Introduction

The C4 photosynthetic pathway has independently evolved more than 60 times, making it one of the most convergent of complex traits in the biosphere (Christin et al. 2011; Sage et al. 2011a). Phylogenetic analyses of known C4 clades demonstrate that the C4 pathway is relatively young in geological time, having repeatedly arisen in the past 30 million years (Christin et al. 2008, 2011; Vincentini et al. 2008). A leading hypothesis for the rise of C4 photosynthesis is the carbon starvation hypothesis, which proposes that, in warm environments, low atmospheric CO2 conditions promoted high rates of photorespiration and reduced carboxylation capacity of Rubisco in C3 plants; in response, C4 photosynthesis evolved to compensate for the inhibitory effects of photorespiration (Ehleringer et al. 1991, 1997; Monson 2003; Sage 2004). Over the past 200 million years, atmospheric CO2 levels were well above current levels until the late-Oligocene epoch (30–25 million years ago, MYA). In the late Oligocene, atmospheric CO2 declined to between 200 and 400 μmol CO2 mol−1 air (Pagani et al. 1999; Pearson and Palmer 2000; Royer et al. 2001). During the last 0.8 MYA, they oscillated between 180 and 300 μmol mol−1 (Petit et al. 1999; Gerhart and Ward 2010). Consistent with the CO2 starvation hypothesis, the decline in CO2 in the late-Oligocene corresponds to the first wave of C4 origins, while a possible late-Miocene CO2 decline to below 300 μmol mol−1 (Kurschner et al. 2008; Tripati et al. 2009) corresponds to an additional wave of C4 evolution and the expansion of C4 grasslands in low- to mid-latitudes (Christin et al. 2008, 2011; Vincentini et al. 2008; Arakaki et al. 2011). Climates were as warm or warmer during the Miocene and Pliocene than today (Zachos et al. 2008), such that leaf temperatures >35°C would have regularly occurred at low latitudes, as indicated by measurements in modern ecosystems (Monson and Jaeger 1991; Vogan 2010). The synchrony between low CO2 in the Miocene and C4 origins is also consistent with ecophysiological studies that show superior growth, carbon gain, quantum yield, and resource-use efficiency in C4 relative to C3 plants at low to current CO2 levels and temperatures over 30°C (Ehleringer and Pearcy 1983; Dippery et al. 1995; Tissue et al. 1995; Sage 1995; Ehleringer et al. 1997). For example, C4 biomass dominates C3 biomass in prairie sod grown at 200 μmol mol−1; at 360 μmol mol−1, the C4 to C3 biomass ratio is similar (Johnson et al. 1993; Polley et al. 2002).

While the timing of C4 origins and the ecophysiological data are consistent with the CO2 starvation hypothesis, these results do not address how low CO2 would have actually led to the evolution of the C4 pathway. For the CO2 starvation hypothesis to be valid, the intervening stages between the C3 and C4 conditions would have to enhance fitness in a low CO2, high temperature environment. The intermediate stage during C4 evolution involves the formation of a photorespiratory CO2 concentration mechanism in plants known as C3–C4 intermediates (Monson and Rawsthorne 2000). In C3–C4 intermediates, the photorespiratory cycle is distributed between two tissue compartments, the bundle sheath and the mesophyll, rather than occurring in a single photosynthetic cell as in C3 plants. This two-tissue metabolic cycle restricts photorespiratory CO2 production to the bundle-sheath cells, thus facilitating its recapture by bundle-sheath chloroplasts, and boosting the efficiency of bundle-sheath Rubisco (von Caemmerer 1989). In combination, the changes leading to C3–C4 intermediacy compensate for high rates of photorespiration as evidenced by reduced CO2 compensation points of photosynthesis (Γ) and elevated rates of net CO2 assimilation in C3–C4 relative to C3 species at low atmospheric CO2 concentration (Morgan and Brown 1979; Ku et al. 1983, 1991; Rajendrudu et al. 1986; Monson and Rawsthorne 2000; Vogan et al. 2007).

To best evaluate the CO2 starvation hypothesis, closely related C3, C3–C4 intermediate, and C4 species should be grown and measured in the hot, low CO2 conditions proposed to favor the evolution of the C4 pathway. Because plants can acclimate to variation in growth CO2 and temperature, inferences based on plants grown away from the conditions where C4 photosynthesis evolved could be misleading. The large majority of C4 lineages have arisen in warm, sub-tropical to tropical locations, and most C3–C4 species grow in hot environments (Monson and Rawsthorne 2000; Edwards and Smith 2010; Sage et al. 2011a, b), hence, it is best to evaluate their performance at the elevated temperatures common in their habitat. Also, multiple evolutionary lineages should be simultaneously studied. To address evolutionary hypotheses where a genetic model is lacking, comparisons of numerous independent lineages are preferred, as each lineage is but one replicated observation (Felsenstein 1985). Previous studies addressing the carbon starvation hypothesis rarely met these criteria. No study has compared responses of multiple evolutionary lines of C3, C3–C4, and C4 species grown and measured at subambient CO2. C3–C4 intermediates have elevated photosynthetic capacities relative to C3 species at sub-ambient CO2; however, this observation is based on measurements of plants grown at the atmospheric CO2 levels at the time of the experiment (330–375 μmol mol−1; Ku et al. 1983, 1991; Rajendrudu et al. 1986; Vogan et al. 2007). High temperatures aggravate photorespiration in C3 plants and thus the selection pressures for C4 evolution would be greatest in hot environments (Sage 2004), but few studies have examined performance of C3–C4 intermediates relative to C3 species growing in hot environments. Schuster and Monson (1990) and Monson and Jaeger (1991) did evaluate C3–C4 Flaveria species grown in warm to hot conditions, but the plants grew at atmospheric CO2 levels of the recent past (350–360 μmol mol−1).

One potential outcome of growth in low CO2 and elevated temperature is leaf-level biochemical acclimation which might enhance carbon acquisition in C3 species and reduce the relative benefit of the C3–C4 mechanism. In C3 species, this can occur by increased investment in Rubisco content (Sage and Coleman 2001). C3–C4 species may also acclimate and thus retain their advantage; for example, they may enhance the efficiency of photorespiratory CO2 re-fixation. A third acclimation response, increased stomatal conductance leading to higher intercellular CO2 levels, has been observed in low CO2-grown plants of the C3 forb Solanum dimidiatum (Maherali et al. 2002). Increased stomatal conductance, however, would greatly enhance water loss in hot environments and could thus be of limited value. In hot environments with high vapor pressure differences between leaf and air, it is possible that C3–C4 intermediate plants have reduced stomatal conductance relative to C3 species, since they may rely on the photorespiratory CO2 pump to offset reduced intercellular CO2 levels. This could lead to greater water-use efficiency in C3–C4 relative to C3 species growing in warm, low CO2 conditions.

In this study, we evaluated photosynthetic and stomatal acclimation to low CO2 in closely related C3, C4 and C3–C4 plants from three separate evolutionary lineages of C4 photosynthesis: Flaveria (Asteraceae), Heliotropium (Boraginaceae), and Alternanthera (Amaranthaceae). Each species studied occurs in hot environments where daytime temperatures during the growing season exceed 35°C (Supplemental Fig. S1; Monson and Jaeger 1991). The C3–C4 species Heliotropium convolvulaceum grows on sand dunes in the Mojave and Chihuahuan deserts of southwestern North America. Episodic monsoon rains support summer activity of H. convolvulaceum, when leaf temperatures exceed 35°C for much of the day (Fig. S1). Heliotropium calcicola (C3), Flaveria robusta (C3), and F. ramosissima (C3–C4) are common in semi-arid regions of central Mexico where they also exploit monsoon rains to remain active on hot summer days (Frohlich 1978; Powell 1978). Alternanthera sessilis (C3) and A. tenella (C3–C4) are herbaceous species native to moist (sub)tropical climates of the Caribbean basin, Central America, and South America (http://www.tropicos.org). Plants were grown at high temperatures (37/29°C) in low and current atmospheric CO2 concentrations of 180 and 380 μmol mol−1. The response of net CO2 assimilation rate (A) to intercellular CO2 concentration (C i ) was measured at 30 and 40°C and the response of A to variation in leaf temperature (T) was assessed at the growth CO2 concentrations. Stomatal acclimation to low CO2 was evaluated using responses of stomatal conductance (g) to C i (Santrucek and Sage 1996). Associated with these measurements, we measured leaf nitrogen (N), Rubisco, and chlorophyll contents to determine if N allocation to photosynthetic processes is altered by growth CO2.

Materials and methods

Source material and growth conditions

Flaveria bidentis (C4) seeds were obtained from offspring of seeds originally collected by Powell (1978). E. Sudderth provided collected seeds of F. ramosissima (C3–C4) from Tehuacan, Mexico (Sudderth et al. 2009). R.F. Sage collected seeds of F. robusta (C3; 20 km W of Colima, Mexico), Heliotropium convolvulaceum (C3–C4; 10 km NE of Overton, NV, USA), and H. texanum (C4; 6 km E of Zapata, TX, USA; Vogan et al. 2007). Heliotropium calcicola (C3) seeds were collected by Michael Frohlich in northern Mexico (Vogan et al. 2007). Seeds of Alternanthera sessilis (C3) and A. tenella (C3–C4) were from offspring of plants collected by A.S. Ragavendra (Hyderabad, India), while A. caracasana (C4) seeds were collected by A.M. Powell in Alpine, TX, USA. Within their respective genera, these species are closely related (Frohlich 1978; McKown et al. 2005; Sanchez-del Pino et al. 2009). Vouchers specimens are stored at the Royal Ontario Museum, Toronto, ON.

Plants were grown from rooted cuttings in growth chambers (GC-20; BioChambers, Winnipeg, MB, Canada) set to deliver day/night temperatures of 37/29°C, a 14-h photoperiod, and a mean irradiance at plant height (±SE) of 561 ± 11 μmol photons m−2 s−1. Plants grew in 4-L pots containing equal parts sand, Pro-Mix (Premier Horticulture, Quakertown, PA, USA) and sterilized topsoil. Plants were watered daily, fertilized three times weekly with a full-strength Hoagland’s solution, and trimmed monthly. Growth CO2 levels (±SE) were 184 ± 5 and 381 ± 8 μmol mol−1, and were maintained with a soda-lime scrubber controlled by a PP Systems WMA-2 gas analyzer (PP Systems International, Amesbury, MA, USA). To minimize growth chamber effects, plants were rotated weekly within a chamber, and the treatments were switched between chambers for the three replicated sets of measurements.

Photosynthetic gas exchange and leaf nitrogen analysis

Gas exchange of a youngest fully expanded leaf was measured using an LI-6400 photosynthesis system (Li-Cor, Lincoln, NE, USA) at a saturating light intensity of 1,500 μmol photons m−2 s−1. The response of A and g to leaf temperature (T) was measured at 380 and 180 μmol CO2 mol−1 air by first lowering leaf T from 25 to 15°C and then raising it in 5°C increments to 45°C. Photosynthetic responses to intercellular CO2 (A/C i responses) were logged at 30 and 40°C during the course of an A/T response measurement. Leaf-to-air vapor pressure difference during the A/T responses rose from 1.5 kPa at low temperature to 3.5 kPa at the warmer temperatures, but was maintained at 2–2.7 kPa for the A/C i measurements. The photosynthetic thermal optimum was calculated by determining maximum A from a cubic regression of the A/T responses. CO2 compensation points (Γ) of A were estimated by extrapolating the initial slope of the A/C i plots to the x-intercept.

After each gas exchange response, two 2.6 cm2 leaf discs were sampled from measured leaves. One was assayed for N content using a C:N analyzer (Costech Analytical Technologies, Valencia, CA, USA); the second was assayed for Rubisco content using the 14C-CABP binding method (Sage et al. 1993). Chlorophyll was spectrophotometrically assayed following extraction in acetone (Evans 1983). Nitrogen allocation to electron transport components was calculated from leaf chlorophyll content based on a value of 55 mol N allocated to electron transport for every mol of chlorophyll (Evans 1983). Rubisco N was estimated assuming a protein N content of 16% (Sage et al. 1987).

Results

Growth CO2 concentration did not affect leaf N, relative N allocation to Rubisco and electron transport, leaf C:N ratios, or specific leaf mass (Supplemental Table S1). Leaf Rubisco content was 60% lower and C:N was roughly 25% higher in C4 species than in the other functional types, but there was no apparent biochemical acclimation to low CO2 in any species or photosynthetic type (Table S1).

Growth CO2 had no effect on the A/C i responses of a given species. The CO2 compensation point and net CO2 assimilation rate at the highest measurement CO2 level (A 800) were the same between plants grown at 180 versus 380 μmol mol−1 (Figs. 1, 2 and 3; Table 1). Increasing the measurement temperature from 30 to 40°C increased Γ in C3 species by an average of 25 μmol mol−1, while in C3–C4 intermediates, the increase was 16 μmol mol−1. Flaveria ramosissima, a C3–C4 species with some C4-cycle activity, exhibited lower Γ than Alternanthera tenella and Heliotropium convolvulaceum which lack significant C4-cycle activity (Table 1; Monson et al. 1986; Devi et al. 1995; Vogan et al. 2007). These species’ gas exchange results did not otherwise differ. Net CO2 assimilation rates at the respective growth CO2 levels were lower in both C3 and C3–C4 species at 40°C compared to 30°C; the decline in A with increasing temperature was approximately 30% greater in C3 than in C3–C4 species (Table 1). At 380 μmol mol−1 and 40°C, A was slightly higher in the C3–C4 species than C3 plants, while at 180 μmol mol−1, A was 67% greater in the C3–C4 species. C4 species exhibited the highest rates of A at all measurement temperatures and CO2 levels.

Fig. 1
figure 1

The response of net CO2 assimilation rate to intercellular CO2 concentration for current ambient CO2-grown (380 μmol CO2 mol−1; a, b low CO2-grown (180 μmol mol−1; c, d Alternanthera species. Measurements were conducted at a photon flux density of 1,500 μmol m−2 s−1, leaf-to-air vapor pressure difference of 2–2.7 kPa and leaf temperatures of 30°C (a, c) and 40°C (b, d). Each symbol is the mean ± SE of three measurements from independent plants

Full size image
Fig. 2
figure 2

The response of net CO2 assimilation rate to intercellular CO2 concentration for current ambient CO2-grown (380 μmol CO2 mol−1; (a, b) low CO2-grown (180 μmol mol−1; c, d) Flaveria species measured at 30°C a and c and 40°C (b, d). Measurement conditions were as described for Fig. 1. Each symbol is the mean ± SE of three measurements from independent plants

Full size image
Fig. 3
figure 3

The response of net CO2 assimilation rate to intercellular CO2 concentration for current ambient CO2-grown (380 μmol CO2 mol−1; a, b) low CO2-grown (180 μmol mol−1) Heliotropium species measured at 30°C (a, c) and 40°C (b, d). Measurement conditions were as described for Fig. 1. Each symbol is the mean ± SE of three measurements from independent plants

Full size image
Table 1 CO2 compensation points (Γ) and net CO2 assimilation rates at growth CO2 concentrations and near CO2 saturation for Flaveria, Alternanthera, and Heliotropium
Full size table

The thermal optimum of A (T OPT) at 380 μmol mol−1 ranged from 30 to 32°C for the C3 species, from 31 to 32°C for the C3–C4 intermediates and from 34 to 42°C for the C4 species (Table 2). When the measurement CO2 concentration was lowered to 180 μmol mol−1, T OPT of the C4 species was unaffected, while it decreased to 29°C in the C3–C4 species and 25°C in the C3 species (Table 2; Fig. 4; Supplemental Table S2). At low measurement CO2, A at T OPT was highest in the C4 species of each genus, and was 34% greater in the C3–C4 intermediates than the C3 species. At the daytime growth temperature (37°C), A in the C3–C4 species was not different from C3 species at the current ambient CO2, but at low CO2, A in C3–C4 species was 1.6–2 times higher.

Table 2 The photosynthetic thermal optimum (T OPT) of net CO2 assimilation rate (A) and net CO2 assimilation rate at T OPT and at 37°C for C3, C4 and C3–C4 intermediate photosynthetic types
Full size table
Fig. 4
figure 4

The response of net CO2 assimilation rate to leaf temperature for C3, C3–C4 and C4 species of Alternanthera, Flaveria, and Heliotropium. Plants were measured near their growth CO2 concentration: 180 μmol mol−1 (a, c, e) and 380 μmol mol−1 (b, d, f). Results from Alternanthera species are presented in (a, b), Flaveria species in (c, d), and Heliotropium species (e, f)

Full size image

The g versus C i responses were not affected by growth CO2 in any species, and were similar in all three photosynthetic types of Alternanthera and Flaveria (Figs. 5, 6 and 7). The C i/C a ratio increased with reduction in C a below 300 μmol mol−1 in both the C3 and C3–C4 species. There was no consistent evidence for stomatal acclimation within any species or photosynthetic type as the C i/C a versus C i responses were similar between CO2 treatments. At 180 μmol mol−1, the larger A values in the C3–C4 intermediates relative to the respective C3 species of Flaveria and Heliotropium resulted in significantly lower C i/C a in the C3–C4 species than in C3 plants; this difference was not apparent at current ambient CO2 levels. Given the similarity in g between the two photosynthetic types at low CO2, the difference in C i /C a between C3 and C3–C4 species at sub-ambient CO2 was driven entirely by higher A in the C3–C4 species. Stomatal conductance also increased at low CO2 in all three C4 species, but to a lesser extent than in the C3 and C3–C4 species, such that the C i/C a values declined or showed little change at the lowest measurement C i.

Fig. 5
figure 5

The responses of stomatal conductance (open symbols) and C i /C a ratio (filled symbols) to variation in intercellular CO2 concentration at 30°C (a, c) or 40°C (b, d) for three Alternanthera species grown near current CO2 conditions (380 μmol mol−1 (a, b) or low CO2 conditions (180 μmol mol−1 (c, d). Measurement conditions were the same as indicated in Fig. 1. Each symbol represents the mean ± SE of three measurements from separate plants

Full size image
Fig. 6
figure 6

The responses of stomatal conductance (open symbols) and C i /C a ratio (filled symbols) to variation in intercellular CO2 concentration at 30°C (a, c) or 40°C (b, d) for three Flaveria species grown near current CO2 conditions (380 μmol mol−1 (a, b) or low CO2 conditions (180 μmol mol−1 (c, d). Measurement conditions were the same as indicated in Fig. 1. Each symbol represents the mean ± SE of three measurements from separate plants

Full size image
Fig. 7
figure 7

The responses of stomatal conductance (open symbols) and C i /C a ratio (filled symbols) to variation in intercellular CO2 concentration at 30°C a, c) or 40°C (b, d) for three Heliotropium species grown near current CO2 conditions (380 μmol mol−1 (a, b) or low CO2 conditions (180 μmol mol−1 (c, d). Measurement conditions were the same as indicated in Fig. 1. Each symbol represents the mean ± SE of three measurements from separate plants

Full size image

Photosynthetic water-use efficiencies (PWUE) in C3 and C3–C4 species of each genus at 40°C were similar at 380 μmol mol−1. There was a trend for PWUE to be 85 and 186% greater in the intermediates F. ramosissima and H. convolvulaceum, respectively, than in the respective C3 species F. robusta and H. calcicola at 180 μmol mol−1 and 40°C (P = 0.07 and 0.08, respectively; Table 3). In Alternanthera, there was no significant difference in PWUE between the C3–C4 species and the C3 species. C4 species exhibited significantly greater PNUE than C3 and C3–C4 species at low and current CO2 concentrations. There were no significant PNUE differences between C3 and C3–C4 species at the current atmospheric CO2 level. At low growth CO2, C3–C4 PNUE was 63% greater than C3 PNUE. Because leaf N, chlorophyll and Rubisco N content were not different between these two functional types, the difference in PNUE resulted from greater A in the C3–C4 than C3 species at 180 μmol mol−1.

Table 3 Photosynthetic water-use and nitrogen-use efficiencies (PWUE and PNUE, respectively) of Flaveria, Alternanthera and Heliotropium species at cuvette CO2 concentrations ([CO2]) of 180 and 380 μmol mol−1
Full size table

Discussion

Acclimation responses to low CO2

Hypothesized acclimation responses to low CO2 in C3 plants include increased Rubisco content and reduced investment in electron transport capacity (Sage and Reid 1992; Sage and Coleman 2001). Such responses have been observed sporadically in C3 plants; for example, Abutilon theophrasti and Solanum dimidiatum increase Rubisco content relative to electron transport capacity in low growth CO2 (Tissue et al. 1995; Anderson et al. 2001). None of the species in this study, however, showed significant photosynthetic acclimation to low CO2. There was no increase in Rubisco content or apparent change in electron transport investment in any species grown at 180 μmol mol−1 CO2 relative to 380 μmol mol−1. We hypothesized that, if changes in leaf N allocation did occur, they might be attenuated in C3–C4 relative to C3 species. This hypothesis was not supported as there was no significant change in N allocation in any species of any functional type when grown at low CO2. This lack of plasticity could have constrained the ability of C3 plants to adjust to periods of low atmospheric CO2 in order to ameliorate carbon starvation. As a result, evolutionary selection for CO2 scavenging mechanisms may have been enhanced. For example, re-fixation of photorespired CO2 within the bundle-sheath may have improved carbon balance of some C3 species sufficiently to initiate the evolution of C3–C4 intermediacy and, subsequently, C4 photosynthesis (Monson and Rawsthorne 2000; Muhaidat et al. 2011).

Our results also demonstrate a low ability of stomata to compensate for CO2 deficiency. The responses of g versus C i and C i /C a versus C i were measured as an index of stomatal acclimation to low CO2. A common acclimation response to high growth CO2 is a reduction in g at high measurement CO2 and reduced sensitivity of stomata to variation in C i (Santrucek and Sage 1996; Tognetti et al. 2000; Lodge et al. 2001; Herrick et al. 2004). Stomata can acclimate to low CO2 by increasing g and the sensitivity to CO2 variation, as has been shown for Solanum dimidiatum (Maherali et al. 2002). This increases C i and thus partially compensates for low CO2. Such a response was not observed in any species in this experiment, as the g versus C i responses generally overlapped between the plants grown at low- and present-day levels of CO2. Correspondingly, the response of C i/C a ratio to C i did not differ between plants grown at 180 or 380 μmol mol−1. The responses of C i/C a to intercellular CO2 concentrations were similar between C3 and C3–C4 species above 300 μmol mol−1, but at sub-ambient CO2 concentrations, C i/C a was significantly lower in the C3–C4 species. The absence of differences between the two photosynthetic types in the g versus C i responses indicate that the lower C i/C a values of the C3–C4 species are driven by inherently greater A at low CO2 rather than differences in stomatal acclimation. In C4 species, the greater A at low CO2 enabled by the C4 cycle prevented any rise in C i /C a at low measurement CO2. This difference in the C i /C a response between C4, C3 and C3–C4 plants demonstrates the assimilatory strength of the C4 metabolic cycle in CO2-depleted atmospheres (Huxman and Monson 2003; Vogan and Sage 2011).

Carbon balance of photosynthetic types under low CO2 and high temperature

While the A/C i responses of all nine species were unaffected by growth at low relative to current CO2 levels, there are substantial differences between the A/C i responses of the three photosynthetic types that impact carbon balance. The primary differences were lower Γ and 40% (at 30°C) and 67% (at 40°C) greater A at 180 μmol mol−1 in the C3–C4 species compared to the C3 species. There was also a trend towards greater PWUE in the C3–C4 species F. ramosissima and H. convolvulaceum, and greater PNUE in the C3–C4 intermediates, relative to their respective C3 relatives at low but not current levels of atmospheric CO2. These differences are likely due to the enhanced capacity of the C3–C4 species to recapture photorespired CO2 (Ku et al. 1983, 1991; Rajendrudu et al. 1986; Vogan et al. 2007). The A/T responses of the different photosynthetic types further illustrate the benefits of the C3–C4 pathway to photosynthetic performance at low CO2 and high temperature. While T OPT and A at T OPT are not significantly different between C3 and C3–C4 species at ambient CO2, the reduction of CO2 to 180 μmol mol−1 resulted in a significantly greater decline in T OPT in C3 species (6.5°C) than in C3–C4 species (2.6°C). Greater ability to recapture photorespired CO2 moderates the effects of lowering CO2 on T OPT, A at T OPT, and A at the growth temperature of 37°C in the C3–C4 species.

The enhanced photosynthetic performance of C3–C4 intermediates under conditions of low CO2 and high temperature are important given the contribution of A to growth and fitness. Reductions in A at sub-ambient CO2 resulted in equivalent reductions in biomass yield in Abutilon theophrasti, Avena sativa, Brassica kaber, Nicotiana tabacum, Phaseolus vulgaris, and Prosopis glandulosa (Polley et al. 1992; Johnson et al. 1993; Dippery et al. 1995; Cowling and Sage 1998; Ward et al. 1999; Campbell et al. 2005). The amount of vegetative biomass has a large influence on fruit number and seed set, and reductions in plant growth often result in substantially lower plant fitness (Solbrig 1981; Farris and Lechowicz 1990). Consistently, reproductive output is substantially reduced by low CO2 (Sage 1995; Campbell et al. 2005). Several studies have also evaluated fitness of plants with mutations that reduce photosynthetic rate. For example, a 25–30% reduction in A in single-gene mutants of Amaranthus hybridus lowers seed number, seed size and seedling survival at light levels above 400 μmol photons m−2 s−1 (Arntz et al. 2000a, b). Arabidopsis thaliana mutants with reduced chlorophyll content and 15% lower A have a 15% lower growth rate and 25% lower seed weight than wild-type plants (Janacek et al. 2009).

Reduced fitness in C3 species during the low CO2 conditions of the past 25 million years likely created opportunities for novel mechanisms of carbon scavenging. At the atmospheric CO2 concentrations of the late-Pleistocene (180–280 μmol mol−1), for example, photorespiration in hot climates would have reduced A by over 50% (Ehleringer et al. 1991). While initially inhibitory, high rates of photorespiration create an opportunity to improve Rubisco efficiency and A if the release of photorespired CO2 occurs within an inner compartment where Rubisco is present (Monson and Rawsthorne 2000). In C3–C4 species, photorespired CO2 is produced in the inner region of the bundle sheath cells, resulting in observed 67% enhancements of A relative to C3 species. In hot environments during low CO2 intervals, as would have been present on subtropical sand dunes where H. convolvulaceum grows, the ability to recapture photorespired CO2 using the C3–C4 mechanism could have maintained fitness in habitats too extreme for C3 species. Such extreme settings may have thus enabled natural experiments in CO2 scavenging, such as C3–C4 intermediacy, to proceed without interference from C3 competitors (Sage 2004).

Conclusion

The results of this study show that improved carbon economy is the primary benefit of C3–C4 intermediacy under low CO2 and high temperature. There was no acclimation to low CO2 via shifts in leaf N allocation or stomatal behavior in any photosynthetic type. The inherently greater A at high temperature and low CO2 of the C3–C4 intermediates compared to their C3 relatives stands out as the major difference between these two photosynthetic types. The greater carbon assimilation capacity of C3–C4 intermediates under these conditions and their somewhat greater PWUE and PNUE likely provided advantages in growth and reproduction, and sustained these species in the hot, low latitude environments of recent geological time. In contrast, C3 species appear to be severely limited in their ability to alleviate CO2 starvation through photosynthetic or stomatal acclimation. In the past century, CO2 levels have risen to values not seen for millions of years (Pagani et al. 2005), and, given the similarity in photosynthetic performance between C3 and C3–C4 species at current atmospheric CO2 concentrations, the critical environmental stress selecting for C3–C4 photosynthesis appears to have already disappeared. As a consequence, the potential for C4 evolution in the future may have been eliminated.