only 0·3 at 5 °C, but this increased substantially at temperatures equivalent to equatorial regions, and at pre-industrial [CO2] (e.g. RA = 0·81 at 35 °C and 280 μmol mol–1). Both of these models assume that Ci/Ca is constant which may be reasonable if stomata adjust in parallel with mesophyll photosynthetic activity. This seems to be the case over the long-term for vegetation acclimated and adapted to particular environments (Amthor 1995; Ehleringer & Cerling 1995). In the short term, temperature responses of stomata and photosynthesis are not the same, particularly at low temperatures, when stomatal response is slow. Therefore Ci/Ca may change, and as Lloyd and Farquhar have shown (Lloyd & Farquhar 1996, Fig. 2), if Ci/Ca is low, then the photosynthetic sensitivity to [CO2] is increased, due to the curvature of the A/Ci response. It is not yet clear whether stomatal responses generally acclimate in parallel as photosynthesis does (Sage 1994; Morison 1998). However, a recent review of long-term experiments on plants at increased [CO2] (Drake et al. 1997) indicates that Ci/Ca ratio was almost identical for plants grown in normal ambient and increased CO2 (33 observations over 26 species, across a wide range of growing conditions from cool to warm).

Figure 2.

. Diagram indicating processes determining, and linkages between, assimilate supply (‘source’) and assimilate demand (‘sinks’) and the points of action of changing CO2 concentration (‘C’) and temperature (‘T’). PCR, photosynthetic carbon reduction; TCA. See text for explanation.

The prediction that stimulation of A by high [CO2] is larger at higher temperatures is supported by many, but not all, reports in the literature (e.g. Long 1991; Sage, S˘antrůc˘ek & Grise 1995). For example in Pinus taeda the relative stimulation of photosynthetic rate over several seasons was correlated with temperature in one study (Tissue et al. 1997) but not in another (Teskey 1997). Falk et al. (1996) concluded that photosynthetic acclimation of plants grown at low temperatures (0–10 °C) results in different photosynthetic responses in different species (particularly between herbaceous annuals and perennial evergreens) and therefore CO2 responsiveness may well be different. There are a few reports that indicate much larger than expected CO2 stimulation at cool temperatures (Greer, Laing & Campbell 1995; Bowler & Press 1996; Teskey 1997) and Bunce (1998) has suggested, that in wheat and barley at least, this is due to changes in the apparent specific activity of Rubisco with growth temperature. Clearly such changes could result in very different photosynthetic sensitivity to [CO2] at different temperatures. There is no evidence of change in the Rubisco specificity factor with environmental conditions; rather differences in relative [CO2]/[O2] and possibly regulatory processes are more likely sources of variation. Given the importance of such effects, there is surprisingly little experimental information on possible mechanisms of Rubisco regulation under interacting T × CO2 conditions during growth. Lawlor et al. (1995) and Delgado et al. (unpublished) found that changes in activation state and catalytic constant did occur due to both [CO2] and temperature and there were interactions which affected the photosynthetic rate demonstrating the underlying complexity of the photosynthetic regulation mechanisms.

The response of photosynthesis to environment is obviously complex: studies on winter wheat (cv. Mercia, Delgado et al. 1994) under simulated field conditions in one season showed only a 10% increase in photosynthetic capacity with 2 ×[CO2] compared with normal [CO2] over the full life of the crop. There were no changes in leaf protein (including Rubisco) or chlorophyll, although T + 4 °C above the ambient (and N deficiency) did decrease the photosynthetic capacity and carboxylation efficiency. Also the leaf duration was reduced and the leaf composition altered. However, in a similar study in another season on the same variety, there were progressive changes in leaf composition and CO2 assimilation which showed acclimation to increased [CO2] and an interaction with T in the leaves that were formed later (Lawlor et al. 1995).

The above discussion concerns only C3 plants. It has been generally considered that because of the CO2 concentrating mechanism in C4 plants they will show little CO2 stimulation irrespective of temperature (e.g. Gifford 1992; Bowes 1996). However, recent studies have shown substantial stimulation of biomass in C4 species (e.g. 28% increase in Panicum antidotale, Ghannoum et al. 1997) even if there are no apparent changes in mature leaf assimilation rates. Substantial responses of A to [CO2] have also been found (up to 30% increase with 2 ×[CO2] in Amaranthus retroflexus, Ziska & Bunce 1997), although the magnitude varies with species. The response of C4 species to T × CO2 interactions requires further examination. As the pattern of response to T of C4 and C3 plants is similar (although the quantitative characteristics are highly species-specific), it is likely that the complex responses of C4 plants to T × CO2 interactions will be similar to those of C3 plants.

As a general point, it is important to note that A per unit leaf area is, arguably, not the most important factor for overall plant growth. Rather it is the total supply of carbohydrate due to the combination of photosynthetic rate and leaf area, and its relationship to the demands for growth, respiration, storage, root exudates, etc., of the whole plant that must be considered (Lawlor 1995).

Response of plant growth to increasing [CO2] involves not only the photosynthetic responses of leaves but also the whole-plant respiration (Gifford 1992), and as Lloyd & Farquhar (1996) have emphasized ‘it is incorrect to simply assume that the [CO2] and temperature dependencies of plant growth are well reflected by the temperature and [CO2] dependencies of photosynthesis’. The growth response to [CO2] depends on the respiration rates per unit of dry matter, so that in theory plants with high maintenance respiration per unit of photosynthesis should show a greater relative sensitivity to [CO2] (Lloyd & Farquhar 1996). Respiration rates increase with temperature, so there should be a greater sensitivity to [CO2] at warmer temperatures. Generally, carbohydrate content decreases in plants grown in warm conditions and may limit respiration and growth (Rowland-Bamford et al. 1996). Therefore additional photosynthate as a consequence of higher [CO2] should allow a greater response of respiration to warmer temperatures. This is probably too great a simplification. Gifford (1994, 1995) has shown that, in wheat plants at least, the large increases in respiration in response to short-term (periods of hours) temperature increase were very different from the response of respiration during long-term growth at different temperatures.

There is still no model for plant respiration equivalent in detail to that for photosynthesis although most workers accept and use the approach of McCree (1974) which distinguishes between growth and maintenance components. The growth respiration rate is related to RGR, and its coefficient is widely held to be insensitive to temperature, and the maintenance respiration is assumed to respond exponentially to temperature according to the Arrhenius concept. However, Amthor (1995) and Gifford (1994, 1995) have both indicated that these generalizations do not hold in all cases. There is a large and growing number of observations that the respiration rate per unit dry weight declines with growth at high [CO2] (see Amthor 1997). For example, in rice grown in sunlit, controlled environment chambers, the crop respiration rate (per unit dry weight) declined throughout crop growth, and was fastest in very low [CO2] (160 μmol mol–1) and declined (up to 45%) as concentrations increased to 900 μmol mol–1 (Baker, Laugel & Boote 1992c). However, Poorter et al. (1992), from analysis of the literature, show that higher [CO2]increased respiration expressed per unit leaf area by 16% but decreased it by 14% when expressed by unit of leaf dry mass. It is necessary to distinguish between a short-term (‘direct’) response to increased [CO2] and the longer term effect (‘indirect’) of growth in different [CO2] (e.g. Amthor 1995; Drake et al. 1997). The direct effect, typically about a 20% reduction in specific respiration rate (from a survey of 53 observations with 23 species, Drake et al. 1997) is thought to be caused by inhibition of two mitochondrial electron transport enzymes, cytochrome c oxidase and succinate dehydrogenase (see Drake et al. 1999 this issue; Drake et al. 1997). The indirect effect is less pronounced, at least in long-term experiments (only 5% reduction, mean of 37 observations on 17 species, Drake et al. 1997), and often closely matches the reduction in tissue nitrogen content and increase in soluble carbohydrates in plants grown at higher [CO2] (e.g. Baker et al. 1992c; Gifford 1995).

In conclusion, although it is generally considered that warmer temperatures will increase the ratio of respiration to photosynthesis and increased [CO2] will decrease it, there is still considerable uncertainty about the balance of these processes and the extent and direction of acclimation (both regulation and changes in capacity) of both processes to T and [CO2] alone and when interacting.

Plant development and growth of organs

Accumulation of biomass and its partitioning between different organs depends on the net carbon balance (considered above), and on the ontogenetic development (i.e. the ordered initiation of meristems and determination of their characteristics, vegetative or reproductive) and on the expansion of the meristems into organs (morphogenesis). Plant ontogeny and morphogenesis are strong functions of temperature, and the relationships are often conveniently summarized in terms of thermal time, θ (e.g. Ritchie & NeSmith 1991). In determinate annual species (which includes most of the major food crops) warmer temperatures increase the rate of ontogenetic development and result in substantial shortening of the growth period. This results in less time for carbon fixation and biomass accumulation before seed set (e.g. Rawson 1992; Morison 1996; Bowes 1996). The effect can be very large. For example, in winter wheat Mitchell et al. (1993) found that time of 50% anthesis in cv. Mercia was 161 d after sowing in ambient T but 21d earlier in a 4 °C warmer treatment. In cv. Hereward temperature increases of only 2 and 4 °C (above the reproductive period mean of 14 °C) resulted in reductions of 16 and 29 d, respectively, in the duration of the reproductive period, compared with 130 d in the ambient treatment (Batts et al. 1997).

Initiation and expansion of roots, leaves, shoots, tillers, branches and reproductive organs are strongly driven by temperature. For any organ or structure, the important temperature relations are the rate of initiation, the rate and duration of expansion, and the start and rate of senescence. Warmer conditions both accelerate rate of organ initiation and shorten duration of organ growth (expansion) but the product of duration and rate is usually not constant over the range of temperature. In cooler temperatures leaves are often larger than in warm temperatures because of a slow rate of expansion but long duration and possibly because of continued assimilate supply (Rawson 1988). So, although it might be expected that the size of the plant would be the same after a certain θ is reached, this is not generally the case (Ritchie & NeSmith 1991). On the scale of the temperature differences associated with GEC, it might be expected that the impacts will be small. However, it is clear that even quite a small increase in T has considerable cumulative effects on the growth of leaves and other organs. This has been nicely illustrated in the FACE study on wheat in Arizona where the apparent effect of increased [CO2] in speeding up development by a few days could be largely explained by the slightly warmer temperatures in the high [CO2] treatments, caused by the reduced transpiration rates during the day and the increased mixing at night in the enriched CO2 treatment (Pinter et al. 1996 & 1997 and see earlier discussion).

The stimulation of total plant leaf area by increased [CO2] has been very widely observed, but many papers do not distinguish between greater leaf numbers (i.e. caused by initiation of new leaves perhaps on new tillers or branches) or increased leaf size. Certainly it has been long established in grasses and other monocots that the main effect of increased [CO2] is through tillering not by increasing leaf size at any particular insertion point, nor from more leaves on any stem (for recent examples see Tissue & Oechel 1987; Gifford & Morison 1993; Mitchell et al. 1993; Hakala & Mela 1996; Ghannoum et al. 1997; Batts et al. 1997, 1998a). In wheat, stimulation of tillering is pronounced (dependent on cultivar, e.g. Rawson 1995), even up to high [CO2] (e.g. 2500 μmol mol–1, Grotenhuis & Bugbee 1997) and only small or zero effects on the phyllochron have been reported (e.g. Frank & Bauer 1996; Batts et al. 1996, 1997; Li et al. 1997; Slafer & Rawson 1997). In rice increased [CO2] produced shorter phyllochron intervals in some experiments (Imai, Coleman & Yanagisawa 1985; Ingram et al. 1995), but not in others (Allen et al. 1995; Baker et al. 1996; Ziska et al. 1997). As with wheat, rice also shows the stimulation due to tillering. For example, Ziska et al. (1997) observed greater tillering with increased [CO2] at both ambient and ambient +4 °C temperatures, which increased early leaf area although the effect was no longer significant as maturity was approached. Part of the [CO2] stimulation of total leaf area may be due to continued expansion of organs already initiated. For example, in both spring and winter wheat cultivars the tiller survival is increased at 2 ×[CO2] (Batts et al. 1996; Mitchell et al. 1993, 1996). Most interestingly, Kinsman et al. (1996, 1997) have shown that higher [CO2] can substantially reduce the cell doubling times in the vegetative shoot meristem of Dactylis glomerata. This was similar to the effect of a 10 °C increase in temperature in a higher-latitude population but not in a population from the Mediterranean. Cell division is, presumably, responding directly to enhanced availability of photosynthate, and this effect probably underlies the enhanced morphogenetic development usually seen in monocots, although little detailed work has been done at this scale. In one study in rice, the apical dome increased in size in response to 2 ×[CO2] by 8 d after planting, and extra tiller buds had appeared within a further 8 d (Jitla et al. 1997). This latter work also showed that delaying enrichment largely removed the tillering response, emphasizing that we must be circumspect in extrapolating results from experiments where CO2 enrichment is delayed or not continuous.

In dicots, in contrast to monocots, both increased initiation rate and increased expansion rates and final size have been observed (see Taylor et al. 1994) in response to [CO2]. For example, Cure, Rufty & Israel (1989) found that in soybean in 700 μmol mol–1 CO2 the main stem and branch leaves emerged and expanded more rapidly than in 350 μmol mol–1 leading to a much larger plant (75% increase in mass), with many more leaves (31% increase in area), but with final size per leaf not much affected. In another study, with the same species, the plastochron interval was reduced substantially by a 10 °C rise in temperature, but by a much smaller amount by 2 ×[CO2] (Baker et al. 1989; 4·2 to 3·2 and 3·9 to 2·6 d per leaf, across 26/19 °C and 36/29 °C growth temperature regimes at 330 and 660 μmol mol–1 concentrations, respectively). In Abutilon theophrasti and Amaranthus retroflexus there were large increases (approximate doubling) in plastochron with temperature increases from 18 °C to 28 °C (Ackerly et al. 1992) but while there were statistically significant increases of plastochron with increase in [CO2] from 400 to 700 μmol mol–1 they were small or even negligible in comparison with the T effect. In three native British herbs, there were increased rates of leaf initiation (Ferris & Taylor 1993) and leaf size was increased, mostly caused by increased cell expansion, but also partly due to increased cell numbers per leaf (Ferris & Taylor 1994). It is obvious that even small effects on leaf size, duration and overall leaf area duration can have major impacts on total plant photosynthesis (Lawlor 1995) and improved understanding of the mechanisms controlling the responses of morphogenesis to T and [CO2] is urgently required.

Temperature is a key modifier of ontogenetic development rate, as well as interacting with other environmental conditions, for example, photoperiod, which act as ‘switches’ in development. As discussed above, the initiation rate is rarely dependent on carbohydrate supply, although expansion is, so little effect of increasing [CO2] on ontogenetic development is anticipated. In general this is the case, as discussed by Rawson (1992) who reviewed a number of papers about [CO2] effects on development of crops. Using the example of wheat again only small or zero effects on rates of vegetative and reproductive development rates have been reported (e.g. Marc & Gifford 1984; Mitchell et al. 1993; Batts et al. 1996, 1997), and certainly much smaller than the effect of temperature. In general, little interaction has been observed between T and CO2, although there appear to be small differences in responses of ontogenetic development between cultivars of wheat (G.R. Batts, personal communication). In other species larger effects have been observed. In rice the effects of increased [CO2] on ontogeny observed in different experiments are not consistent with some reports of little or no effect (Manola et al. 1994; Baker et al. 1996; Ziska et al. 1997), and others of earlier flowering (Imai et al. 1985). The effects of relatively large changes of [CO2] are generally considerably smaller than the effects of small changes in temperature. Similarly, the patterns of flowering response to [CO2] in soybean are very variable with no effect, advancing, or retarding all being reported (see Ellis et al. 1995). It is interesting that in sorghum (Sorghum bicolor) flowering was substantially delayed (e.g. from 45 to 59 d with increase in [CO2] from 360 to 720 μmol mol–1 at a temperature of 30/21 °C, Ellis et al. 1995) which confirmed a number of earlier studies (e.g. Marc & Gifford 1984). Although the suggestion has been made that the direction of the effect of [CO2] on duration to flowering is related to whether species are short- or long-day plants (Reekie, Hicklenton & Reekie 1994) this does not always appear to be the case (Ellis et al. 1995). No mechanism has yet been found for the effect of increased [CO2] on sorghum, but it may be related to the substantial delay in development that sorghum shows in response to drought.

In non-crop species the effects of [CO2] on development are less well documented, probably in part because the basic information on developmental scales and temperature and photoperiod responses has not been described in the same detail as for the major crops. Studies on annual species of both C3 and C4 types, have demonstrated a number of effects of increased [CO2] on development which clearly differ between species. For example, Garbutt & Bazzaz (1984) showed that Phlox drummondii and Datura stramonium flowered 2–4 d earlier (usual duration approximately 38 and 30 d, respectively) at 600 compared to 300 μmol mol–1 but there was no effect on A. theophrasti. Larger effects were found in A. retroflexus plants which flowered at 27 compared with 36 d for 700 compared with 350 μmol mol–1, while Setaria faberii plants were slowed down by increased [CO2] and flowered at 71 rather than 56 d (Garbutt, Williams & Bazzaz 1990). Bazzaz and colleagues make the point that such changes in flowering date, and in numbers of flowers produced may have important consequences for insect pollinators and for the community dynamics in general. From the viewpoint of understanding mechanisms the ‘direct CO2 effect’ must be distinguished from the ‘indirect CO2 effect’ but the ecological consequences are the same.

Biomass accumulation

Accumulation of biomass and its partitioning between different organs depends on both the net carbon balance and the ordered initiation and expansion of organs (considered above). Expanding organs are sinks for assimilate which interact with the assimilate sources and an understanding of this interaction is critical to understanding plant responses to environment (Lawlor & Keys 1993; Farrar 1996). As development and growth rates of sinks are strongly reduced at low temperatures, so is demand for assimilates. Therefore, as Rawson (1988) and Gifford (1992) have pointed out, at low temperatures sink growth and metabolism are the controlling processes and not carbohydrate supply, so plants may show very little response to increased CO2, depending on the response of CO2 assimilation to temperature. As Paul, Driscoll & Lawlor (1991) have shown, low T generally results in accumulation of carbohydrates, indicating that growth or storage are limiting. Conversely, at high temperatures, sink demand is large and assimilate is depleted so there should be a marked positive response to increased [CO2]. Temperature sensitivities and optima for photosynthesis, respiration, development, and for overall growth vary between species, between ecotypes and varieties, and with acclimation and developmental stage (e.g. Björkman 1981; Coleman & Bazzaz 1992; Rawson 1992; Leegood & Edwards 1996; Falk et al. 1996). Given these differences, ‘perhaps it is naive to expect generalized patterns of growth [stimulation]’ to quote Rawson (1992). However, a basic picture or ‘model’ of interactions between processes does emerge from the wide range of experimental results, with the same processes in different species or experiments being affected by environmental conditions in a similar general way, but with large quantitative differences. As a consequence the relation between, for example, total biomass and T × CO2 interactions is extremely variable (Rawson 1992). Biomass allocation patterns between organs are a key determinant of plant growth, and will profoundly influence CO2 responsiveness as well as possibly be influenced by them (see, e.g. Callaway et al. 1994; Jitla et al. 1997).

Determinate and indeterminate plants

Several previous reviews have distinguished between the expected responses of determinate and indeterminate plants to [CO2] and T. Determinate plants are considered, in the more simplistic analyses, to offer only limited sink capacity for assimilates because their developmental regulation is strictly controlled and the number and size of organs made are limited and relatively fixed (Lawlor & Keys 1993). Thus the amount of material required for growth is also limited, and likely to offer only a small response to increased [CO2]. Indeterminate plants do not suffer, so it is hypothesized, from developmentally restricted capacity for growth or storage. Rather they are considered to be source limited and, given adequate assimilates, will show a very large response to increased [CO2]. This difference in determinacy has been suggested as the reason for the higher response of biomass to [CO2] in, say, cotton or soybean than in sunflower or wheat (see Poorter 1993; and Pinter et al. 1996). In a comparison of closely related cultivars of soybean (Ziska, Bunce & Caulfield 1998) it is striking that the cultivar with the smallest seed yield response to 2 ×[CO2] (13%, compared to 48% for the largest response) was the only determinate cultivar. Some species are not as indeterminate as usually considered. One example is potato, where 2 ×[CO2] stimulated a 40% increase in tuber yield, due to increased tuber number, but the canopy senesced earlier, perhaps due to tubers reaching maximum size earlier (Miglietta et al. 1998). Another example is sugar beet, which has been regarded as indeterminate because it grows as a vegetative crop with an early formed, large tap root which accumulates sucrose during a long growing season. In theory, availability of assimilate should determine production, with continued growth providing the sink to exploit extra assimilates in increased [CO2]. However, this is not the case (Demmers-Derks et al. 1998). Sugar beet grown in simulated field conditions at ambient temperatures increased in total dry mass by approximately 25% due to 2 ×[CO2]; this is a comparable increase to that in winter wheat grown in very similar conditions (Mitchell et al. 1993). Increasing the temperature by 3 °C throughout the growing season accelerated growth in the early periods, giving more leaves and leaf area and increased light interception and dry matter. However, senescence of leaves accelerated and the plants matured earlier. Increased [CO2] did compensate for the warmer growing conditions but not proportionally more than at normal temperature. Thus, there was no T × CO2 interaction for sugar-beet, a so-called indeterminate plant. An explanation is that the developmental processes determine the sink growth and thus capacity and the duration of growth. Warmer conditions accelerate development and also maturation, even in an apparently indeterminate plant, thus assimilate supply is not the limiting factor. As with cereals it is sink capacity which is altered by higher [CO2] but in sugar beet the capacity changes through modification of organ size rather than number, and the increase in capacity is limited.


In the preceding sections the discussions of T ×[CO2] interactions in carbon balance, development, expansion and biomass accumulation emphasize strongly that growth responses cannot be readily predicted from photosynthesis responses alone. In summarizing the results of over 10 field CO2 enrichment studies, Koch & Mooney (1996; p. 417) suggested that the growth stimulation is much less and more variable than stimulation of the rate of photosynthesis, particularly in natural ecosystems. A number of papers and reviews emphasize lack of CO2 effect on growth at low temperature, although there is very little firm evidence. Idso et al. (1987) predicted (by extrapolation of experimental data) that [CO2] would have a negative effect on growth below 18·5 °C, later revised to below 11 or 12 °C (Idso & Kimball 1989). In winter wheat, increased [CO2] at temperatures below approximately 10 °C did not significantly increase biomass production (Mitchell et al. 1993 and unpublished). In those experiments, cool temperatures were correlated with decreased irradiance and it is likely that this affected the response and the estimate of T × CO2 interactions.

Rawson (1992) carefully reviewed the topic and found no evidence for a negative response of growth to increased [CO2] at 12 °C. The CO2 stimulation effect did increase at higher temperatures, by approximately 2% per °C, for results from the literature that spanned the range 12 to 28 °C. In the same study Rawson (1992) reported a specific comparison for plants grown well-spaced and fertilized for the same thermal time (from sowing up to approximately first flowering), and the mean shoot growth stimulation caused by a doubling of [CO2] for cotton, wheat, sunflower, cowpea and soybean was 1·32 at a mean temperature of 20 °C and 1·47 at 30 °C. However, Rawson (1992) did highlight the large variation in the responses to [CO2] and temperature and clearly identified from studies on several crop-plant species that the variation can be partly attributed to plant age (i.e. growth stage), partly to plant spacing effects, and partly to interaction with other environmental conditions, particularly radiation and nutrients. Growth stimulation by [CO2] is usually higher in young plants, because of the high RGR, and the positive feedback caused by expansion of canopies, before self-shading limits growth.

In a more detailed study using temperature gradient tunnels with winter wheat in both winter (cool, ≈ 10 °C) and summer conditions (warm, ≈ 20 °C) Rawson (1995) found that response of biomass and grain yield to 2 ×[CO2] was larger in the summer (average increase 32 and 35%, respectively) than in winter (increase of 7 and 6%, respectively). However, these averages may reflect the very high temperatures in the summer experiment (> 40 °C on many days, and daily maxima above 30 °C on almost all days), and the markedly larger grain yield response to [CO2] during the summer experiment caused by only a 2 °C rise (change from 34 to 61% (Rawson 1995). Possibly this was an example of the alleviation of high temperature stresses by increased [CO2] (see below), resulting in much larger yield increases which are not related to simple photosynthetic T × CO2 interaction models. In experiments under more temperate conditions, Mitchell et al. (1993) measured no significant T × CO2 interaction, except for a decrease due to higher [CO2] at ambient temperature with deficient nitrogen supply.

Clearly, there is considerable variation in the responses noted in the studies discussed in the preceding paragraphs. Is there any general relationship of CO2 responsiveness with temperature discernible from the literature? In Table 1 we list the relative growth response to [CO2] (Rb) for a number of different species and experiments (18 species, 106 observations in all), attempting to extend the results summarized by Rawson (1992; Fig. 4). Obviously this is a very incomplete list, but we have particularly included studies where different temperatures were examined in the same experiment, or at least in the same experimental system, and where mean temperatures over the whole experiment were clearly defined or could be estimated from the information given. We have chosen to highlight species where there are several independent observations, which inevitably biases the information towards the major crops. We express the growth response to increased [CO2] in two ways. Firstly, simply as the change in biomass (M) due to increased [CO2] relative to the biomass at low [CO2], or Rb = (MhighMlow)/Mlow. As the range in [CO2] used in these disparate experiments is large this cannot be ignored, so we have also used the ‘biota growth factor’, or β-value = Rb/ln (CO2high–CO2low), which has been widely used in modelling. Amthor & Koch (1996) have highlighted the problems in use of β in global change modelling, but here we are simply using it to summarize experimental data. One problem with β is that because of the asymptotic growth response to [CO2] it usually declines with increasing [CO2] (Amthor & Koch (1996). In considering the data in Table 1 it should also be noted that the experiments were of very different durations, and are for a mixture of experimental systems and conditions, some with spaced plants, some in stands. Furthermore, the data used are a mixture of either just shoot biomass (above-ground biomass) or total biomass. However, the studies used are not unrepresentative, as indicated by the overall average Rb and β-values {0·301 ± 0·212, standard deviation (SD) and 0·503 ± 0·330 respectively}, which are close to the average values usually reported for the growth increase with a doubling of [CO2] (see, e.g. Amthor & Koch 1996).

Table 1.  . Relative growth stimulation by increased CO2 concentration at different temperatures. Stimulation is expressed as Rb = (MhighMlow)/Mlow (where M is dry mass of plant material) and β = Rb/ln (CO2high/CO2low) Thumbnail image of

The first feature to note in Table 1 is the variability of CO2 response, either expressed as Rb or β, and neither of them show obvious relationships with temperature (Figs 1a & b), with as large a range in values at both warm temperatures (> 25 °C) as at cool ones (< 15 °C). This is the case when considering all the information together (Fig. 1) or individual experiments which included different temperatures (Table 1). For example, there is little consistency in whole plant biomass β-values across the extensive experiments of Baker et al. (1996) with paddy rice with temperature ranging from 24 to 34 °C. The β-values ranged from 0·12 to 0·53, with these two extremes coming from two different experiments at almost the same temperature (24·2 and 25·1 °C). Indeed, Baker & Allen (1993) conclude that there were no significant interactive T × CO2 effects on above-ground biomass, grain yield or yield components in IR30 rice. In contrast, in later studies with cultivar IR72 the CO2 response did increase with increasing temperature, although the overall effect of [CO2] was not significant (Allen et al. 1995). In irrigated field experiments with IR72, Ziska et al. (1997) found a large reduction in Rb with an increase in T of 4 °C in a wet season experiment, but a small increase in the dry (= higher radiation) season. Table 1 shows several other examples where a reduction of CO2 stimulation with increasing temperature have been found (e.g. soybean Baker et al. 1989). Another example is a growth chamber study over 35 d in warm conditions on soybean and five weed species (Tremmel & Patterson 1993), where there was either no difference in CO2 response of growth between 26/19 °C and 30/23 °C or there was less response at the warmer temperatures.

Figure 1.

. Relationship between (a) relative stimulation of biomass, Rb by increased CO2 concentration and (b) biota growth factor, β, and mean growth temperature, derived from values in the literature (see Table 1). Symbols indicate: □, rice; ▵, spring wheat: ◊, winter wheat: ▿, soybean; ○, other species.

Secondly, the dearth of data for temperatures less than 10 °C is obvious. The few studies at low temperatures are mostly those from longer term quasi-‘field’, multiyear experiments, where because temperature varies a simple mean seasonal temperature may be inappropriate. This contrasts with the studies that Rawson (1992) examined from controlled environment studies with single species, and where temperature was well-defined (and usually constant over the season). Results from the field or simulating it, where T (and other factors) was variable (Rawson 1992; Mitchell et al. 1993; Rawson 1995; Batts et al. 1996, 1997, 1998b; Wheeler et al. 1996a,[176] b) show just how difficult it is to derive a single response function. We agree with Rawson (1992) who analysed single species, controlled environment studies with often constant T, that it is ‘rather naive’ to expect a simple response from such disparate studies and that ‘results cannot be forced into a simple framework’.

Responses at low temperature

It is noteworthy how few experiments have examined stimulation of growth by increased CO2 at low temperatures (e.g. in the range 0–10 °C). This may simply reflect the convenience of working at close to ‘room temperature’ (i.e. 20–25 °C). However, it should be remembered that (by definition) temperate agriculture zones, and many ecosystems covering very large areas have mean annual temperatures, and even growing season temperatures, well below 20 °C. Long (1991), emphasizing the importance of the T × CO2 interaction, referred to two long-term field studies, in the Alaskan tundra, and in the Chesapeake Bay wetland, which confirmed the expectation that the warmer site would show a larger CO2 stimulation. More recently Oechel & Vourlitis (1996; p. 172) attributed the lack of response of the tundra ecosystem after 3 years of growth in high [CO2] both to low availability of nutrients and responsiveness of the species to nutrients, and to low temperatures limiting sink growth potential. Plants adapted to such conditions may also have slow potential growth rates. In the warmer Chesapeake Bay salt marsh, Drake et al. (1996; p. 209) attributed the large year-to-year variation in stimulation of net ecosystem CO2 exchange to environmental interactions, in particular the larger stimulations in very warm (and dry) years when increased [CO2]‘moderated the impact of stress on productivity’. Körner et al. (1996; p. 178) have pointed out that much of the photosynthesis of a Carex curvula sedge mat community in alpine grassland growing in cooler conditions takes place at warm leaf temperatures (> 20 °C) due to microclimate effects on the short vegetation, but that low night-time temperatures in the very short growing season (2–4 months) may inhibit cell division, differentiation and expansion and hence prevent formation of sinks for additional carbohydrate. They observed substantial stimulation of net ecosystem CO2 exchange (41% for an increase in [CO2] from 355 to 675 μmol mol–1) but no change in biomass after 2 years (Körner et al. 1996) and again nutrient modification was very important. They suggest that this ecosystem evolved to be non-carbon limited (Körner et al. 1997). Large root systems may use and store considerable amounts of carbohydrates but such systems are likely to be in approximate source–sink balance. However, there were large differences between the responses of individual species within the grassland communities. Certainly, there was no evidence that net carbon uptake increased in the warmer part of the growing season, compared with the cooler part (Körner et al. 1997).

In other experiments with grasses at cool temperatures, above-ground biomass did not respond to increased [CO2] in some experiments (Saebo & Mortensen 1995, 1996; Hakala & Mela 1996), but did in others, with the same species (Mortensen 1997). In one experiment with Festuca pratensis (Hakala & Mela 1996) there was little or no response of yield of the swards at normal temperatures (≈ 5–20 °C daily mean temperatures over the growing season) although leaf photosynthetic rate increased by 40–60% with 2 ×[CO2]. However, when temperatures were increased by 3 °C there was a 10 to 29% increase in sward yield with increased [CO2]. Similarly, Newton et al. (1994) found an interaction with temperature in the above-ground growth rate of pasture turves (dominated by ryegrass and white clover), in that an effect of 2 ×[CO2] was only detectable at 22/16 °C and not at 16/10 °C or 10/4 °C. In contrast, Mortensen (1997) could not detect a difference in the growth response of three species (Phleum pratense, Lolium perenne and F. pratensis) to 2 ×[CO2] at 14 °C or 19 °C mean temperature.

In studies where there is increased assimilation rate, but no increase in biomass, it is not clear where the increased assimilate goes. Possibly it is partitioned into root structure or storage or it is exuded into the soil or perhaps respired. In several other studies with temperate crop species it is clear that growth stimulation by increased [CO2] can occur at temperatures below 15 °C (see Table 1). Indeed, considerable growth stimulation may occur in winter wheat during the winter, at mean temperatures as low as 6 to 8 °C (e.g. Wheeler et al. 1996b) but not invariably (Mitchell et al. 1993). It is thus clear that there is considerable variation between experiments, with the same species or closely related varieties. This variation echoes that found by Hunt et al. (1991), in their survey of the CO2 response of seedlings of 27 herbaceous species, grown at 18 °C, where the increase in the ratio of total biomass at 540 μmol mol–1 to that at 350 μmol mol–1 ranged from 0·01 to 0·49.

Effects on cold and freezing tolerance

There is little and contradictory evidence that [CO2] affects low temperature and freezing tolerance. In Vaccinium myrtillus doubling of [CO2] did not affect subsequent freezing resistance (Taulavuori et al. 1997) but freezing resistance of overwintered buds of Betula alleghaniensis was improved (Wayne et al. 1998). The frost hardiness of Pinus sylvestris saplings increased during exposure to higher temperature and [CO2] (Repo, Hanninen & Kellomaki 1996). However, seasonal timing was also altered so that buds hardened later and de-hardened earlier (Repo et al. 1996). In contrast, in Picea sitchensis increased [CO2] delayed bud-burst, so reducing the risk of late frost damage. Furthermore, as the buds set earlier, the growing season was appreciably (average of 24 d) shorter, although it varied between clones (Murray et al. 1994). In both Picea sitchensis (Murray et al. 1994) and Salix phylicifolia (Silvola & Alholm 1993), nutrient status clearly interacted with [CO2], presumably through carbohydrate supply, in determining bud phenology and response to temperature.

Responses at high temperatures

At the other end of the ecosystem temperature scale, comparatively little work has been done with plants of warm environments, although responses are similar to more temperate plants (see Hogan, Smith & Ziska 1991). The growth response, Rb, of seedlings of five C3 species in an experiment in the tropics (36·5 and 21·2 °C maximum and minimum temperatures), at 712 compared to 354 μmol mol–1 CO2, ranged between 0·10 and 1·72 suggesting that there is as much variation between tropical as temperate species (Ziska et al. 1991). The tropical plant okra (Abelmoschus esculentus) survived to maturity in low temperature (20/14 °C) when grown in 450 μmol mol–1 CO2 or above but not at 350 μmol mol–1 (Sionit, Strain & Beckford 1981). This, and other possible effects of [CO2] in alleviating chilling stress in tropical plants (also see below) led Hogan et al. (1991) to suggest that increased atmospheric [CO2] might allow a shift of tropical lowland species into cooler zones, for example to higher elevations.

High temperature stress is expected to be alleviated by increased [CO2] due to the improved photosynthesis under hot conditions (Long 1991), especially if plants are limited by assimilate supply, such as during rapid growth, e.g. rice during growth at very high temperatures (Allen et al. 1995) or after flowering (see Ahmed et al. 1993; Rawson 1988, 1992; Baker & Allen 1993; Rowland-Bamford et al. 1996) but no direct benefit was observed from CO2 enrichment in high temperature studies with Gossypium barbadense (Reddy, Hodges & McKinion 1992) or cowpea (Ahmed et al. 1993). Indeed if increased [CO2] does decrease evaporative cooling crop temperatures may be higher, thus increasing the likelihood of periods of extreme temperature (e.g. Bassow, McConnaughay & Bazzaz 1994). Conversely, reduced stomatal conductance may slow the rate of transpiration and decrease the onset, severity and impact of water stress (Morison 1993) although the effects in the field may be smaller than in controlled conditions (Lawlor & Mitchell 1991). There may also be other interactions between [CO2] and water status that affect temperature responses. For example, chilling tolerance may be affected; improved water relations due to doubling of [CO2] at low temperature (6·5 °C for 24 h) substantially decreased chilling damage to bean (Phaseolus vulgaris), cucumber and maize (Boese, Wolfe & Melkonian 1997).

High temperatures can greatly damage specific growth stages with flowering and fertilization being particularly affected, for example in rice (Allen et al. 1995), where very brief (≤ 1 h) high temperatures (> 35 °C) at anthesis can cause sterility but not in the preceding or following day (Yoshida 1981). Wheat is also similarly affected (Mitchell et al. 1993; Wheeler et al. 1996b). There is little evidence that higher [CO2] itself greatly affects these responses but may do so via the energy balance feedback (e.g. Matsui et al. 1997) or other mechanisms such as improving water status enough to allow pollen to be shed in high temperatures (Ingram et al. 1995).

More general alleviation of high temperature effects by increased [CO2] could come from reduced respiration (found in Betula alleghaniensis, Wayne et al. 1998; Dactylis, Ziska & Bunce 1993; Pinus sylvestris,Wang, Kellomaki & Laitinen 1995) or improved antioxidant systems due to greater availability of carbohydrates (Sage 1996; Rao, Hall & Ormrod 1995). Conversely, if photosynthetic acclimation to increased [CO2] occurs, this could increase the risk of photo-inhibition with bright light and high temperatures (see Roden & Ball 1996). Also if decreased nitrogen content resulted in impaired synthesis of protectant systems then greater damage could ensue. However, no effect of increased [CO2] on heat-shock resistance was found in three species (Coleman et al. 1991). Hogan et al. (1991) considered that increasing [CO2] would alleviate high temperature stress, but the evidence is very sparse and some comes from work with very high [CO2]. We do not believe that it is possible to generalize about the effects of increased [CO2] in extreme temperatures, either on specific mechanisms or globally, and more evidence is required.


The term ‘yield’ refers to the plant part harvested, for example grain in cereals or sugar (sucrose) extracted from the tap root of sugar beet. Similarly ‘quality’ is defined in terms of characteristics deemed of value. The terms have little biological meaning as the components and composition of the plant that constitute yield are often very different. Therefore, it is not surprising that comparison of effects of environment on yields for different species and experiments is often inconclusive. In the case of cereals, such as wheat and rice, where the grain yield per area (Y) is determined by the number of ears or panicles per area (Ne), the number of grain per ear (Ng) and the mass per grain (Mg), comparison is more meaningful. The primary effect of increased [CO2] in these species is to increase numbers of tillers (see earlier discussion) and this largely determines Ne in wheat (Rawson 1992; Mitchell et al. 1993; Batts et al. 1996, 1997) and rice (Baker, Allen & Boote 1992a,[11]b; Ziska et al. 1997). Increased [CO2] increases Ne both by stimulating production of tillers early on and by increasing survival, whilst warmer temperatures may increase initiation of tillers but increase their death (Mitchell et al. 1993; Batts et al. 1996). Coleoptile and late tillers are affected most and these produce the smallest ears with fewest grains. Hence the effects of T × CO2 interactions on ear production and survival and on yield are complex. There is considerable variation between years or seasons (Batts et al. 1998a,b; Ziska et al. 1997) and with nitrogen supply and light (Mitchell et al. 1993, 1996; Ziska et al. 1996) and water supply in both of these crops. The average Ng depends on the number of grains in different tiller classes and on the conditions experienced during growth. Increased [CO2] generally increases the number of grains, the proportion of grains reaching maturity and the number of filled grains, whereas warmer temperatures decrease these attributes particularly when exceptional temperatures induce sterility (for wheat examples see Mitchell et al. 1993; Wheeler et al. 1996a,b). In rice high temperature may depress Ng substantially (Baker et al. 1992a,b) with the effect greater at ambient than increased [CO2] (Ziska et al. 1997). Finally, Mg depends on conditions during grain filling as well as on the number of grain per ear or panicle. Increased [CO2] generally does not increase Mg greatly (unless Ng is not able to respond) but warmer temperatures decrease it substantially, both in rice (Baker et al. 1992a,[11]b; Ziska et al. 1997) and wheat (Mitchell et al. 1993; Wheeler et al. 1996a,b).

However, such generalizations must be qualified. Production of yield is influenced by conditions throughout the different developmental and growth stages. As final yield is determined by several processes and each is potentially altered in different ways in different crops and with different conditions, including [CO2], T and solar radiation, the permutations are numerous. Thus, there is considerable variation between different experiments, even on the same variety under very similar conditions. Rather than attempting to define a set of ‘true responses’ for a particular variety or, even more difficult, for a range of varieties, it is more instructive to consider the possible factors which may be responsible for the variation in the T × CO2 response of yield.

Mechanism of response to [CO2] and temperature

Consider Fig. 2 as a general statement of the mechanisms determining plant growth, biomass accumulation and yield. The balance between the source and the different sink processes will change with developmental stage and with conditions. That is, there is not a fixed relation of source to sink activity (Farrar 1996). Rather, in the short term the balance will be very dynamic and very dependent on the store of CH2O, source of N assimilates, etc. There may be large imbalance between source (photosynthesis) and sink (growth) with relatively little effect on growth, whereas in the long term (periods of days) it would tend towards a more steady state and have larger effects on growth.

Remaining with the analysis of cereals, in vegetative growth the formation of leaves on the main stem and then growth of tillers are the primary sinks for assimilates. Their rate of growth and respiration (both maintenance and growth) depends on the temperature, whereas the production of assimilates depends on temperature but especially on light intensity and [CO2]. Warm conditions accelerate growth and respiration so demand increases but this may not be matched by assimilate supply if light is limiting, due to latitude, season or cloud. Rapid growth in warm but dull weather may result in an assimilate deficit, which in the short term is met from storage. However, if the longer-term demand for assimilates cannot be met then growth of organs, for example the later-formed tillers, ears and grains, is impaired. Regulatory mechanisms in this important area are poorly understood for crops grown in current or future conditions (Fiel 1992). If such starvation persists as the development progresses towards formation of reproductive organs, then the impact on ear formation and yield will be marked. Mitchell et al. (1996), showed that reduced light during development, between terminal spikelet and start of grain filling, decreased yields of winter wheat.

Shade during a period encompassing anthesis had the greatest effect, presumably because the assimilate demands during this period were large. This effect is well established in many crops (e.g. rice, see Islam & Morison 1992). However, in wheat even though a doubling of [CO2] increased biomass prior to the shade treatments, there was no significant interaction between [CO2] and shade on yield (Mitchell et al. 1996). The interactions between shading at different periods and [CO2] on Mg were complex, suggesting that the source-sink processes (including assimilate fluxes to and from storage) at the time of grain formation and the early stages of grain set are of considerable importance. They are, however, little understood. These results also warn us that the reduced radiation in many experimental systems combined with abnormal temperatures may cause very different responses to those in the field.

Grain ‘filling’ (growth) may also be analysed using the model of Fig. 2, but simplified because the formation of new structures is largely stopped and assimilate supply to the grain from recent photosynthesis and from stem and root storage dominates. Higher [CO2] increases the assimilate supply but the response of the grain will depend on how many grains are drawing on the supply. With increased Ne and Ng the total yield increases but not necessarily Mg. If Ng is low, due to a partial failure of grain set such as with high temperature inhibition of anthesis, etc. or to N deficiency, then large grain may be produced. Conversely, if there are high numbers of grain to be supplied this may lead to decreased Mg. Possibly, deficiency of nitrogen during early grain growth may lead to competition between grains resulting in abortion. Warmth during grain filling has the most profound effects for it determines the duration of the process as well as rates of photosynthesis and of storage remobilization and transport. The shorter the duration the smaller the grain; sometimes the rate of grain growth is unimpaired and only duration is affected, sometimes both. Clearly the impact of increased [CO2] is to increase the rate of grain growth (Wheeler et al. 1996a), particularly at warmer temperatures, but there is little good evidence of the relative importance of supply versus developmental limitation. It is highly likely that this balance changes with growing seasons and experimental conditions and this probably accounts for the variation in responses seen. The interactions between grain number, the rate of grain filling, and assimilate supply per grain are the key to understanding what regulates production under different environmental conditions including [CO2] and T (Rawson 1988; Rogers et al. 1996). This applies equally to other seed crops.


The effect of GEC on the quality of yield products is critically important for both the biological success of plants and for the nutritional and industrial quality of the products. Quality encompasses many attributes and depends on the components of importance. Relatively little attention has been directed towards quality, with most focused on crops although the importance of ‘quality’ in determining the ecological performance of species and their interaction with insects and other organisms has long been recognized. Two aspects of quality of most concern are changes in size and altered composition of individual components of yield. In wheat, for example, the size of individual grain may affect the suitability for particular processes and the economics of milling. In sugar beet the size of the tap root and its shape may influence the efficiency of harvesting, transport and cleaning before sugar extraction. However, changes in the composition of seed or root and the impacts on economics under GEC have not been extensively analysed. The tendency for the nitrogen content of plant material, ranging from grain to herbage, to decrease with increased CO2 has been widely reported (e.g. see Cotrufo, Ineson & Scott 1998). These trends appear to be exacerbated by deficient N supply, but it is not clear in all experiments (see Cotrufo et al. 1998). In natural ecosystems with N limitation such changes may affect herbivory and decomposition and the energy and nutrient balances and flows. Similarly the effect on domestic animal nutrition from pastures may well be of importance (e.g. Seligman & Sinclair 1995).

Temperature often has large effects on quality of products. Smaller, more fibrous leaves are often a result of warmer temperatures (especially if water stress occurs), and there may be changes in nutritional quality, for example decreases in nitrogen and increasing tannin and phenols (e.g. Dury et al. 1998; Lindroth 1996). In cereals smaller grain with a larger proportion of husk is commonly observed at warmer temperature (e.g. Tester et al. 1995). Small grain may have larger N concentration as carbohydrate decreases due to the shorter grain-filling period. Sanhewe et al. (1996) found that the seed viability and vigour in wheat increased with mean temperature over the grain-filling period and no detrimental effect of high T was observed. Potential longevity of seed increased with increasing T but increased [CO2] had no effect. The ability of seed to germinate and desiccation tolerance were the same in grain from a range of T and normal compared with increased [CO2] (Sanhewe et al. 1996). Indeed, as mentioned earlier, the lack of effect of [CO2] on mass per grain in wheat suggests no great change in this aspect of quality.

Specific changes in biochemical composition which are important in quality have been shown for a range of crops. Generally there is a decrease in the concentration of proteins, indicating lower nutritional value (e.g. in rice, Ziska et al. 1997). Wheat grown with increased [CO2] also has decreased protein (Rogers et al. 1998), amino acid content (Manderscheid et al. 1995) and N percentage (Mitchell et al. 1993) but no changes in proportions of grain storage proteins (Shewry et al. 1994). In wheat, small increases in temperature (2–4 °C) had a larger effect than 2 ×[CO2] on grain quality (Williams et al. 1995; Tester et al. 1995). Starch content, starch grain size and numbers, and gelatinization were altered in complex ways with temperature, but there was little effect of increased [CO2]. The lipid and starch lipid concentrations also changed qualitatively and quantitatively (Williams et al. 1995; Tester et al. 1995). Decreased protein concentration could affect human nutrition and altered starch characteristics could also affect the acceptability of the product. For example, the ‘stickiness’ of rice increased due to increased amylose after growth in warmer temperatures (Ziska et al. 1997).

The quality of other crops is also likely to alter. An example is sugar beet (Demmers-Derks et al. 1998) which depends on the concentrations of sucrose, and components which interfere with the extraction of sucrose, such as glycinebetaine and α-amino N. Neither 2 ×[CO2] or a temperature increase of + 4 °C affected sucrose concentration per unit root dry matter but glycinebetaine and amino N did decrease with increased [CO2] in similar proportion. Only amino N was increased by warmer temperature. There was substantial interaction with other factors: amino-N increased by a greater proportion with abundant N supply than with deficient N. However, the roots contained a larger proportion of sucrose but smaller proportion of α-amino N and glycinebetaine when grown with deficient N.

While it seems that temperature differences of only a few degrees have more pronounced effects on quality than increased [CO2], our ability to assess effects of T × CO2 interactions on quality will be severely limited until our understanding of the biochemical mechanisms underlying the interactions shown in Fig. 2 is improved. In addition, it is evident that GEC conditions could affect the quality and economics of crop production, but it should be remembered that many aspects of both are subject to relatively rapid social or cultural (food preferences) and technical (crop varieties, processing) changes.


Many papers have considered [CO2] effects alone on wild species, particularly focusing on inter- and intraspecific competition (see review by Patterson & Flint 1990). However, little has been done on the effect of [CO2] and its interaction with temperature on processes in natural plant populations compared to the volume of work on crop species. Of course, the T × CO2 interaction and effects on source and sink as outlined in Fig. 2 will apply in wild species just as in cultivated ones (e.g. Coleman & Bazzaz 1992). In one detailed study (Morse & Bazzaz 1994) the response of artificial populations of the two annual species A. theophrasti (C3) and A. retroflexus (C4) to enrichment was examined at different temperatures. At the higher temperatures germination and initial growth were faster, creating more competition. In the C3 species CO2 enrichment increased competition and mean size as expected, implying increased fecundity. In both species mortality and size inequalities between individuals were increased by both higher temperature and enrichment, which Morse & Bazzaz (1994) attributed to faster canopy closure, and earlier, stronger competition for light. In addition, increased [CO2] caused flowering in A. theophrasti at smaller plant sizes. From earlier studies (e.g. with P. drummondii,Garbutt & Bazzaz 1984) there is also evidence of differences in response to [CO2] for different populations. As plant populations from different climatic zones differ in responses to temperature there are highly likely to be T × CO2 interactions in growth and reproduction, and possibly changes in the relationships between vegetative and reproductive phases and allocation of growth in short-lived species (see Farnsworth & Bazzaz 1995).


The responses of plant growth and development to changes in temperature and CO2 described above are very complex but as generalizations we suggest that in comparison with current [CO2] and temperatures:

1 increased [CO2] stimulates growth in biomass but in almost no system studied is the effect unlimited;

2 in many cases in increased [CO2] there is acclimation of A but over the long-term rates are still somewhat higher than in the present atmospheric [CO2];

3 accumulation of carbohydrates is greater in increased [CO2];

4 increased [CO2] increases the number of organs and size (the relative effects being highly specific to species) but in most cases does not affect development;

5 warmer temperatures accelerate the rate of organ development and expansion but decrease the duration so total biomass production is reduced, especially of annual, determinate crops;

6 increased [CO2] partially compensates for the effect of T on biomass but does not have a proportionately larger effect at higher than at normal T (no T × CO2 interaction);

7 warmer T may decrease the accumulation of carbohydrates but not consistently; and

8 at very cool temperatures the stimulation of growth by increased [CO2] may be decreased and may cease at low temperatures, though the effect will depend on plant acclimation and adaptation.

A semi-mechanistic explanation of these effects is suggested and summarized in Fig. 2. Consider the processes determining biomass production as a source of assimilates (P), which supplies the sink processes, which are growth (G), storage (S) and respiration (R) so P = G + S + R, all with units of mol C plant–1 time–1. The rate of assimilate supply is determined, ultimately, by the available light, the total leaf area and the efficiency of the photosynthetic system per unit area of leaf in converting light into CH2O as a function of temperature. The total respiration (R) is determined by the mass of respiring material, the temperature response and the availability of CH2O. There may be complex acclimation of both photosynthesis and respiration to [CO2] and temperature. The conceptual model may accommodate such changes as a relative decrease in these rates with time and conditions. Storage (S) is a component with an upper limit determined by the size of the storage compartments and the efficiency of the mechanism for accumulating stored materials. Growth (G) is a function of temperature and assimilate supply (see earlier discussion). However, and most importantly, it is ultimately determined and regulated by the developmental processes, such as the switch from vegetative to reproductive development and the initiation of meristems, and the onset and rate of senescence.

With increased [CO2] the rate of production of CH2O is higher than in normal [CO2] due to reduction in photorespiration and faster carboxylation. This increased rate of CH2O supply stimulates the expansion of meristems so larger G results. The relative growth rate may be higher than in normal [CO2] during early growth (den Hertog, Stulen & Lambers 1993). However, this process is not persistent, otherwise there would be a continuously increasing difference between production of dry matter between plants grown in increased and normal [CO2]. In particular, for determinate plants the ontogenetic developmental processes limit meristem production, curtailing further growth. In addition plant competition for light restricts growth. Thus increased [CO2] produces a larger plant, with more and/or larger organs but, after the initial growth stimulation, with a similar RGR to that of plants in normal [CO2]. Other environmental factors, such as irradiance and nutrient supply regulate the formation of meristems and organs and their growth so affecting the basic responses. Generally, even if the RGR of high-[CO2] grown plants returns to that of the plants in normal [CO2], their biomass is larger.

With this larger biomass in increased [CO2], and if R is not affected by [CO2], the total R of plants is higher. With P still in excess of G + R then S will be large, both because the size of the compartment is increased and in some cases, but not all (sugar beet for example), the concentration of S per unit cell volume is increased, but ultimately S may reach the intrinsic capacity of the storage compartments. As a consequence, the feed-back mechanisms of regulation are engaged, with carbohydrate down-regulation of P (e.g. decreased activity of Rubisco) and/or acclimation (e.g. reduction of Rubisco protein amount, Rowland-Bamford et al. 1991). This may have the consequence of releasing some N, etc. for additional growth in nutrient-limited conditions. This is probably a small component of the nutrient budget but, depending on the nutrient supply, it may be of consequence. Possibly there are a number of different mechanisms which are species-dependent or affected by other conditions (e.g. the availability of N) so that similar overall response, in terms of balancing processes, may be achieved by different routes. The effect is to decrease P and bring it into a balance with G.

The value of R may be considered fixed, depending on temperature, so that in warmer conditions, when a faster rate of supply of ATP, NAD(P)H and metabolic intermediates to G is required, the plant uses more CH2O than in cooler conditions. However, evidence suggests a more dynamic system. If P is high then possibly R is increased due to the increased supply of CH2O thus using more P and serving to avoid down-regulation or acclimation of P. However, the extensive feed-back regulation on R, for example, via ATP concentration, which is needed for cellular homeostasis, may not allow this option. In this case, consumption of CH2O by uncoupled respiration (alternative oxidase) would serve as a ‘safety valve’ to regulate CH2O concentration and allow P to continue at a substantial rate (Amthor 1995). If decrease in R occurs with increased [CO2], then it will increase S, and affect the balance of other processes, exacerbating acclimation. Although little is known of the acclimatization responses of respiration to increased [CO2] and warmer temperatures such changes could result in adjustment of source and sink. However, the role of R in determining responses to increased [CO2] is unclear and deserves further analysis (see Drake et al. 1999, this issue).

Temperature affects many processes in the scheme outlined, and these concepts may explain some experimental results. A simple view of the effects of warmer temperature on biomass production is that stimulation of growth of existing meristems/organs causes faster development but the duration of phases is shortened, decreasing the time for supply of assimilated to the growing organs, resulting in fewer, smaller organs and less biomass accumulation. Accumulation of CH2O may indicate the balance between P and the sink processes. Increased [CO2] would increase P, allowing higher G and R and maintaining or increasing [CH2O]. However, the developmental processes regulate G ultimately via production of new meristems and organs.

The simple scheme presented also includes the likely effects of nutrient supply (considered as nitrogen N but similar concepts apply to other nutrients). If N is abundant then the processes of protein synthesis, cell formation and expansion will occur with an upper limit determined by the developmental processes. The response to increased [CO2] will then be positive and large. If N is severely limiting, then the absolute response to increased [CO2] would be small (and may be negative). Such limitations caused by nutrient deficiency may also interact with T and [CO2] and must be considered when seeking to understand and explain their effects.


Response of plant development, growth and biomass accumulation to increased [CO2] at different temperatures is not as expected from an analysis of the photosynthetic processes alone. In adapted and acclimated vegetation there is little firm evidence of T × CO2 interactions across the differences in temperature regime that characterize different biomes. However, ecosystems at low T and with deficient nutrients, where growth is severely limited, may show little response to increased [CO2]. Interspecific differences in response, due to different growth forms, rates of development and expansion of meristems and organs, and patterns of assimilate allocation to organs obscure simple generalizations about temperature and CO2 responsiveness. This may also be the case with different genotypes and cultivars, although little information is available about quantitative differences in mechanisms which may underlie such responses. These effects will determine ecosystem responses. Investigations of T × CO2 interactions in individual species or genotypes show little consistent pattern due largely to the many different interactions of temperature and assimilate supply (for example on carbon balance, meristem initiation and expansion rates, senescence, fertilization, and carbohydrate storage and mobilization) at different stages in the growth cycle or in different weather and edaphic conditions. Extreme conditions of temperature and nutrition for example, may affect processes independently of [CO2] and thus prevent or modify the responses.

We finish by emphasizing that examination of CO2 response in different temperature regimes is clearly important both because of past and future GEC, and the different temperature regimes around the globe. However, other aspects of the environment, in particular light, water and nutrient supply, are obviously critical in the assessment and interpretation of effects of increased [CO2] so the impacts of T × CO2 interactions should not be considered in isolation. Indeed, many of these interactions may be already included in the experiments reported, but in few are these effects quantified. A conceptual model of the interactions of different processes in the biochemistry and physiology of plants and the impacts of environmental conditions on them is advanced to aid understanding of plant responses to GEC. Progress in understanding responses will require quantification of these processes under a range of GEC conditions.


The preparation of this review was assisted by a research contract (ModexCrop) to J.I.L.M. from the EU Environment programme, Contract No. ENV4-CT95–0142. We thank Kevin Ansell, University of Essex, for suggestions about literature sources, Kirsten Lawlor for production of Fig. 2, and Rowan Mitchell, IACR-Rothamsted for discussions about plant responses to T × CO2 interactions. IACR-Rothamsted is grant supported by the Biotechnology and Biological Research Council.