The evolution of plants has resulted in increasing levels of complexity, from the earliest algal mats, through bryophytes, lycopods, ferns to the complex gymnosperms and angiosperms of today. While many of the groups which appeared earlier continue to thrive, especially in the environments in which they evolved, for a time each new grade of organisation became more "successful" than its predecessors.
In the Ordovician period, around 450 million years ago, the first land plants appeared. These began to diversify in the late Silurian Period, around 420 million years ago, and the results of their diversification are displayed in remarkable detail in an early Devonian fossil assemblage from the Rhynie chert. This chert preserved early plants in cellular detail, petrified in volcanic springs.
By the middle of the Devonian Period, most of the features recognised in plants today are present, including roots, leaves and secondary wood; and, by late Devonian times, seeds had evolved. Late Devonian plants had thereby reached a degree of sophistication that allowed them to form forests of tall trees.
Evolutionary innovation continued into the Carboniferous period and is still ongoing today. Most plant groups were relatively unscathed by the Permo-Triassic extinction event, although the structures of communities changed. This may have set the scene for the appearance of the flowering plants in the Triassic (~200 million years ago), and their later diversification in the Cretaceous and Paleogene. The latest major group of plants to evolve were the grasses, which became important in the mid-Paleogene, from around 40 million years ago. The grasses, as well as many other groups, evolved new mechanisms of metabolism to survive the low CO2 and warm, dry conditions of the tropics over the last 10 million years.
Land plants evolved from chlorophyte algae, perhaps as early as 510 million years ago; some molecular estimates place their origin even earlier, as much as 630 million years ago. Their closest living relatives are the charophytes, specifically Charales; assuming that the Charales' habit has changed little since the divergence of lineages, this means that the land plants evolved from a branched, filamentous alga dwelling in shallow fresh water, perhaps at the edge of seasonally desiccating pools. The alga would have had a haplontic life cycle: it would only very briefly have had paired chromosomes (the diploid condition) when the egg and sperm first fused to form a zygote; this would have immediately divided by meiosis to produce cells with half the number of unpaired chromosomes (the haploid condition). Co-operative interactions with fungi may have helped early plants adapt to the stresses of the terrestrial realm.
Plants were not the first photosynthesisers on land; weathering rates suggest that organisms were already living on the land 1,200 million years ago, and microbial fossils have been found in freshwater lake deposits from 1,000 million years ago, but the carbon isotope record suggests that they were too scarce to impact the atmospheric composition until around 850 million years ago. These organisms, although phylogenetically diverse, were probably small and simple, forming little more than an "algal scum".
The first evidence of plants on land comes from spores of Mid-Ordovician age (early Llanvirn, ~470 million years ago). These spores, known as cryptospores, were produced either singly (monads), in pairs (diads) or groups of four (tetrads), and their microstructure resembles that of modern liverwort spores, suggesting they share an equivalent grade of organisation. They are composed of sporopollenin – further evidence of an embryophytic affinity. It could be that atmospheric 'poisoning' prevented eukaryotes from colonising the land prior to this, or it could simply have taken a great time for the necessary complexity to evolve.
Trilete spores similar to those of vascular plants appear soon afterwards, in Upper Ordovician rocks. Depending exactly when the tetrad splits, each of the four spores may bear a "trilete mark", a Y-shape, reflecting the points at which each cell squashed up against its neighbours. However, this requires that the spore walls be sturdy and resistant at an early stage. This resistance is closely associated with having a desiccation-resistant outer wall—a trait only of use when spores must survive out of water. Indeed, even those embryophytes that have returned to the water lack a resistant wall, thus don't bear trilete marks. A close examination of algal spores shows that none have trilete spores, either because their walls are not resistant enough, or in those rare cases where it is, the spores disperse before they are squashed enough to develop the mark, or don't fit into a tetrahedral tetrad.
The earliest megafossils of land plants were thalloid organisms, which dwelt in fluvial wetlands and are found to have covered most of an early Silurian flood plain. They could only survive when the land was waterlogged. There were also microbial mats.
Once plants had reached the land, there were two approaches to dealing with desiccation. The bryophytes avoid it or give in to it, restricting their ranges to moist settings, or drying out and putting their metabolism "on hold" until more water arrives. Tracheophytes resist desiccation: They all bear a waterproof outer cuticle layer wherever they are exposed to air (as do some bryophytes), to reduce water loss, but—since a total covering would cut them off from CO2 in the atmosphere—they rapidly evolved stomata, small openings to allow gas exchange. Tracheophytes also developed vascular tissue to aid in the movement of water within the organisms (see below), and moved away from a gametophyte dominated life cycle (see below). Vascular tissue also facilitated upright growth without the support of water and paved the way for the evolution of larger plants on land.
The establishment of a land-based flora caused increased accumulation of oxygen in the atmosphere, as the plants produced oxygen as a waste product. When this concentration rose above 13%, wildfires became possible. This is first recorded in the early Silurian fossil record by charcoalified plant fossils. Apart from a controversial gap in the Late Devonian, charcoal is present ever since.
Charcoalification is an important taphonomic mode. Wildfire drives off the volatile compounds, leaving only a shell of pure carbon. This is not a viable food source for herbivores or detritovores, so is prone to preservation; it is also robust, so can withstand pressure and display exquisite, sometimes sub-cellular, detail.
All multicellular plants have a life cycle comprising two generations or phases. One is termed the gametophyte, has a single set of chromosomes (denoted 1N), and produces gametes (sperm and eggs). The other is termed the sporophyte, has paired chromosomes (denoted 2N), and produces spores. The gametophyte and sporophyte may appear identical – homomorphy – or may be very different – heteromorphy.
The pattern in plant evolution has been a shift from homomorphy to heteromorphy. The algal ancestors of land plants were almost certainly haplobiontic, being haploid for all their life cycles, with a unicellular zygote providing the 2N stage. All land plants (i.e. embryophytes) are diplobiontic – that is, both the haploid and diploid stages are multicellular. Two trends are apparent: bryophytes (liverworts, mosses and hornworts) have developed the gametophyte, with the sporophyte becoming almost entirely dependent on it; vascular plants have developed the sporophyte, with the gametophyte being particularly reduced in the seed plants.
It has been proposed that the basis for the emergence of the diploid phase of the life cycle as the dominant phase, is that diploidy allows masking of the expression of deleterious mutations through genetic complementation. Thus if one of the parental genomes in the diploid cells contains mutations leading to defects in one or more gene products, these deficiencies could be compensated for by the other parental genome (which nevertheless may have its own defects in other genes). As the diploid phase was becoming predominant, the masking effect likely allowed genome size, and hence information content, to increase without the constraint of having to improve accuracy of replication. The opportunity to increase information content at low cost is advantageous because it permits new adaptations to be encoded. This view has been challenged, with evidence showing that selection is no more effective in the haploid than in the diploid phases of the lifecycle of mosses and angiosperms.
There are two competing theories to explain the appearance of a diplobiontic lifecycle.
The interpolation theory (also known as the antithetic or intercalary theory) holds that the sporophyte phase was a fundamentally new invention, caused by the mitotic division of a freshly germinated zygote, continuing until meiosis produces spores. This theory implies that the first sporophytes bore a very different morphology to the gametophyte they depended on. This seems to fit well with what is known of the bryophytes, in which a vegetative thalloid gametophyte is parasitised by simple sporophytes, which often comprise no more than a sporangium on a stalk. Increasing complexity of the ancestrally simple sporophyte, including the eventual acquisition of photosynthetic cells, would free it from its dependence on a gametophyte, as seen in some hornworts (Anthoceros), and eventually result in the sporophyte developing organs and vascular tissue, and becoming the dominant phase, as in the tracheophytes (vascular plants). This theory may be supported by observations that smaller Cooksonia individuals must have been supported by a gametophyte generation. The observed appearance of larger axial sizes, with room for photosynthetic tissue and thus self-sustainability, provides a possible route for the development of a self-sufficient sporophyte phase.
The alternative hypothesis is termed the transformation theory (or homologous theory). This posits that the sporophyte appeared suddenly by a delay in the occurrence of meiosis after the zygote germinated. Since the same genetic material would be employed, the haploid and diploid phases would look the same. This explains the behaviour of some algae, which produce alternating phases of identical sporophytes and gametophytes. Subsequent adaption to the desiccating land environment, which makes sexual reproduction difficult, would result in the simplification of the sexually active gametophyte, and elaboration of the sporophyte phase to better disperse the waterproof spores. The tissue of sporophytes and gametophytes preserved in the Rhynie chert is of similar complexity, which is taken to support this hypothesis.
To photosynthesise, plants must absorb CO2 from the atmosphere. However, this comes at a price: while stomata are open to allow CO2 to enter, water can evaporate. Water is lost much faster than CO2 is absorbed, so plants need to replace it, and have developed systems to transport water from the moist soil to the site of photosynthesis. Early plants sucked water between the walls of their cells, then evolved the ability to control water loss (and CO2 acquisition) through the use of a waterproof cuticle perforated by stomata. Specialised water transport tissues soon evolved in the form of hydroids, tracheids, then secondary xylem, followed by an endodermis and ultimately vessels.
The high CO2 levels of Silurian-Devonian times, when plants were first colonising land, meant that the need for water was relatively low. As CO2 was withdrawn from the atmosphere by plants, more water was lost in its capture, and more elegant transport mechanisms evolved. As water transport mechanisms, and waterproof cuticles, evolved, plants could survive without being continually covered by a film of water. This transition from poikilohydry to homoiohydry opened up new potential for colonisation. Plants then needed a robust internal structure that contained long narrow channels for transporting water from the soil to all the different parts of the above-soil plant, especially to the parts where photosynthesis occurred.
During the Silurian, CO2 was readily available, so little water needed to be expended to acquire it. By the end of the Carboniferous, when CO2 levels had lowered to something approaching today's, around 17 times more water was lost per unit of CO2 uptake. However, even in these "easy" early days, water was at a premium, and had to be transported to parts of the plant from the wet soil to avoid desiccation. This early water transport took advantage of the cohesion-tension mechanism inherent in water. Water has a tendency to diffuse to areas that are drier, and this process is accelerated when water can be wicked along a fabric with small spaces. In small passages, such as that between the plant cell walls (or in tracheids), a column of water behaves like rubber – when molecules evaporate from one end, they literally pull the molecules behind them along the channels. Therefore transpiration alone provided the driving force for water transport in early plants. However, without dedicated transport vessels, the cohesion-tension mechanism cannot transport water more than about 2 cm, severely limiting the size of the earliest plants. This process demands a steady supply of water from one end, to maintain the chains; to avoid exhausing it, plants developed a waterproof cuticle. Early cuticle may not have had pores but did not cover the entire plant surface, so that gas exchange could continue. However, dehydration at times was inevitable; early plants cope with this by having a lot of water stored between their cell walls, and when it comes to it sticking out the tough times by putting life "on hold" until more water is supplied.
To be free from the constraints of small size and constant moisture that the parenchymatic transport system inflicted, plants needed a more efficient water transport system. During the early Silurian, they developed specialized cells, which were lignified (or bore similar chemical compounds) to avoid implosion; this process coincided with cell death, allowing their innards to be emptied and water to be passed through them. These wider, dead, empty cells were a million times more conductive than the inter-cell method, giving the potential for transport over longer distances, and higher CO2 diffusion rates.
The earliest macrofossils to bear water-transport tubes are Silurian plants placed in the genus Cooksonia. The early Devonian pretracheophytes Aglaophyton and Horneophyton have structures very similar to the hydroids of modern mosses.
Plants continued to innovate new ways of reducing the resistance to flow within their cells, thereby increasing the efficiency of their water transport. Thickened bands on the walls of tubes are apparent from the early Silurian onwards are adaptations to ease the flow of water. Banded tubes, as well as tubes with pitted ornamentation on their walls, were lignified and, when they form single celled conduits, are referred to as tracheids. These, the "next generation" of transport cell design, have a more rigid structure than hydroids, preventing their collapse at higher levels of water tension. Tracheids may have a single evolutionary origin, possibly within the hornworts, uniting all tracheophytes (but they may have evolved more than once).
Water transport requires regulation, and dynamic control is provided by stomata. By adjusting the amount of gas exchange, they can restrict the amount of water lost through transpiration. This is an important role where water supply is not constant, and indeed stomata appear to have evolved before tracheids, being present in the non-vascular hornworts.
An endodermis probably evolved during the Silu-Devonian, but the first fossil evidence for such a structure is Carboniferous. This structure in the roots covers the water transport tissue and regulates ion exchange (and prevents unwanted pathogens etc. from entering the water transport system). The endodermis can also provide an upwards pressure, forcing water out of the roots when transpiration is not enough of a driver.
Once plants had evolved this level of controlled water transport, they were truly homoiohydric, able to extract water from their environment through root-like organs rather than relying on a film of surface moisture, enabling them to grow to much greater size. As a result of their independence from their surroundings, they lost their ability to survive desiccation – a costly trait to retain.
During the Devonian, maximum xylem diameter increased with time, with the minimum diameter remaining pretty constant. By the mid Devonian, the tracheid diameter of some plant lineages had plateaued. Wider tracheids allow water to be transported faster, but the overall transport rate depends also on the overall cross-sectional area of the xylem bundle itself. The increase in vascular bundle thickness further seems to correlate with the width of plant axes, and plant height; it is also closely related to the appearance of leaves and increased stomatal density, both of which would increase the demand for water.
While wider tracheids with robust walls make it possible to achieve higher water transport pressures, this increases the problem of cavitation. Cavitation occurs when a bubble of air forms within a vessel, breaking the bonds between chains of water molecules and preventing them from pulling more water up with their cohesive tension. A tracheid, once cavitated, cannot have its embolism removed and return to service (except in a few advanced angiosperms[verification needed] that have developed a mechanism of doing so). Therefore, it is well worth plants' while to avoid cavitation occurring. For this reason, pits in tracheid walls have very small diameters, to prevent air entering and allowing bubbles to nucleate. Freeze-thaw cycles are a major cause of cavitation. Damage to a tracheid's wall almost inevitably leads to air leaking in and cavitation, hence the importance of many tracheids working in parallel.
Cavitation is hard to avoid, but once it has occurred plants have a range of mechanisms to contain the damage. Small pits link adjacent conduits to allow fluid to flow between them, but not air – although ironically these pits, which prevent the spread of embolisms, are also a major cause of them. These pitted surfaces further reduce the flow of water through the xylem by as much as 30%. Conifers, by the Jurassic, developed an ingenious improvement, using valve-like structures to isolate cavitated elements. These torus-margo structures have a blob floating in the middle of a donut; when one side depressurises the blob is sucked into the torus and blocks further flow. Other plants simply accept cavitation; for instance, oaks grow a ring of wide vessels at the start of each spring, none of which survive the winter frosts. Maples use root pressure each spring to force sap upwards from the roots, squeezing out any air bubbles.
Growing to height also employed another trait of tracheids – the support offered by their lignified walls. Defunct tracheids were retained to form a strong, woody stem, produced in most instances by a secondary xylem. However, in early plants, tracheids were too mechanically vulnerable, and retained a central position, with a layer of tough sclerenchyma on the outer rim of the stems. Even when tracheids do take a structural role, they are supported by sclerenchymatic tissue.
Tracheids end with walls, which impose a great deal of resistance on flow; vessel members have perforated end walls, and are arranged in series to operate as if they were one continuous vessel. The function of end walls, which were the default state in the Devonian, was probably to avoid embolisms. An embolism is where an air bubble is created in a tracheid. This may happen as a result of freezing, or by gases dissolving out of solution. Once an embolism is formed, it usually cannot be removed (but see later); the affected cell cannot pull water up, and is rendered useless.
End walls excluded, the tracheids of prevascular plants were able to operate under the same hydraulic conductivity as those of the first vascular plant, Cooksonia.
The size of tracheids is limited as they comprise a single cell; this limits their length, which in turn limits their maximum useful diameter to 80 μm. Conductivity grows with the fourth power of diameter, so increased diameter has huge rewards; vessel elements, consisting of a number of cells, joined at their ends, overcame this limit and allowed larger tubes to form, reaching diameters of up to 500 μm, and lengths of up to 10 m.
Vessels first evolved during the dry, low CO2 periods of the late Permian, in the horsetails, ferns and Selaginellales independently, and later appeared in the mid Cretaceous in angiosperms and gnetophytes. Vessels allow the same cross-sectional area of wood to transport around a hundred times more water than tracheids! This allowed plants to fill more of their stems with structural fibres, and also opened a new niche to vines, which could transport water without being as thick as the tree they grew on. Despite these advantages, tracheid-based wood is a lot lighter, thus cheaper to make, as vessels need to be much more reinforced to avoid cavitation.
For a more ecological discussion on the evolution of flowers, go to Flower
Flowers are modified leaves possessed only by the angiosperms, which are relatively late to appear in the fossil record; the group originated and diversified during the Early Cretaceous and became ecologically significant thereafter. Flower-like structures first appear in the fossil records some ~130 mya, in the Cretaceous period. Colourful and/or pungent structures surround the cones of plants such as cycads and gnetales, making a strict definition of the term "flower" elusive.
The main function of a flower is reproduction, which, before the evolution of the flower and angiosperms, was the job of microsporophylls and megasporophylls. A flower can be considered a powerful evolutionary innovation, because its presence allowed the plant world to access new means and mechanisms for reproduction.
The flowering plants have long been assumed to have evolved from within the gymnosperms; according to the traditional morphological view, they are closely allied to the gnetales. However, as noted above, recent molecular evidence is at odds to this hypothesis, and further suggests that gnetales are more closely related to some gymnosperm groups than angiosperms, and that extant gymnosperms form a distinct clade to the angiosperms, the two clades diverging some 300 million years ago.
The relationship of stem groups to the angiosperms is important in determining the evolution of flowers. stem groups provide an insight into the state of earlier "forks" on the path to the current state. Convergence increases the risk of misidentifying stem groups. Since the protection of the megagametophyte is evolutionarily desirable, probably many separate groups evolved protective encasements independently. In flowers, this protection takes the form of a carpel, evolved from a leaf and recruited into a protective role, shielding the ovules. These ovules are further protected by a double-walled integument.
Penetration of these protective layers needs something more than a free-floating microgametophyte. Angiosperms have pollen grains comprising just three cells. One cell is responsible for drilling down through the integuments, and creating a conduit for the two sperm cells to flow down. The megagametophyte has just seven cells; of these, one fuses with a sperm cell, forming the nucleus of the egg itself, and another joins with the other sperm, and dedicates itself to forming a nutrient-rich endosperm. The other cells take auxiliary roles.[clarification needed] This process of "double fertilisation" is unique and common to all angiosperms.
In the fossil record, there are three intriguing groups which bore flower-like structures. The first is the Permian pteridosperm Glossopteris, which already bore recurved leaves resembling carpels. The Triassic Caytonia is more flower-like still, with enclosed ovules – but only a single integument. Further, details of their pollen and stamens set them apart from true flowering plants.
The Bennettitales bore remarkably flower-like organs, protected by whorls of bracts which may have played a similar role to the petals and sepals of true flowers; however, these flower-like structures evolved independently, as the Bennettitales are more closely related to cycads and ginkgos than to the angiosperms.
However, no true flowers are found in any groups save those extant today. Most morphological and molecular analyses place Amborella, the nymphaeales and Austrobaileyaceae in a basal clade dubbed "ANA". This clade appear to have diverged in the early Cretaceous, around 130 million years ago – around the same time as the earliest fossil angiosperm, and just after the first angiosperm-like pollen, 136 million years ago. The magnoliids diverged soon after, and a rapid radiation had produced eudicots and monocots by 125 million years ago. By the end of the Cretaceous 66 million years ago, over 50% of today's angiosperm orders had evolved, and the clade accounted for 70% of global species. It was around this time that flowering trees became dominant over conifers 
The features of the basal "ANA" groups suggest that angiosperms originated in dark, damp, frequently disturbed areas. It appears that the angiosperms remained constrained to such habitats throughout the Cretaceous – occupying the niche of small herbs early in the successional series. This may have restricted their initial significance, but given them the flexibility that accounted for the rapidity of their later diversifications in other habitats.
Phylogeny of anthophytes and gymnosperms, from 
|Traditional view||Modern view|
The family Amborellaceae is regarded as being the sister clade to all other living flowering plants. The complete genome of Amborella trichopoda is still being sequenced as of March 2012. By comparing its genome with those of all other living flowering plants, it will be possible to work out the most likely characteristics of the ancestor of A. trichopoda and all other flowering plants, i.e. the ancestral flowering plant.
It seems that on the level of the organ, the leaf may be the ancestor of the flower, or at least some floral organs. When some crucial genes involved in flower development are mutated, clusters of leaf-like structures arise in place of flowers. Thus, sometime in history, the developmental program leading to formation of a leaf must have been altered to generate a flower. There probably also exists an overall robust framework within which the floral diversity has been generated. An example of that is a gene called LEAFY (LFY), which is involved in flower development in Arabidopsis thaliana. The homologs of this gene are found in angiosperms as diverse as tomato, snapdragon, pea, maize and even gymnosperms. Expression of Arabidopsis thaliana LFY in distant plants like poplar and citrus also results in flower-production in these plants. The LFY gene regulates the expression of some genes belonging to the MADS-box family. These genes, in turn, act as direct controllers of flower development.
The members of the MADS-box family of transcription factors play a very important and evolutionarily conserved role in flower development. According to the ABC Model of flower development, three zones — A,B and C — are generated within the developing flower primordium, by the action of some transcription factors, that are members of the MADS-box family. Among these, the functions of the B and C domain genes have been evolutionarily more conserved than the A domain gene. Many of these genes have arisen through gene duplications of ancestral members of this family. Quite a few of them show redundant functions.
The evolution of the MADS-box family has been extensively studied. These genes are present even in pteridophytes, but the spread and diversity is many times higher in angiosperms. There appears to be quite a bit of pattern into how this family has evolved. Consider the evolution of the C-region gene AGAMOUS (AG). It is expressed in today's flowers in the stamens, and the carpel, which are reproductive organs. Its ancestor in gymnosperms also has the same expression pattern. Here, it is expressed in the strobili, an organ that produces pollen or ovules. Similarly, the B-genes' (AP3 and PI) ancestors are expressed only in the male organs in gymnosperms. Their descendants in the modern angiosperms also are expressed only in the stamens, the male reproductive organ. Thus, the same, then-existing components were used by the plants in a novel manner to generate the first flower. This is a recurring pattern in evolution.
Another floral feature that has been a subject of natural selection is flowering time. Some plants flower early in their life cycle, others require a period of vernalization before flowering. This decision is based on factors like temperature, light intensity, presence of pollinators and other environmental signals: genes like CONSTANS (CO), Flowering Locus C (FLC) and FRIGIDA regulate integration of environmental signals into the pathway for flower development. Variations in these loci have been associated with flowering time variations between plants. For example, Arabidopsis thaliana ecotypes that grow in the cold, temperate regions require prolonged vernalization before they flower, while the tropical varieties, and the most common lab strains, don't. This variation is due to mutations in the FLC and FRIGIDA genes, rendering them non-functional.
Quite a few players in this process are conserved across all the plants studied. Sometimes though, despite genetic conservation, the mechanism of action turns out to be different. For example, rice is a short-day plant, while Arabidopsis thaliana is a long-day plant. Now, in both plants, the proteins CO and FLOWERING LOCUS T (FT) are present. But, in Arabidopsis thaliana, CO enhances FT production, while in rice, the CO homolog represses FT production, resulting in completely opposite downstream effects.
There are many theories that propose how flowers evolved. Some of them are described below.
The Anthophyte Theory was based on the observation that a gymnospermic group Gnetales has a flower-like ovule. It has partially developed vessels as found in the angiosperms, and the megasporangium is covered by three envelopes, like the ovary structure of angiosperm flowers. However, many other lines of evidence show that Gnetales is not related to angiosperms.
The Mostly Male Theory has a more genetic basis. Proponents of this theory point out that the gymnosperms have two very similar copies of the gene LFY, while angiosperms just have one. Molecular clock analysis has shown that the other LFY paralog was lost in angiosperms around the same time as flower fossils become abundant, suggesting that this event might have led to floral evolution. According to this theory, loss of one of the LFY paralog led to flowers that were more male, with the ovules being expressed ectopically. These ovules initially performed the function of attracting pollinators, but sometime later, may have been integrated into the core flower.
Photosynthesis is not quite as simple as adding water to CO2 to produce sugars and oxygen. A complex chemical pathway is involved, facilitated along the way by a range of enzymes and co-enzymes. The enzyme RuBisCO is responsible for "fixing" CO2 – that is, it attaches it to a carbon-based molecule to form a sugar, which can be used by the plant, releasing an oxygen molecule along the way. However, the enzyme is notoriously inefficient, and just as effectively will also fix oxygen instead of CO2 in a process called photorespiration. This is energetically costly as the plant has to use energy to turn the products of photorespiration back into a form that can react with CO2.
These two pathways, with the same effect on RuBisCO, evolved a number of times independently – indeed, C4 alone arose 62 times in 18 different plant families. A number of 'pre-adaptations' seem to have paved the way for C4, leading to its clustering in certain clades: it has most frequently been innovated in plants that already had features such as extensive vascular bundle sheath tissue. Many potential evolutionary pathways resulting in the C4 phenotype are possible and have been characterised using Bayesian inference, confirming that non-photosynthetic adaptations often provide evolutionary stepping stones for the further evolution of C4.
The C4 construction is most famously used by a subset of grasses, while CAM is employed by many succulents and cacti. The trait appears to have emerged during the Oligocene, around 25 to 32 million years ago; however, they did not become ecologically significant until the Miocene, 6 to 7 million years ago. Remarkably, some charcoalified fossils preserve tissue organised into the Kranz anatomy, with intact bundle sheath cells, allowing the presence C4 metabolism to be identified without doubt at this time. Isotopic markers are used to deduce their distribution and significance. C3 plants preferentially use the lighter of two isotopes of carbon in the atmosphere, 12C, which is more readily involved in the chemical pathways involved in its fixation. Because C4 metabolism involves a further chemical step, this effect is accentuated. Plant material can be analysed to deduce the ratio of the heavier 13C to 12C. This ratio is denoted δ13C. C3 plants are on average around 14‰ (parts per thousand) lighter than the atmospheric ratio, while C4 plants are about 28‰ lighter. The δ13C of CAM plants depends on the percentage of carbon fixed at night relative to what is fixed in the day, being closer to C3 plants if they fix most carbon in the day and closer to C4 plants if they fix all their carbon at night.
It's troublesome procuring original fossil material in sufficient quantity to analyse the grass itself, but fortunately there is a good proxy: horses. Horses were globally widespread in the period of interest, and browsed almost exclusively on grasses. There's an old phrase in isotope pal?ontology, "you are what you eat (plus a little bit)" – this refers to the fact that organisms reflect the isotopic composition of whatever they eat, plus a small adjustment factor. There is a good record of horse teeth throughout the globe, and their δ13C has been measured. The record shows a sharp negative inflection around 6 to 7 million years ago, during the Messinian, and this is interpreted as the rise of C4 plants on a global scale.
While C4 enhances the efficiency of RuBisCO, the concentration of carbon is highly energy intensive. This means that C4 plants only have an advantage over C3 organisms in certain conditions: namely, high temperatures and low rainfall. C4 plants also need high levels of sunlight to thrive. Models suggest that, without wildfires removing shade-casting trees and shrubs, there would be no space for C4 plants. But, wildfires have occurred for 400 million years – why did C4 take so long to arise, and then appear independently so many times? The Carboniferous period (~300 million years ago) had notoriously high oxygen levels – almost enough to allow spontaneous combustion – and very low CO2, but there is no C4 isotopic signature to be found. And there doesn't seem to be a sudden trigger for the Miocene rise.
During the Miocene, the atmosphere and climate were relatively stable. If anything, CO2 increased gradually from 14 to 9 million years ago before settling down to concentrations similar to the Holocene. This suggests that it did not have a key role in invoking C4 evolution. Grasses themselves (the group which would give rise to the most occurrences of C4) had probably been around for 60 million years or more, so had had plenty of time to evolve C4, which, in any case, is present in a diverse range of groups and thus evolved independently. There is a strong signal of climate change in South Asia; increasing aridity – hence increasing fire frequency and intensity – may have led to an increase in the importance of grasslands. However, this is difficult to reconcile with the North American record. It is possible that the signal is entirely biological, forced by the fire- (and elephant?)- driven acceleration of grass evolution – which, both by increasing weathering and incorporating more carbon into sediments, reduced atmospheric CO2 levels. Finally, there is evidence that the onset of C4 from 9 to 7 million years ago is a biased signal, which only holds true for North America, from where most samples originate; emerging evidence suggests that grasslands evolved to a dominant state at least 15Ma earlier in South America.