Abstract
The role of land plants in establishing our present day atmosphere is analysed. Before the evolution of land
plants, photosynthesis by marine and fresh water organisms was not intensive enough to deplete CO2 from
the atmosphere, the concentration of which was more than the order of magnitude higher than present.
With the appearance of land plants, the exudation of organic acids by roots, following respiratory and
photorespiratory metabolism, led to phosphate weathering from rocks thus increasing aquatic productivity.
Weathering also replaced silicates by carbonates, thus decreasing the atmospheric CO2 concentration. As a
result of both intensive photosynthesis and weathering, CO2 was depleted from the atmosphere down to
low values approaching the compensation point of land plants. During the same time period, the atmospheric
O2 concentration increased to maximum levels about 300 million years ago (Permo-Carboniferous
boundary), establishing an O2/CO2 ratio above 1000. At this point, land plant productivity and weathering
strongly decreased, exerting negative feedback on aquatic productivity.
Increased CO2 concentrations were triggered by asteroid impacts and volcanic activity and in the Mesozoic era
could be related to the gymnosperm flora with lower metabolic and weathering rates. A high O2/CO2 ratio is metabolically linked to the formation of citrate and oxalate, the main factors causing weathering, and to the production of reactive
oxygen species, which triggered mutations and stimulated the evolution of land plants. The development of
angiosperms resulted in a decrease in CO2 concentration during the Cenozoic era, which finally led to the
glacial-interglacial oscillations in the Pleistocene epoch. Photorespiration, the rate of which is directly
related to the O2/CO2 ratio, due to the dual function of Rubisco, may be an important mechanism in
maintaining the limits of O2 and CO2 concentrations by restricting land plant productivity and weathering.
Introduction
The atmosphere on Earth is in an extreme state of disequilibrium in which highly reactive gases, such
as oxygen and methane, exist together at concentrations that are different by many orders of magnitude from the photochemical
steady state (Lenton1998). Living organisms regulate the composition of the Earth’s atmosphere via large biogenic fluxes of gases (Lovelock 1972; Vernadsky [1926] 1998). These fluxes can be modelled by daisyworld
(a model describing a very simple planet that has only two species of life on its surface – white and black daisies) scenarios, which are based on feedback regulation of the global environment (Watson and Lovelock 1983; Adams et al. 2003). The turnover of CO2 and O2 through the biosphere dramatically exceeds the turnover of inorganic geochemical cycles. The annual flux of CO2 through the biosphere is approximately 10% of the atmospheric CO2, or slightly more than 0.1% of the totalcarbon in the biosphere.
Thus the turnover time of atmospheric CO2 is about 10 years, while the turnover of all carbon could be less than 1 ka (thousand years). Each year, 120 Pg (120·1015 g) of carbon is exchanged in each direction between terrestrial ecosystems and the atmosphere; another 90 Pg is exchanged between the ocean and the atmosphere,
while 6.3 Pg is emitted by burning fossil fuels (Scholes and Noble 2001). For atmospheric O2, the
turnover time is 4.5 ka, while the inorganic cycle is approximately 3.2 Ma (million years) (Lenton
1998). Photosynthetic organisms producing O2 and utilizing CO2 have played a key role in regulating the
gaseous content of the atmosphere, from their first appearance and continuing evolution. We analyse
here the idea that the gaseous balance of the modern atmosphere, is adjusted mainly by the activity of land plants.
Gaseous content of the atmosphere during Phanerozoic
The Phanerozoic eon covers roughly 550 Ma back to the time when diverse hard-shelled animals first appeared (Figure 1). It is divided into three
eras – Palaeozoic, Mesozoic, and Cenozoic. The border between Phanerozoic and Precambrian
eons (which includes the Proterozoic and Archaean eras) is clear-cut and corresponds to the
first appearance of abundant metazoan fossils. The borders between Palaeozoic and Mesozoic
(245 Ma) and Mesozoic and Cenozoic (65 Ma) eras are also well-defined and are proven to correspond to major asteroid impacts (Ahrens and Jahren 2000; Benton and Twitchett 2003). The Palaeozoic era is divided into six periods (Cambrian, Ordovician, Silurian, Devonian, Carboniferous and Permian), and the Mesozoic era into three periods (Triassic, Jurassic and Cretaceous). The Cenozoic era is divided into seven epochs (Palaeocene, Eocene, Oligocene, Miocene, Pliocene, Pleistocene and Holocene). The first three are united as the Palaeogene period and the last four
as the Neogene period. The previous classification considered the last two as a separate Quaternary
period. The Holocene epoch started 12 ka ago and corresponds to the development of human civilization after the last ice age.
There has been a suggestion of introducing a separate epoch (Anthropocene) (Crutzen 2002), which started
with the industrial revolution (the exact time designated from the invention of the steam engine in 1784
by James Watt), corresponding to major anthropogenic effects, including the tremendous rise in atmospheric CO2, which became evident only in the last century.
Concentrations of O2 during Phanerozoic
Life originated in an anoxic atmosphere and the first available O2 was produced by photosynthesis.
The oxygenic photosynthesis of cyanobacteria is a very ancient process and probably appeared in the
Archaean era, as early as 3800 Ma (Lenton 2001). The presence of O2 sinks including photochemical
destruction, reduced volcanic and metamorphic gases and continental weathering, prevented the
rise of atmospheric O2, until the sinks became saturated. After all the inorganic reductants had become exhausted, photosynthesis, which global overall rate is reflected in organic carbon burial, was needed for oxygen to accumulate in the atmosphere (Bjerrum and Canfield 2004). The O2 concentration did not rise above 800 ppm until the
middle Proterozoic (2200–2000 Ma), when it increased to 2000 ppm (0.2%) (Rye and Holland 1998). Later in the Proterozoic era, the O2 concentration increased to values of 2–3% and rose again towards the end of Proterozoic
(1000–570 Ma), probably triggering the Cambrian ‘explosion’ (the evolutionary diversification of large metazoans) (von Bloh et al. 2003), when it increased to more than 10%. The apparent diversification corresponding to the Cambrian explosion came from a sudden capacity of metazoans to be calcified and preserved as CaCO3. There could
be a significant diversification accompanying the rise of O2 preceding the border between the Proterozoic and Cambrian, as the molecular clock data indicate (Knoll 1994). Biological colonization of the land surface began in the late Proterozoic, leading to phosphate and silicate weathering from rocks and a decreasing CO2 concentration,
while the O2 concentration increased (Lenton and Watson 2004).
Existing geochemical models indicate an O2 concentration of 15–17% in early Phanerozoic
(Berner and Canfield 1989; Berner et al. 2000; Berner 2003a), however before the emergence of
land plants, the O2 concentration could have been as low as 12% during the period 570–400 Ma
(Lenton 2001). With the appearance of land plants, which evolved 420 Ma and were widespread by 370 Ma, the O2 concentration increased, reaching a peak near 300 Ma (late Carboniferous). Giant dragonflies, charcoal deposits and indications of intensive fires, provide evidence for concentrations of O2 higher than present. Berner et al.
(2000, 2003) have suggested a maximum of 35% O2 in late Carboniferous, while more moderate
estimates taking into account feedback of phosphorus weathering, give an upper limit of around
25% (Lenton 2001). After this period, the concentration of O2 decreased and was relatively stable, falling during Triassic and Jurassic below 20% (with a corresponding disappearance of fires and
charcoal burial) and increasing in Cretaceous (150 Ma) to the present day value or higher (Lenton 2001; Bergman et al. 2004) (Figure 1).
Although it is likely that 35% O2 is too high an estimate, Wildman et al. (2004) have shown that
even at such high O2 concentration there would be no widespread burning of forests following a single
lightning strike, at moisture contents common to living plants. Times of high O2 agree with observations of fire-resistant plant morphology, large insects and high concentrations of fossil charcoal
(Wildman et al. 2004). The latest estimates of the COPSE (Carbon–Oxygen–Phosphorus–SulphurEvolution) model (which couples four geochemical cycles as compared to earlier single cycle models) provide a value for O2 in the Permo-Carboniferous of 30% with an upper limit of 33%. The COPSE model questions the low O2 levels in the Triassic
and the Jurassic, estimating them as nearer 20–22% (Bergman et al. 2004). The authors of the model, however, state that, although from 350 Ma to present day, the charcoal record offers the basis for the O2 estimates, its interpretation relative to absolute O2 concentrations still varies.
Concentrations of CO2 during Phanerozoic
Modern geochemical modelling based on the distribution of 13C and other isotopes, shows that
before the fall in CO2 caused by the appearance of land plants, the concentration of CO2 was
0.4–0.5% at the beginning of the Phanerozoic era in the Ordovician and in the Cambrian. The CO2
concentration dropped down during the Devonian and the Carboniferous (coinciding with the highest
O2 concentration in the late Carboniferous). It was higher during the Mesozoic era, probably reaching
2000 ppm (0.2%) in some periods (up to 1400 ppm in the COPSE model, Bergman et al.
2004). Later in the Cenozoic era, a gradual decrease of CO2 took place and for nearly 50 Ma
the concentration was not much higher than the present preindustrial level of 300 ppm (Royer
et al. 2001a, b). Detailed analyses of the air bubbles in cores of Antarctic ice have indicated that
the CO2 concentration oscillated between 180 and 280 ppm, during the last 420 ka (Figure 2) (Petit
et al. 1999; Cuffey and Vimeux 2001). The data has been obtained for the 740-ka period showing
eight glacial cycles (EPICA community members 2004). These oscillations were of a 100 ka period,
with a slow fall and a rapid increase (less than 10 ka). Ten to fifteen smaller oscillations with the
amplitude of 10–20 ppm were observed within every large oscillation. A similar occurrence of
oscillating CO2 with corresponding glaciations and deglaciations, probably also took place around the
Carboniferous-Permian boundary (Di Michele et al. 2001).
The evaluation of atmospheric CO2 based mainly on d13C values of pedogenic carbonates
provides a good CO2 estimate for the pre-Cenozoic period but does not resolve short-term excursions
of 5–10 Ma. For the Cenozoic, the delta 13C of the organic remains of phytoplankton could be of use,
both for temporal resolution (up to 1–10 ka) and calculated CO2 concentrations (Royer et al.
2001a). At elevated CO2 concentrations as detected in the Mesozoic, this method is not useful, it also
needs corrections for changes in growth rate and O2 concentration. Significant concerns have been
raised about the use of isotope palaeobarometry even at moderate CO2 concentrations (Laws et al.
2002) and the proponents of this method now state that the palaeobarometric data can only be used
for a rough estimate of CO2 concentration (Rau et al. 1996, 1997; Royer et al. 2001a).
In Berner’s GEOCARB models, the latest being GEOCARB III (Berner and Kothavala
2001), the isotopic data are used with corrections for different factors (mostly due to the rates of
weathering, but also to global degassing and mountain uplift), to obtain a reliable model of
CO2 changes during Phanerozoic. Models of CO2 in the atmosphere based on stomatal densities of
fossil leaves (Figure 3) (Retallack 2001), show more precise timing for outbursts of CO2 caused
e.g. by asteroid impacts, which took place at the Permian–Triassic, Triassic–Jurassic and Cretaceous–Palaeogene boundaries, also some smaller events which occurred throughout Phanerozoic. Stomata adapt to local and global changes on all timescales (Hetherington and Woodward 2003). The measurement of fossil stomatal densities provides excellent temporal resolution (less than 100 years), with high precision at low CO2 but
Temperature
The deviations in the temperature of the Earth’s atmosphere can be correlated with the CO2 concentration by the equation (Kothavala et al. 1999): DT ¼ 4 lnðRCO2Þ ð1Þ where RCO2 is the CO2 concentration expressed as a ratio to the preindustrial CO2 level (290 ppm).
The data based on foraminifer shells extracted from impermeable clay-rich sediments, show that
the distribution and replacement of the species of these temperature-sensitive organisms in clay-rich
sediments during late Cretaceous and Eocene epochs is in agreement with the variations in CO2
concentrations (Pearson et al. 2001). Correlations between temperature and CO2 concentration during the last 400 ka have been shown to be very precise (Cuffey and Vimeux 2001). Some deviations in earlier periods have been demonstrated, e.g. in Miocene, where the medium temperatures were 6 C higher than predicted from a CO2 concentration of 250–290 ppm, which could be explained by the presence of other greenhouse gasses, e.g. methane (Zachos et al. 2001).
We have used Equation (1) and the estimates of CO2 from the GEOCARB III model to calculate
the mean surface temperature-deviations during the Phanerozoic eon, whilst taking into consideration the 5% increase in the brightness of the sun from the Proterozoic era to present (Tajika 2003).
From the data on O2, CO2 and temperature, we can conclude that these parameters stabilised following the appearance of land plants, with concentrations in the range of 17–30% for O2, 0.02–0.2% for CO2 and average temperatures 3 C higher (with the deviation of 3–4 C) than present day (Figures 1 and 4).
Land plant photosynthesis and O2/CO2 balance
In the photosynthesis process, CO2 is fixed via the Calvin–Benson cycle, in which ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is the primary CO2 assimilatory enzyme in C3 plants. During evolution, the enzyme somewhat surprisingly, has preserved a capacity to use O2 as a substrate, which in an atmosphere of high O2 and
low CO2, makes CO2 fixation less efficient due to the photorespiratory process, starting with the
oxygenation reaction of Rubisco (see below). Metabolic pathways have evolved to overcome the
loss of CO2 during photorespiration, either via a CO2 concentrating mechanism (CCM) in algae, or
via the C4 pathway of photosynthesis in some advanced higher plants. In the latter, primary CO2
fixation is carried out by another enzyme (phosphoenolpyruvate carboxylase), which has no
oxygenase reaction and CO2 is delivered to the Calvin–Benson cycle at a higher concentration in a
specific compartment. We have discussed below, the biospheric consequences of the development of
photosynthesis on Earth and its role in preserving the O2/CO2 balance.
The dual function of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco)
The Rubisco enzyme has maintained during evolution the affinity for oxygen, although there is evidence of its reduction in some species (Bird et al. 1982). This activity becomes physiologically important in the atmosphere with low CO2 and high O2 causing the photorespiration process. The Rubisco specificity factor (Laing et al. 1974),
which defines the preference of Rubisco for CO2 compared to O2 is: s ¼ VcKo=KcVo
where Ko is the Michaelis constant for O2, Kc is the Michaelis constant for CO2, Vc is the maximal rate with CO2, Vo is the maximal rate with O2.
The value of s is about 100 at 25 C in the angiosperm C3 species, while it is lower in C4 plants,
in conifers and in ferns, and it is significantly lower in some green algae (Jordan and Ogren 1981, 1983).
If we consider the O2/CO2 ratio in the atmosphere, which is about 36 times higher than in solution at
25 C because of higher solubility of CO2 compared to O2, we get a value for the specificity factor of
Rubisco of about 3600 in relation to the atmospheric O2/CO2 ratio. This indicates that the Rubisco enzyme will have a 3600 times higher affinity for CO2 than for O2 in the atmosphere, provided there are no limitation effects of stomatal conductance. In the real situation in plants, we have to consider also the concentration of CO2 near the sites of carboxylation, which can be 50% less than that in the air (Evans and von Caemmerer 1996), or even less for
xeromorphic plants (Di Marco et al. 1990), for estimating the specificity of Rubisco in the surrounding atmosphere.
Only the unusual Rubisco enzymes from certain thermophilic red algae (cyanidiophytes) have
very high carboxylation to oxygenation (Vc/Vo) ratios (Uemura et al. 1997), which can be
explained by a higher affinity for CO2, rather than a lower affinity for oxygen. However the s values of the cyanidiophytes are much less extreme when the in vitro assays for s are conducted at the growth temperatures of the organisms (Uemura et al. 1997). The Rhodophyta species living at moderate temperatures have Rubisco with much
less variation in s values (Uemura et al. 1997), while the highest s value (310 at 90 C) was
reported for the Rubisco of the hyperthermophilic archaeon, Pyrococcus kodakaraensis (Ezaki et al. 1999).
On the other hand, Tortell (2000) has shown that plotting the Rubisco specificity factor for a
range of algae against date of evolution, gives a good correlation (showing the increase of s during
the course of evolution) and the red algae really stand out as having different s values. However,
the question why this high affinity enzyme was not positively selected for, during the evolution of
other groups leading to the appearance of land plants, still remains open. It has been shown that
the Form 1 Rubisco from red algae is expressed abundantly in transgenic higher plant chloroplasts
but is not assembled to form an active enzyme (Whitney et al. 2001). However, it has been possible to express active bacterial Rubisco enzymes in higher plants and the CO2 assimilation parameters of these plants correlated with the kinetic properties of the inserted Rubisco enzyme (Parry et al. 2003; Whitney and Andrews 2003). A higher
specificity factor of Rubisco in some C3 plants (e.g. in sunflower by 30%), leads to a marked increase
in net photosynthesis and biomass accumulation (Kent et al. 1992; Kent and Tomany 1995).
The specificity factor of Rubisco can impose limits on the CO2 and O2 concentrations in the
atmosphere. Without special concentrating mechanisms for CO2, plants can only exist in a range of
specified concentrations of O2 and CO2. Above a certain concentration of O2 and below a certain
concentration of CO2 (compensation points), the oxygenase reaction of Rubisco will dissipate more
carbon than is fixed in the carboxylase reaction (Tolbert et al. 1995). Even small changes in the
concentrations of O2 and CO2 in the atmosphere, can lead to drastic changes in metabolism,
favouring either reductive or oxidative reactions (Tolbert et al. 1995; Cen et al. 2001).
At low CO2 levels, the CO2 concentrating mechanisms (CCM) become important (Badger
and Price 2003). The CCMs evolved as a response to the decrease in CO2 concentration during Paleozoic. The CCMs use energy for the active transport of protons, CO2 or bicarbonate. Some CCMs can work with acid-catalysed bicarbonate
conversion to carbon dioxide in a compartment in which the equilibrium carbon dioxide concentration is greater than that in the bulk phase (Badger and Price 2003). To this compartment, bicarbonate can diffuse from the bulk phase (Walker et al. 1980), where the energisation of the CCM operates on the basis of extracellular acid zones produced by the plasmalemma ATPase.
Difference in photosynthesis of aquatic and land plants
In algae, high rates of photosynthetic carbon assimilation occur even at CO2 concentrations as low as 5 ppm, because of an effective CO2 concentrating mechanism (CCM), based on carbonic anhydrase (Raven 2003). This mechanism is induced by low CO2 concentrations, being preceded by photorespiratory release of glycolate (and to lesser extent glyoxylate, glycine and CO2). The rate of photorespiratory decarboxylation of glycine is low in algal cells, due to the operation of a CCM, thus preventing the loss of essential ammonia (Ramazanov and Cardenas 1992; Igamberdiev and Lea 2002). The data showed that the C3 land plants lose ammonia from the leaves at a rate in the order of 0.1–1 nmol m)2 (leaf area) s)1 and this rate increased with temperature and nitrogen nutrition (Husted et al. 2002). The leaves also exhibit a compensation point for ammonia, which ranges from 0.1 to 20 nmol mol)1 in air (Schjoerring et al. 2000). Due to the very high water solubility of ammonia, its loss by aquatic plants would deplete them in nitrogen very rapidly. So, in contrast to land plants, the role of peroxisomes in photorespiration in algae is not very important, photorespiratory peroxisomes with the same enzyme content as in land plants appeared only in their immediate ancestors, the Charophyceae (Huss and Kranz 1997; Karol et al. 2001).
The CO2 enrichment caused by a CCM decreases the degree of 13C fractionation carried out by Rubisco, while in ambient CO2 concentrations this fractionation is usually comparable with the values characteristic for C3 higher plants (Kaplan and Reinhold 1999). Ambient CO2 is not sufficient to completely repress CCMs in most algae; large discrimination against 13C can be found at very high CO2 and low growth rates. Thus in algae, in contrast to C3 land plants, assimilated carbon is characterized by a decrease in 13C fractionation with decreasing CO2 concentration (Laws et al. 2002). On this basis, Rothman (2001, 2002) estimated the atmospheric CO2 concentration during the Phanerozoic period, which is in a good agreement with other models, except that it does not predict a lower CO2 levels in Carboniferous-Permian suggested by the GEOCARB models. This can be explained by the noninclusion of the effects of O2 concentration on 13C fractionation. An elevation in O2 causes in algae, in a similar manner to land plants, an increase in the rate of 13C fractionation (Berner et al. 2000). In C3 land plants, lowering the CO2 concentration
leads to the opening of stomata, thus decreasing the effect of stomatal conductance on 13C fractionation (Igamberdiev et al. 2004). Photorespiration itself contributes to carbon isotope fractionation (Igamberdiev et al. 2004), and it is likely that the O2/CO2 ratio was the main factor determining the carbon isotope composition of C3 land plants
during the Phanerozoic eon (Strauss and PetersKottig 2003).
The absence, or very low activity of a CCM and active peroxisomal metabolism together with a
high glycine decarboxylase capacity in mitochondria, has made the photorespiratory process very
intensive in land plants. Initially, the marine phytoplankton always had a higher biomass and
higher rates of photosynthesis than land plants. A higher global rate of photosynthesis does not
necessarily mean higher biomass in the case of phytoplankton; today there are similar primary
productivities in the sea and on land (Mu¨llerKarger et al. 2005), yet there is about three orders
of magnitude less primary producer biomass in the ocean than on land due to a much greater turnover
of primary producers in the ocean. New satellite observations have shown that phytoplankton
productivity is even higher than earlier estimates (Behrenfeld et al. 2005).
The role of phytoplankton in O2 release was significant from the time when all inorganic
reductants had been exhausted and organic carbon burial began to take place (Bjerrum and Canfield
2004), while CO2 depletion was probably mostly due to the activity of land plants via weathering
(Berner 1997), although the direct role of photo synthesis in this process remains to be quantified.
When the CO2 concentration in the atmosphere decreased to the minimum corresponding to the
ecological compensation point in land plants, the productivity of land plants diminished. The ecological compensation point is a CO2 concentration (in average 150–180 ppm at 21% O2) below which
plants are unable to complete their lifecycles (Sage and Coleman 2001). During glacial periods, temperature gradients drive oceanic circulation causing a greater supply of oxygen to deep waters (Berner 2003a). This would lead to an increase in atmospheric carbon dioxide via inhibition of the formation of anoxic waters in the bottom layers
and a reduction of organic burial (Berner 2003a).
Respiration and photorespiration
The reduction level of higher plants (reflecting the reduction state of reaction centres in chloroplasts
and pyridine nucleotides in all subcellular compartments, primarily in mitochondria) is strongly affected by the O2/CO2 ratio. At low CO2 and high O2, the overreduction of chloroplast reaction centres occurs, due to the low rate of CO2 fixation in the Calvin cycle. Under these conditions, photorespiratory ammonia assimilation together with
other energy utilizing and/or wasting processes, consume the excess reduction power in the chlo
roplasts (Givan et al. 1988; Heber et al. 1996; Osmond et al. 1997; Igamberdiev et al. 2001; Igamberdiev and Lea 2002). In addition, in the mitochondria, photorespiration and respiration increase the reduction level (moderated by the
operation of non-coupled pathways). This results in the incomplete oxidation of glycolytic and photorespiratory substrates leading correspondingly to the efflux of citrate from mitochondria (Igamberdiev and Gardestro¨m 2003) and to the suppression of glycine oxidation resulting in oxalate formation in peroxisomes (Igamberdiev and Lea 2002), both compounds being further excreted into the soil (Figure 5).
Photorespiration is essential even at much higher atmospheric CO2 concentrations than at
present. High CO2 concentrations result in an elevation of temperature, which increases photorespiration and dark respiration rates more than the rate of assimilatory carboxylation (Brooks and Farquhar 1985). However, the development of large leaves may have prevented overheating due to higher transpiration rates (Beerling et al. 2001).
Based on the temperature dependence of the oxygenase to carboxylase ratio of Rubisco (Brooks and Farquhar 1985; Bernacchi et al. 2001) and the estimated deviations of surface temperature from the mean value, we can calculate the oxygenase to carboxylase ratio during Phanerozoic according to Berner’s estimate of CO2 and Lenton’s estimate of
O2 in the atmosphere. The ratio of photorespiratory rate to photosynthetic assimilation rate is half of the oxygenase to carboxylase ratio (Sharkey 1988). Palaeozoic photorespiratory rates could be slightly higher as the carboxylase function of Rubisco was lower by 10–20% (Bird et al. 1982), but these rates could also respond to a lower
brightness of the sun (5%), which would result in lower temperatures (Tajika 2003).
Land plant photosynthesis and photorespiration in the regulation of O2 and CO2
Tolbert et al. (1995) developed the concept that CO2 depletion/O2 release caused by photosynthesis is counterbalanced by CO2 release/O2 uptake during photorespiration. Tolbert et al. proposed that the equilibrium establishes at concentrations of CO2 and O2 near the preindustrial level. In an experiment with tobacco and spinach in closed chambers, assimilation of CO2 led to its relative exhaustion down to low concentrations and to an increase of O2 with the establishment of an equilibrium O2/CO2 ratio (Tolbert et al. 1995). The equilibrium concentration of O2 was established at 23% with a CO2 concentration of 220 ppm – which is close to the glacial CO2 level. The CO2 concentration of 350 ppm was in equilibrium with 27–28% O2 and the CO2 concentration of 700 ppm was in equilibrium with 35% O2. These values indicate that in the Permo–Carboniferous boundary, a predicted CO2 concentration of 300–350 ppm (Berner 1997; Beerling et al. 2001; Beerling 2002) would correspond with the moderate estimate of O2 concentration (25%) made by Lenton (2001), which takes into account phosphorus weathering, rather than the more extreme (35%) value proposed by Berner (2003a,b). However, some experiments have shown that it
is possible for plants to complete their lifecycles even at 35% O2 and 350 ppm CO2 (Beerling and Berner 2000).
The latest estimate of O2 and CO2 concentrations based on the COPSE model (Bergman et al. 2004), uniting earlier GEOCARB (Berner and Kothavala 2001) and the feedback-based atmospheric O2 model (Lenton 2001), suggests even higher late Carboniferous CO2 levels, closer to 800–900 ppm. With the suggested high late Carboniferous O2 levels, this fits nicely with the O2/ CO2 equilibration by land plants suggested by Tolbert et al. (1995). The COPSE model also gives
higher O2 levels for the late Mesozoic (Cretaceous) (30% and above), which agrees well with the suggested equilibration at high CO2 concentrations (1000 ppm) estimated for that time.
In our view, the approach taken by Tolbert et al. (1995) is of value as long as it is remembered that the experiments were carried out in closed chambers over short periods of time, and therefore have some limitations when applied to the whole biosphere. The enclosed system lacked an ocean and a rock cycle, both of which regulated the atmospheric CO2 during the evolution of land plants. To work well, the mechanism proposed by Tolbert et al. (1995) should provide a faster and more efficient O2/CO2 equilibration in the atmosphere as compared to other factors, especially the
oceanic one. In O2 evolution, the role of photosynthesis (both marine and land) was clearly predominant, but the role of different factors in CO2 depletion/evolution, requires numerous contro-versial estimates. The inclusion of the ocean in estimates of CO2 production by marine phytoplankton (Raven and Falkowski 1999) can explain the rapid recovery from glacial low CO2 concentrations, however the CO2 depletion is always a slower process.
It took many millions years in Palaeozoic to deplete CO2 during the emergence of land plants and this took place via a weathering process (Berner 1997), rather than direct photosynthetic assimilation. After asteroid impacts, the depletion
of excess CO2 took hundreds of thousands or even millions of years (McElwain et al. 1999; Ahrens
and Jahren 2000) and the interglacial rise of CO2 occupied almost a hundred thousand years (Falkowski et al. 2000). All this means is that a simple photosynthetic/photorespiratory O2/CO2 balance mechanism is not sufficient to explain gaseous homeostasis of the atmosphere. Even together with other biospheric processes, it is not able to
respond to dramatic changes such as the industrial rise of CO2, over short time periods (Berner 2003a).
However, land plant photosynthesis does mediate the main processes of CO2 depletion such as weathering and the ocean sink and this depletion never falls below the ecological CO2 compensation point, i.e. the feedback mechanisms
switch upon depletion, resulting in a rise of CO2. Thus the biospheric equilibrium of CO2/O2 concentration works with a feedback mechanisms that may be responsible for the oscillatory regime, probably together with synchronization of these oscillations with the cycles of solar activity. This indicates that the approach of Tolbert et al. (1995) has a rational basis and requires further development and clarification to explain the observed biospheric phenomena.
The role of weathering by land plants in CO2 depletion in the atmosphere
It has been suggested that the removal of CO2 from the biosphere occurs more via the weathering of silicates and formation of carbonates, during root activity and soil formation, rather than via photosynthetic assimilation (Berner 1997). The plant activity in this process is apparently more important than are temperature and rainfall. The weathering process is directly connected with the photosynthetic activity of land plants and a requirement for phosphate, being caused mainly by the excretion of citrate and oxalate from roots (Diatloff et al. 2004). Oxalate is formed as a side product of the glycolate oxidase reaction in leaves, when the reduction level in mitochondria established during high rates of photorespiration, suppresses the oxidation of glycine (Igamberdiev and
Lea 2002). Other pathways of oxalate formation that occur in plants, involve isocitrate lyase that
may also operate in the cytosol (Igamberdiev et al. 1986), oxaloacetate lyase (Raven et al. 1982), and
the catabolism of ascorbate (de Bolt et al. 2004; Green and Fry 2005). These pathways are not
linked directly to photorespiration but they relate to high photosynthetic and respiratory activities.
Citrate is also formed when the reduction level in mitochondria increases, due to switching from the
complete to the partial TCA cycle (Igamberdiev and Gardestro¨m 2003). Organic acids can be
transported to roots, or alternatively citrate can be formed in roots due to high activities of citrate
synthase (Kihara et al. 2003). At the pH of phloem sap, oxalate salts are transported together with
citrate (which constitutes one third of all acids transported) and malate. The transport through
the phloem flow together with the formation of organic acids in the rhizosphere constitutes an
effective mechanism of organic acid excretion by the root system (Jones 1998).
In the weathering process, citrate and oxalate release phosphate and other mineral compounds,
during soil formation. The release of phosphate led to the eutrophication of the ocean and to th
increase of photosynthetic productivity by algae, as 
well as to the substitution of silicates by carbonates, both processes consuming atmospheric CO2 (Lenton 2001). The weathering was more intensive than the release of CO2 to the atmosphere due to geological and biological activities including fires and respiration in extended periods, except during the short times of volcanic eruptions etc. This led to a decrease in the atmospheric concentration of CO2
up to the time of glaciation, when the CO2 production rate increased. Glaciation strongly decreased the process of weathering (due to both lower rates of metabolism in the cooler and drier climate and less plant cover) and the CO2 concentration began to rise mostly via negative feedback on oceanic phytoplankton, which release CO2 more intensively as the CO2 concentration falls (Raven 1994, 2000; Raven and Falkowski 1999;
Riebesell et al. 2000). At low CO2 concentration, fast-growing grasslands are predominant, while when the CO2 increases, slower growing trees become widespread (Bond et al. 2003), contributing to different rates of weathering and CO2 production. In simulation experiments with changing temperature in enclosed environments, it was
shown that at higher temperatures the release of CO2 increases sharply (Gerber et al. 2004).
Vascular plants amplify the rate of weathering by about an order of magnitude relative to lichens and mosses (Lenton 2001). Photosynthetic uptake of CO2 (and release of O2) (currently 8.4·1015 mol y)1) is counterbalanced by respiration (and photorespiration in land plants) (currently8.39·1015 mol y)1). The photosynthesis process will deplete CO2 if burial of organic C takes place, the rate of which is currently 1013 mol y)1 and was higher in the periods with elevated CO2 concentrations at the Permian–Carboniferous boundary (Lenton 1998, 2001). The opposite process
(organic carbon weathering and degassing) is estimated to be approximately the same rate as photosynthetic CO2 assimilation, but this is a matter of debate, especially for earlier times, e.g. late Devonian and Carboniferous (Lenton 2001). Thus it is difficult to estimate the relative input of Tolbert’s factor (photosynthesis/photorespiration
balance) and Berner’s factor (silicate weathering) on CO2 depletion. But even if depletion of CO2
is mainly due to silicate weathering, it would take place down to values comparable with the existence of land plants, and then feedback processes would start to act. Interactions of CO2 forming and CO2 depleting factors may contribute to the glacial-interglacial oscillations (see below). If burial of carbon takes place, as in the Carboniferous, or a total increase of biomass takes place as in the interglacial periods, this will also be a cause of CO2 depletion. Probably the
initial depletion was mainly due to the weathering process, while near the CO2 minimum/O2 maximum, the depletion is mainly due to photosynthetic assimilation. Near this critical point, plants at lower latitudes and also altitudes (at
higher temperatures) will produce CO2 while at higher latitudes (lower temperatures) they will consume it, but this will be limited by the decrease in temperature. Together with the uptake of CO2 by the ocean, the processes of
CO2 assimilation and CO2 release would have contributed to the glacial-interglacial oscillations (see Sigman and Boyle 2001), which were also synchronized with solar activity (Rial 2004).
Figure 6 shows the global effects of the biosphere on the balance of CO2 in the atmosphere, which includes photosynthetic CO2 assimilation, photorespiration, weathering, temperature etc. For the modelling of these events, it is necessary to know the rate constants of all the processes and their dependence on temperature and CO2 concentrations. The removal of CO2 by weathering and by direct photosynthetic activity can be estimated from the relative flux intensities of these processes. At the present time, this information is incomplete.
At the junction of the Carboniferous and Permian periods, plant diversity decreased (Figure 7). This decrease was also affected by the strongest asteroid impact at the Permian–Triassic boundary (Benton and Twitchett 2003). During
the Mesozoic era, the weathering process was probably less intensive because of the type of flora (gymnosperm) that existed (Moulton et al. 2000).This could correspond to higher concentrations of CO2. The distribution of angiosperms in the Cretaceous could cause further increase of weathering and a decrease of CO2 in the atmosphere.
O2/CO2 compensation ratio
An introduction to the concept of the compensation point (G) is needed, because it is established both by the CO2 and the O2 concentration. The CO2 compensation point is determined as the concentration of CO2, when the rate of photo synthetic CO2 assimilation is equal to CO2 release by respiration and photorespiration. If we ignore leaf respiration, which is relatively low in the light and probably does not exceed 5% of the rate of assimilation (Atkin et al. 2000), we get gamma star (G*), the compensation point taking into consideration only photorespiration. Definitions of the
CO2 compensation point are based on a constant O2 concentration, as it has been shown that the dependence of G on O2 is linear (Farquhar et al. 1980). The value of G for C3 plants at 21% O2 is about 50 ppm CO2 at 25 C.
However, the real ecological compensation point is higher (because of the necessity of photosynthetic and respiration costs for maintenance, growth and productivity), being in general about 180 ppm depending on temperature, irradiance, humidity and other factors (Sage and Coleman 2000). Respiration by plant and non-plant organisms may contribute to the value of the ecological compensation point or at least have some feedback impact on the decrease of the atmospheric CO2. Below some lesser CO2 value than the ecological compensation point (at the O2 value 21%), plants are unable to complete their lifecycles (Sage and Coleman 2000). Lack of carbon reduces the capacity of plants to assimilate nutrients, particularly nitrogen (Andrews et al. 2004). In addition, Tolbert et al. (1995) introduced the definition of O2 compensation point, which means the concentration of O2 at a constant concentration of CO2,
when the assimilation and the release of CO2 are equal (Goyal 2001).
Combining both definitions, we get the definition of O2/CO2 compensation ratio, which represents the ratio of O2 concentration to CO2 compensation point at this concentration. For G = 50 ppm CO2, the O2/CO2 compensation ratio
is 21% O2/50 ppm CO2=4200. However, the ecological O2/CO2 compensation ratio (considering ecological compensation point of 180 ppm) is about 1200. This is probably the maximum ratio that occurred in the Carboniferous–Permian time and the Pleistocenic glacial times. The value of the compensation point for CO2
strongly depends on temperature (Jordan and Ogren 1983, 1984). The average temperature can be calculated approximately from the CO2 concentration (see below) and the way that it changed during Phanerozoic can be seen in Figure 4. The temperature dependence of the compensation point was originally proposed to be linear (Brooks
and Farquhar 1985), now it is assumed to be exponential, which is more expressed upon the increase of temperature (Bernacchi et al. 2001). We can use the dependence suggested by Brooks and Farquhar (1985) (which approximates well over the moderate mean temperature values of the Earth’s surface) to determine the effects of temperature changes on the ratio of O2 and CO2 concentrations, which is reflected in the value of the compensation ratio and in the rate of photorespiration.
An estimate of the Vo/Vc ratio of Rubisco (Figure 7), taking into account the effect of temperature, shows that it changed slightly less drastically during the past 350 Ma, than the O2/CO2 ratio. Since variation in the O2/CO2 ratio regulated by temperature is somewhat higher than that estimated by Brooks and Farquhar (1985), and since respiration (not included in our calculation) also increases drastically with a temperature rise, the real change of Vo/Vc will be 5 times or even less, while the O2/CO2 ratio changed more than ten-fold, being quite low during long periods of the Mesozoic era. At higher temperatures, the effects of O2 (e.g. photoinhibition due to a higher reduction level in the chloroplast) will be pronounced even at higher CO2 concentrations. During an increase of CO2 in the atmosphere (as in Mesozoic), which corresponds to a higher mean temperature according to Equation (1), the O2/CO2 compensation ratio will be lower, than under depletion of CO2 (this was actually observed in Earth’s history).
Evolution of land plants and evolution of the atmosphere
Correlation of O2 and CO2 in the atmosphere with the evolutionary process
If the diversity of land plants expressed as the number of land plant families according to Rothman (2001) and Benton (1993), is compared with the concentrations of O2 and CO2 in the atmosphere, there is no obvious correlation between the diversity and concentration of either O2 or CO2. However, there is a surprisingly good correlation between the O2/CO2 ratio in the atmosphere and biodiversity (Figure 7). The correlation is even better if biodiversity is plotted against the effective O2/CO2 ratio (ratio of oxygenase to carboxylase rates of Rubisco, Vo/Vc), calculated on the basis of
the effect of temperature on the compensation point according to Brooks and Farquhar (1985). It
can be seen that the increase in this ratio correlates well with diversity, while a decrease accompanied the extinction of land plant families. A similar correlation has been observed with the number of animal species (Rothman 2001), which has not been discussed in this paper. Elevation of CO2 under constant O2, i.e. a decrease of the O2/CO2
ratio, has previously been shown to reduce biodiversity in land communities (Zavaleta et al. 2001, 2003a, b; Shaw et al. 2002).
However, the diversity of marine fauna has been shown to follow directly the CO2 concentration and corresponding temperature changes (Cornette et al. 2002), while some parameters of plants are correlated directly with O2 but not to CO2, e.g. the replacement of woodiness by herbaceousness in evolution and the woodiness indices of plants (Gottlieb et al. 1995; Gottlieb and Borin 1998). In addition, it has been proposed that the evolution of secondary metabolites was triggered by O2, providing the incorporation of oxygen atoms into a series of molecular species (Gottlieb and Borin 1998).The increase in biodiversity during late Devonian and early Carboniferous corresponded to the
increase in O2/CO2 ratio, peaking at the late Carboniferous–Permian boundary. This was the period of highest diversity of land plants before the late Mesozoic. The diversity could be even higher, since e.g. angiosperms became distributed widely only in Cretaceous, but comparisons of gene sequences showed that the differentiation
between gymnosperms and angiosperms arose at the Carboniferous–Permian boundary (Savard
et al. 1994). A Carboniferous–Permian high O2 episode might have triggered this split between
major plant groups. Although carbon isotopebased estimates show an average concentration of CO2 of 0.3–0.4% in this period, recent data has indicated the existence of oscillations of CO2 (and consequently temperature) of a 100 ka period (Sigman and Boyle 2001), possibly with minimum CO2 concentrations (corresponding to glaciations)
close to that in the Pleistocenic oscillations (Tajika 2003). There is also some indication of the appearance of C4-like plants in the late Carboniferous period based on the d13C value(19&) of some fossils (Jones 1994) but this needs further substantiation.
The evolution of land plants was accomplished by the appearance and development of stomata. The stomata evolved in Silurian and the stomatalindex increased as carbon dioxide decreased from the Silurian to the Carboniferous (Royer et al. 2001a). The decrease in CO2 and increase of O2 led to the importance of maximizing CO2 diffusion into the leaf (Beerling and Woodward 1997), thereby raising intercellular CO2 concentrations and reducing CO2 evolution by photorespiration. In addition, high transpiration rates prevented the overheating of leaves, allowing the evolution of
larger leaves (Beerling et al. 2001).The decrease in O2/CO2 ratio during Permian was accompanied by a decrease in the diversity of land plants. The asteroid impact event of the Permian–Triassic boundary, the greatest in the history of Phanerozoic, completed this process destroying 90% of the marine animal species and 70% of the plant species (Rothman 2001).
While all impact events, after a short period of cooling, led to global warming and a rise in atmospheric CO2, this rise probably determined the relatively low diversity of the Triassic period. The determination of atmospheric CO2 concentrations based on stomatal characteristics, shows the periods of instant CO2 increase caused by the impacts of asteroids, more clearly than the isotopic data (Retallack 2001). This can be seen at the boundaries of Permian–Triassic, Triassic–Jurassic and Cretaceous–Palaeogene (Figure 3). Plant metabolism can be more reductive or more oxidative, depending on the O2 and CO2 concentrations in the atmosphere (Cen et al. 2001).
The increased O2/CO2 ratio is more effective at higher temperatures, providing a higher Rubisco Vo/Vc ratio, and consequently a higher compensation O2/ CO2 ratio. This is why biological diversity is higher in tropical areas and the spreading of plants occurs from the tropical areas towards moderate and subpolar latitudes (Meyen 1987). In addition, the rate of evolutionary process triggered by an increase in O2/CO2 ratio is mediated by the amount of reactive oxygen species. At higher altitudes, where active oxygen (ozone) is at a high concentration (Sandroni et al. 1994) and UV radiation is strong at lower partial pressures of atmospheric gases, the rate of evolution will be higher. The reduction level in plants (the flow of reducing power to the synthesis of biomass measured as the reductant utilization rate) strongly depends on the O2/CO2 ratio (Cen et al. 2001). Low CO2 greatly enhances plant stress symptoms (Cowling and Sage 1998), while high CO2 alleviates these effects.
At high O2 to CO2 ratios, oxygen can easily be converted to toxic superoxide and hydrogen peroxide, thus causing gene mutations (Raven 1991). An increase in angiosperm biodiversity during the Cenozoic caused for the second time during Phanerozoic, the global long-term depletion of CO2 in the atmosphere down to the ecological compensation point. The appearance of most C4 plant species is connected with the late Miocene (8–10 Ma), however, in some special areas, they appeared earlier and then distributed over the Earth. The C4 pathway independently evolved over 45 times in 19 families of angiosperms, the earliest likely being in the Chenopodiaceae dating back 15–21 Ma (Sage 2004). The C4 plants appeared in tropical areas where the compensation ratio is lower, and thus photorespiration is higher. However, C4 plants had a limited capacity for spreading into colder because of the additional energy
needed to provide the operation of the C4 cycle (Sage 2004).
Miocene was characterised by quite low CO2 concentrations (250–290 ppm) decoupled from temperature (higher by +6 C than calculated based solely on the CO2 concentration), probably because of outbursts of greenhouse methane (Zachos et al. 2001). This high temperature and a high O2/CO2 ratio caused a climate favourable for C4 metabolism. The C4 plants are adapted to low CO2 concentrations and warm climate and their photosynthetic metabolism is considered as an efficient CO2 pump (von Caemmerer and Furbank 2003). They, however, can also prosper at elevated
CO2 concentrations, particularly at elevated temperatures and in arid conditions (Sage and Kubien2003).
Thus, there is a good indication that the appearance of new genetic material is correlated with periods when the O2/CO2 ratio is maximal. During the last million years this was observed during the glaciation periods. Tropical/subtropical areas and mountain regions with a higher degree of oxygen effects were the centres for the origin of cultivated plants (Vavilov 1926). Recessive genes then drifted to the peripheral regions of the species distribution. Cultivated plants are characterised by a channelling of metabolism towards a higher productivity of certain storage tissues. It is likely that the genetic diversity was built up in tropical areas during glaciation periods, while the spreading occurred during the interglacial periods. Thus the origin of agriculture was linked to the transition from glaciation to higher CO2 concentrations, when agriculture could be effective.
Greenhouse bursts and their consequences
The rate of increase of CO2 concentration in the atmosphere was rapid (several-fold in short time periods) on several occasions during Phanerozoic, mostly after impacts of large meteorites (postapocalyptic greenhouse effects) (Retallack 1999). After these events CO2 was established at a higher concentration. At the Cretaceous–Palaeogene
boundary, there was an increase in CO2 from 350–500 ppm to 2300 ppm within less than 10,000 years according to the data of stomatal densities (McElwain et al. 1999; Retallack 2001). The CO2 concentration then decreased to the initial value during hundreds of thousands years (Beerling et al. 2002). The warming impaired leaf photosynthetic function and severely reduced carbon uptake (McElwain et al. 1999).
During the Cretaceous–Palaeogene boundary, carbon isotopic recovery (return to the 13C values before the asteroid impact) was observed between 65.00± .05 Ma and 65.16±0.04 Ma for the terrestrial biosphere (more rapidly than for the marine biosphere), i.e. it took 100–200 thousand years (Ahrens and Jahren 2000; Beerling et al. 2001). This decrease in CO2 concentration was due to the unique role of land plants in controlling the O2 and CO2 ratio, which is supported by the fact that the terrestrial ecosystems were recovering ahead of marine production (Beerling et al. 2001), however, it was relatively slow as compared to the fast greenhouse bursts.
Higher CO2 concentrations, which persisted for long periods of millions of years, may have resulted either from an unusual solar activity or from the specific properties of the flora with a low weathering rate and slow metabolism (possibly gymnosperms in the Mezozoic). According to the above considerations, we have assumed a major
role of photorespiration in maintaining a climate suitable for biospheric development. During the last 1 million years, oscillations of CO2 concentration near the ecological compensation point were accompanied by corresponding climate changes.
They could be explained by a self-regulatory role of the biosphere (a decrease of weathering at the lowest CO2 concentrations leading to a CO2 increase), however they were also synchronized with solar activity changes (Rial 2004). The period of these oscillations was around 0.1 Ma and the amplitude was between 180 and 280 ppm, i.e. by about 60% (Sigman and Boyle 2001). Now the indications are that these oscillations will stop, and most scientists claim (Falkowski et al. 2000) that this is due to human activity. However, it is possible that when considering increases in temperature and CO2, cause and effect may have become confused (Khilyuk and Chilingar 2003).
If we consider the role of the sun in climate change, which during the periods of higher solar activity resulted in an increase in the temperature of the Earth and CO2 release from oceans to the atmosphere, and even according to some estimates provided about 40% of the present global warming (Beer et al. 2000), we can also attribute the modern increase of temperature and CO2 to solar activity.
The data on sunspot number reconstruction shows that the period of very high solar activity during the last 60 years is unique throughout the last millennium (Usoskin et al.2003). Taking into account the relatively low CO2 concentration in the past million years (Royer et al. 2001b) and its oscillations coincident with solar activity, mainly with Milankovitch cycles (cycles in the Earth’s orbit that influence the amount of solar radiation including a 100 ka eccentricity cycle corresponding to major glaciations) (Rial 2004), we can regard unusually high solar activity from the middle of the 20th century as an important reason for global warming, which may be comparable to the anthropogenic release of CO2. It corresponds to the increase of CO2 occurring exponentially at the present time (Figure 8). A doubling of CO2 concentration is predicted in 100 years or even less (Falkowski et al. 2000).
For such high rate of CO2 emission, there is no natural process, which can cope with this increase. It is possible that we will return to the Pre-Devonian state, when the CO2 concentration was high and not fully regulated by the biosphere, and this situation will be beyond human control.
General conclusion
Photorespiration has played an important role since the appearance of land plants, in the regulation of the O2 and CO2 concentrations in the atmosphere. The limits of photosynthetic/photorespiratory parameters (O2/CO2 compensation ratio) based on Rubisco kinetics, determined the limits of variation in O2 and CO2 concentrations
(O2/CO2 ratio). The established atmospheric O2/ CO2 ratio was coincident with the rates of evolution of land plants. The current increase in CO2 concentration caused both by anthropogenic burning of fossil fuel and by unusual solar
activity is unique, and a similar CO2 increase did not occur from the early Cenozoic. If we consider the predicted CO2 increase later than the year 2100, the possible effect may be compared even with early Devonian.
Acknowledgments
The authors wish to thank Dr. Alf Keys and Dr. Leszek Kleczkowski for their critical reading of the manuscript and are also grateful to an anonymous reviewer for a large number of very helpful suggestions.
References
