The coordination of major events in C4 photosynthesis evolution in the genus Flaveria.

C4 photosynthesis is a remarkable complex trait, elucidations of the evolutionary trajectory of C4 photosynthesis from its ancestral C3 pathway can help us better understand the generic principles of the evolution of complex traits and guide the engineering of C3 crops for higher yields. Here, we used the genus Flaveria that contains C3, C3-C4, C4-like and C4 species as a system to study the evolution of C4 photosynthesis. We first mapped transcript abundance, protein sequence and morphological features onto the phylogenetic tree of the genus Flaveria, and calculated the evolutionary correlation of different features; we then predicted the relative changes of ancestral nodes of those features to illustrate the major events during the evolution of C4 photosynthesis. We found that gene expression and protein sequence showed consistent modification patterns in the phylogenetic tree. High correlation coefficients ranging from 0.46 to 0.9 among gene expression, protein sequence and morphology were observed. The greatest modification of those different features consistently occurred at the transition between C3-C4 species and C4-like species. Our results show highly coordinated changes in gene expression, protein sequence and morphological features, which support evolutionary major events during the evolution of C4 metabolism.

Targeted knockdown of ribulose-1, 5-bisphosphate carboxylase-oxygenase in rice mesophyll cells.

We generated antisense constructs targeting two of the five Rubisco small subunit genes (OsRBCS2 and 4) which account for between 30-40 % of the RBCS transcript abundance in leaf blades. The constructs were driven by a maize phosphoenolpyruvate carboxylase (PEPC) promoter known to have enriched expression in mesophyll cells (MCs). In the resulting lines leaf, Rubisco protein content was reduced by between 30-50 % and CO2 assimilation rate was limited under photorespiratory and non-photorespiratory conditions. A relationship between Rubisco protein content and CO2 assimilation rate was found. This was associated with a significant reduction in dry biomass accumulation and grain yield of between 37-70%. In addition to serving as a resource for reducing Rubisco accumulation in a cell-preferential manner, these lines allow us to characterize gene function and isoform specific suppression on photosynthesis and growth. Our results suggest that the knockdown of multiple genes is required to completely reduce Rubisco accumulation in MCs.

Distinct C4 sub-types and C3 bundle sheath isolation in the Paniceae grasses.

In C4 plants, the enzymatic machinery underpinning photosynthesis can vary, with, for example, three distinct C4 acid decarboxylases being used to release CO2 in the vicinity of RuBisCO. For decades, these decarboxylases have been used to classify C4 species into three biochemical sub-types. However, more recently, the notion that C4 species mix and match C4 acid decarboxylases has increased in popularity, and as a consequence, the validity of specific biochemical sub-types has been questioned. Using five species from the grass tribe Paniceae, we show that, although in some species transcripts and enzymes involved in multiple C4 acid decarboxylases accumulate, in others, transcript abundance and enzyme activity is almost entirely from one decarboxylase. In addition, the development of a bundle sheath isolation procedure for a close C3 species in the Paniceae enables the preliminary exploration of C4 sub-type evolution.

Installation of C4 photosynthetic pathway enzymes in rice using a single construct.

Introduction of a C4 photosynthetic mechanism into C3 crops offers an opportunity to improve photosynthetic efficiency, biomass and yield in addition to potentially improving nitrogen and water use efficiency. To create a two-cell metabolic prototype for an NADP-malic enzyme type C4 rice, we transformed Oryza sativa spp. japonica cultivar Kitaake with a single construct containing the coding regions of carbonic anhydrase, phosphoenolpyruvate (PEP) carboxylase, NADP-malate dehydrogenase, pyruvate orthophosphate dikinase and NADP-malic enzyme from Zea mays, driven by cell-preferential promoters. Gene expression, protein accumulation and enzyme activity were confirmed for all five transgenes, and intercellular localization of proteins was analysed. 13 CO2 labelling demonstrated a 10-fold increase in flux though PEP carboxylase, exceeding the increase in measured in vitro enzyme activity, and estimated to be about 2% of the maize photosynthetic flux. Flux from malate via pyruvate to PEP remained low, commensurate with the low NADP-malic enzyme activity observed in the transgenic lines. Physiological perturbations were minor and RNA sequencing revealed no substantive effects of transgene expression on other endogenous rice transcripts associated with photosynthesis. These results provide promise that, with enhanced levels of the C4 proteins introduced thus far, a functional C4 pathway is achievable in rice.

A Partial C4 Photosynthetic Biochemical Pathway in Rice.

Introduction of a C4 photosynthetic pathway into C3 rice (Oryza sativa) requires installation of a biochemical pump that concentrates CO2 at the site of carboxylation in modified bundle sheath cells. To investigate the feasibility of this, we generated a quadruple line that simultaneously accumulates four of the core C4 photosynthetic enzymes from the NADP-malic enzyme subtype, phosphoenolpyruvate carboxylase (ZmPEPC), NADP-malate dehydrogenase (ZmNADP-MDH), NADP-malic enzyme (ZmNADP-ME), and pyruvate phosphate dikinase (ZmPPDK). This led to enhanced enzyme activity and mild phenotypic perturbations but was largely neutral in its effects on photosynthetic rate. Measurements of the flux of 13CO2 through photosynthetic metabolism revealed a significant increase in the incorporation of 13C into malate, consistent with increased fixation of 13CO2 via PEP carboxylase in lines expressing the maize PEPC enzyme. However, there was no significant differences in labeling of 3-phosphoglycerate (3PGA) indicating that there was no carbon flux through NADP-ME into the Calvin-Benson cycle. There was also no significant difference in labeling of phosphoenolpyruvate (PEP) indicating that there was no carbon flux through PPDK. Crossing the quadruple line with a line with reduced glycine decarboxylase H-protein (OsGDCH) abundance led to a photosynthetic phenotype characteristic of the reduced OsGDCH line and higher labeling of malate, aspartate and citrate than in the quintuple line. There was evidence of 13C labeling of aspartate indicating 13CO2 fixation into oxaloacetate by PEPC and conversion to aspartate by the endogenous aspartate aminotransferase activity. While Kranz anatomy or other anatomical modifications have not yet been installed in these plants to enable a fully functional C4 cycle, these results demonstrate for the first-time a partial flux through the carboxylation phase of NADP-ME C4 metabolism in transgenic rice containing two of the key metabolic steps in the C4 pathway.

Transgenic maize phosphoenolpyruvate carboxylase alters leaf-atmosphere CO2 and 13CO2 exchanges in Oryza sativa.

The engineering process of C4 photosynthesis into C3 plants requires an increased activity of phosphoenolpyruvate carboxylase (PEPC) in the cytosol of leaf mesophyll cells. The literature varies on the physiological effect of transgenic maize (Zea mays) PEPC (ZmPEPC) leaf expression in Oryza sativa (rice). Therefore, to address this issue, leaf-atmosphere CO2 and 13CO2 exchanges were measured, both in the light (at atmospheric O2 partial pressure of 1.84 kPa and at different CO2 levels) and in the dark, in transgenic rice expressing ZmPEPC and wild-type (WT) plants. The in vitro PEPC activity was 25 times higher in the PEPC overexpressing (PEPC-OE) plants (~20% of maize) compared to the negligible activity in WT. In the PEPC-OE plants, the estimated fraction of carboxylation by PEPC (ß) was ~6% and leaf net biochemical discrimination against 13CO2[Formula: see text] was ~ 2‰ lower than in WT. However, there were no differences in leaf net CO2 assimilation rates (A) between genotypes, while the leaf dark respiration rates (Rd) over three hours after light-dark transition were enhanced (~ 30%) and with a higher 13C composition [Formula: see text] in the PEPC-OE plants compared to WT. These data indicate that ZmPEPC in the PEPC-OE rice plants contributes to leaf carbon metabolism in both the light and in the dark. However, there are some factors, potentially posttranslational regulation and PEP availability, which reduce ZmPEPC activity in vivo.

Knockdown of glycine decarboxylase complex alters photorespiratory carbon isotope fractionation in Oryza sativa leaves.

The influence of reduced glycine decarboxylase complex (GDC) activity on leaf atmosphere CO2 and 13CO2 exchange was tested in transgenic Oryza sativa with the GDC H-subunit knocked down in leaf mesophyll cells. Leaf measurements on transgenic gdch knockdown and wild-type plants were carried out in the light under photorespiratory and low photorespiratory conditions (i.e. 18.4 kPa and 1.84 kPa atmospheric O2 partial pressure, respectively), and in the dark. Under approximately current ambient O2 partial pressure (18.4 kPa pO2), the gdch knockdown plants showed an expected photorespiratory-deficient phenotype, with lower leaf net CO2 assimilation rates (A) than the wild-type. Additionally, under these conditions, the gdch knockdown plants had greater leaf net discrimination against 13CO2 (Δo) than the wild-type. This difference in Δo was in part due to lower 13C photorespiratory fractionation (f) ascribed to alternative decarboxylation of photorespiratory intermediates. Furthermore, the leaf dark respiration rate (Rd) was enhanced and the 13CO2 composition of respired CO2 (δ13CRd) showed a tendency to be more depleted in the gdch knockdown plants. These changes in Rd and δ13CRd were due to the amount and carbon isotopic composition of substrates available for dark respiration. These results demonstrate that impairment of the photorespiratory pathway affects leaf 13CO2 exchange, particularly the 13C decarboxylation fractionation associated with photorespiration.

Effects of reduced carbonic anhydrase activity on CO2 assimilation rates in Setaria viridis: a transgenic analysis.

In C4 species, the major ß-carbonic anhydrase (ß-CA) localized in the mesophyll cytosol catalyses the hydration of CO2 to HCO3-, which phosphoenolpyruvate carboxylase uses in the first step of C4 photosynthesis. To address the role of CA in C4 photosynthesis, we generated transgenic Setaria viridis depleted in ß-CA. Independent lines were identified with as little as 13% of wild-type CA. No photosynthetic defect was observed in the transformed lines at ambient CO2 partial pressure (pCO2). At low pCO2, a strong correlation between CO2 assimilation rates and CA hydration rates was observed. C18O16O isotope discrimination was used to estimate the mesophyll conductance to CO2 diffusion from the intercellular air space to the mesophyll cytosol (gm) in control plants, which allowed us to calculate CA activities in the mesophyll cytosol (Cm). This revealed a strong relationship between the initial slope of the response of the CO2 assimilation rate to cytosolic pCO2 (ACm) and cytosolic CA activity. However, the relationship between the initial slope of the response of CO2 assimilation to intercellular pCO2 (ACi) and cytosolic CA activity was curvilinear. This indicated that in S. viridis, mesophyll conductance may be a contributing limiting factor alongside CA activity to CO2 assimilation rates at low pCO2.

Mesophyll Chloroplast Investment in C3, C4 and C2 Species of the Genus Flaveria.

The mesophyll (M) cells of C4 plants contain fewer chloroplasts than observed in related C3 plants; however, it is uncertain where along the evolutionary transition from C3 to C4 that the reduction in M chloroplast number occurs. Using 18 species in the genus Flaveria, which contains C3, C4 and a range of C3-C4 intermediate species, we examined changes in chloroplast number and size per M cell, and positioning of chloroplasts relative to the M cell periphery. Chloroplast number and coverage of the M cell periphery declined in proportion to increasing strength of C4 metabolism in Flaveria, while chloroplast size increased with increasing C4 cycle strength. These changes increase cytosolic exposure to the cell periphery which could enhance diffusion of inorganic carbon to phosphenolpyruvate carboxylase (PEPC), a cytosolic enzyme. Analysis of the transcriptome from juvenile leaves of nine Flaveria species showed that the transcript abundance of four genes involved in plastid biogenesis-FtsZ1, FtsZ2, DRP5B and PARC6-was negatively correlated with variation in C4 cycle strength and positively correlated with M chloroplast number per planar cell area. Chloroplast size was negatively correlated with abundance of FtsZ1, FtsZ2 and PARC6 transcripts. These results indicate that natural selection targeted the proteins of the contractile ring assembly to effect the reduction in chloroplast numbers in the M cells of C4 Flaveria species. If so, efforts to engineer the C4 pathway into C3 plants might evaluate whether inducing transcriptome changes similar to those observed in Flaveria could reduce M chloroplast numbers, and thus introduce a trait that appears essential for efficient C4 function.

Independent and Parallel Evolution of New Genes by Gene Duplication in Two Origins of C4 Photosynthesis Provides New Insight into the Mechanism of Phloem Loading in C4 Species.

UNLABELLED: C4 photosynthesis is considered one of the most remarkable examples of evolutionary convergence in eukaryotes. However, it is unknown whether the evolution of C4 photosynthesis required the evolution of new genes. Genome-wide gene-tree species-tree reconciliation of seven monocot species that span two origins of C4 photosynthesis revealed that there was significant parallelism in the duplication and retention of genes coincident with the evolution of C4 photosynthesis in these lineages. Specifically, 21 orthologous genes were duplicated and retained independently in parallel at both C4 origins. Analysis of this gene cohort revealed that the set of parallel duplicated and retained genes is enriched for genes that are preferentially expressed in bundle sheath cells, the cell type in which photosynthesis was activated during C4 evolution. Furthermore, functional analysis of the cohort of parallel duplicated genes identified SWEET-13 as a potential key transporter in the evolution of C4 photosynthesis in grasses, and provides new insight into the mechanism of phloem loading in these C4 species. KEY WORDS: C4 photosynthesis, gene duplication, gene families, parallel evolution.

Targeted Knockdown of GDCH in Rice Leads to a Photorespiratory-Deficient Phenotype Useful as a Building Block for C4 Rice.

The glycine decarboxylase complex (GDC) plays a critical role in the photorespiratory C2 cycle of C3 species by recovering carbon following the oxygenation reaction of ribulose-1,5-bisphosphate carboxylase/oxygenase. Loss of GDC from mesophyll cells (MCs) is considered a key early step in the evolution of C4 photosynthesis. To assess the impact of preferentially reducing GDC in rice MCs, we decreased the abundance of OsGDCH (Os10g37180) using an artificial microRNA (amiRNA) driven by a promoter that preferentially drives expression in MCs. GDC H- and P-proteins were undetectable in leaves of gdch lines. Plants exhibited a photorespiratory-deficient phenotype with stunted growth, accelerated leaf senescence, reduced chlorophyll, soluble protein and sugars, and increased glycine accumulation in leaves. Gas exchange measurements indicated an impaired ability to regenerate ribulose 1,5-bisphosphate in photorespiratory conditions. In addition, MCs of gdch lines exhibited a significant reduction in chloroplast area and coverage of the cell wall when grown in air, traits that occur during the later stages of C4 evolution. The presence of these two traits important for C4 photosynthesis and the non-lethal, down-regulation of the photorespiratory C2 cycle positively contribute to efforts to produce a C4 rice prototype.

C4 Photosynthesis in the Rice Paddy: Insights from the Noxious Weed Echinochloa glabrescens.

The C4 pathway is a highly complex trait that increases photosynthetic efficiency in more than 60 plant lineages. Although the majority of C4 plants occupy disturbed, arid, and nutrient-poor habitats, some grow in high-nutrient, waterlogged conditions. One such example is Echinochloa glabrescens, which is an aggressive weed of rice paddies. We generated comprehensive transcriptome datasets for C4 E. glabrescens and C3 rice to identify genes associated with adaption to waterlogged, nutrient-replete conditions, but also used the data to better understand how C4 photosynthesis operates in these conditions. Leaves of E. glabrescens exhibited classical Kranz anatomy with lightly lobed mesophyll cells having low chloroplast coverage. As with rice and other hygrophytic C3 species, leaves of E. glabrescens accumulated a chloroplastic phosphoenolpyruvate carboxylase protein, albeit at reduced amounts relative to rice. The arid-grown species Setaria italica (C4) and Brachypodium distachyon (C3) were also found to accumulate chloroplastic phosphoenolpyruvate carboxylase. We identified a molecular signature associated with C4 photosynthesis in nutrient-replete, waterlogged conditions that is highly similar to those previously reported from C4 plants that grow in more arid conditions. We also identified a cohort of genes that have been subjected to a selective sweep associated with growth in paddy conditions. Overall, this approach highlights the value of using wild species such as weeds to identify adaptions to specific conditions associated with high-yielding crops in agriculture.

RNA-Seq based phylogeny recapitulates previous phylogeny of the genus Flaveria (Asteraceae) with some modifications.

BACKGROUND: The genus Flaveria has been extensively used as a model to study the evolution of C4 photosynthesis as it contains C3 and C4 species as well as a number of species that exhibit intermediate types of photosynthesis. The current phylogenetic tree of the genus Flaveria contains 21 of the 23 known Flaveria species and has been previously constructed using a combination of morphological data and three non-coding DNA sequences (nuclear encoded ETS, ITS and chloroplast encoded trnL-F). RESULTS: Here we developed a new strategy to update the phylogenetic tree of 16 Flaveria species based on RNA-Seq data. The updated phylogeny is largely congruent with the previously published tree but with some modifications. We propose that the data collection method provided in this study can be used as a generic method for phylogenetic tree reconstruction if the target species has no genomic information. We also showed that a "F. pringlei" genotype recently used in a number of labs may be a hybrid between F. pringlei (C3) and F. angustifolia (C3-C4). CONCLUSIONS: We propose that the new strategy of obtaining phylogenetic sequences outlined in this study can be used to construct robust trees in a larger number of taxa. The updated Flaveria phylogenetic tree also supports a hypothesis of stepwise and parallel evolution of C4 photosynthesis in the Flavaria clade.

Lineage-specific gene radiations underlie the evolution of novel betalain pigmentation in Caryophyllales.

Betalain pigments are unique to the Caryophyllales and structurally and biosynthetically distinct from anthocyanins. Two key enzymes within the betalain synthesis pathway have been identified: 4,5-dioxygenase (DODA) that catalyzes the formation of betalamic acid and CYP76AD1, a cytochrome P450 gene that catalyzes the formation of cyclo-DOPA. We performed phylogenetic analyses to reveal the evolutionary history of the DODA and CYP76AD1 lineages and in the context of an ancestral reconstruction of pigment states we explored the evolution of these genes in relation to the complex evolution of pigments in Caryophylalles. Duplications within the CYP76AD1 and DODA lineages arose just before the origin of betalain pigmentation in the core Caryophyllales. The duplications gave rise to DODA-α and CYP76AD1-α isoforms that appear specific to betalain synthesis. Both betalain-specific isoforms were then lost or downregulated in the anthocyanic Molluginaceae and Caryophyllaceae. Our findings suggest a single origin of the betalain synthesis pathway, with neofunctionalization following gene duplications in the CYP76AD1 and DODA lineages. Loss of DODA-α and CYP76AD1-α in anthocyanic taxa suggests that betalain pigmentation has been lost twice in Caryophyllales, and exclusion of betalain pigments from anthocyanic taxa is mediated through gene loss or downregulation. [Correction added after online publication 13 May 2015: in the last two paragraphs of the Summary the gene name CYP761A was changed to CYP76AD1.].

Dissecting Molecular Evolution in the Highly Diverse Plant Clade Caryophyllales Using Transcriptome Sequencing.

Many phylogenomic studies based on transcriptomes have been limited to "single-copy" genes due to methodological challenges in homology and orthology inferences. Only a relatively small number of studies have explored analyses beyond reconstructing species relationships. We sampled 69 transcriptomes in the hyperdiverse plant clade Caryophyllales and 27 outgroups from annotated genomes across eudicots. Using a combined similarity- and phylogenetic tree-based approach, we recovered 10,960 homolog groups, where each was represented by at least eight ingroup taxa. By decomposing these homolog trees, and taking gene duplications into account, we obtained 17,273 ortholog groups, where each was represented by at least ten ingroup taxa. We reconstructed the species phylogeny using a 1,122-gene data set with a gene occupancy of 92.1%. From the homolog trees, we found that both synonymous and nonsynonymous substitution rates in herbaceous lineages are up to three times as fast as in their woody relatives. This is the first time such a pattern has been shown across thousands of nuclear genes with dense taxon sampling. We also pinpointed regions of the Caryophyllales tree that were characterized by relatively high frequencies of gene duplication, including three previously unrecognized whole-genome duplications. By further combining information from homolog tree topology and synonymous distance between paralog pairs, phylogenetic locations for 13 putative genome duplication events were identified. Genes that experienced the greatest gene family expansion were concentrated among those involved in signal transduction and oxidoreduction, including a cytochrome P450 gene that encodes a key enzyme in the betalain synthesis pathway. Our approach demonstrates a new approach for functional phylogenomic analysis in nonmodel species that is based on homolog groups in addition to inferred ortholog groups.

Evolutionary convergence of cell-specific gene expression in independent lineages of C4 grasses.

Leaves of almost all C4 lineages separate the reactions of photosynthesis into the mesophyll (M) and bundle sheath (BS). The extent to which messenger RNA profiles of M and BS cells from independent C4 lineages resemble each other is not known. To address this, we conducted deep sequencing of RNA isolated from the M and BS of Setaria viridis and compared these data with publicly available information from maize (Zea mays). This revealed a high correlation (r=0.89) between the relative abundance of transcripts encoding proteins of the core C4 pathway in M and BS cells in these species, indicating significant convergence in transcript accumulation in these evolutionarily independent C4 lineages. We also found that the vast majority of genes encoding proteins of the C4 cycle in S. viridis are syntenic to homologs used by maize. In both lineages, 122 and 212 homologous transcription factors were preferentially expressed in the M and BS, respectively. Sixteen shared regulators of chloroplast biogenesis were identified, 14 of which were syntenic homologs in maize and S. viridis. In sorghum (Sorghum bicolor), a third C4 grass, we found that 82% of these trans-factors were also differentially expressed in either M or BS cells. Taken together, these data provide, to our knowledge, the first quantification of convergence in transcript abundance in the M and BS cells from independent lineages of C4 grasses. Furthermore, the repeated recruitment of syntenic homologs from large gene families strongly implies that parallel evolution of both structural genes and trans-factors underpins the polyphyletic evolution of this highly complex trait in the monocotyledons.

Shared origins of a key enzyme during the evolution of C4 and CAM metabolism.

CAM and C4 photosynthesis are two key plant adaptations that have evolved independently multiple times, and are especially prevalent in particular groups of plants, including the Caryophyllales. We investigate the origin of photosynthetic PEPC, a key enzyme of both the CAM and C4 pathways. We combine phylogenetic analyses of genes encoding PEPC with analyses of RNA sequence data of Portulaca, the only plants known to perform both CAM and C4 photosynthesis. Three distinct gene lineages encoding PEPC exist in eudicots (namely ppc-1E1, ppc-1E2 and ppc-2), one of which (ppc-1E1) was recurrently recruited for use in both CAM and C4 photosynthesis within the Caryophyllales. This gene is present in multiple copies in the cacti and relatives, including Portulaca. The PEPC involved in the CAM and C4 cycles of Portulaca are encoded by closely related yet distinct genes. The CAM-specific gene is similar to genes from related CAM taxa, suggesting that CAM has evolved before C4 in these species. The similar origin of PEPC and other genes involved in the CAM and C4 cycles highlights the shared early steps of evolutionary trajectories towards CAM and C4, which probably diverged irreversibly only during the optimization of CAM and C4 phenotypes.

Getting the most out of natural variation in C4 photosynthesis.

C4 photosynthesis is a complex trait that has a high degree of natural variation, involving anatomical and biochemical changes relative to the ancestral C3 state. It has evolved at least 66 times across a variety of lineages and the evolutionary route from C3 to C4 is likely conserved but not necessarily genetically identical. As such, a variety of C4 species are needed to identify what is fundamental to the C4 evolutionary process in a global context. In order to identify the genetic components of C4 form and function, a number of species are used as genetic models. These include Zea mays (maize), Sorghum bicolor (sorghum), Setaria viridis (Setaria), Flaveria bidentis, and Cleome gynandra. Each of these species has different benefits and challenges associated with its use as a model organism. Here, we propose that RNA profiling of a large sampling of C4, C3-C4, and C3 species, from as many lineages as possible, will allow identification of candidate genes necessary and sufficient to confer C4 anatomy and/or biochemistry. Furthermore, C4 model species will play a critical role in the functional characterization of these candidate genes and identification of their regulatory elements, by providing a platform for transformation and through the use of gene expression profiles in mesophyll and bundle sheath cells and along the leaf developmental gradient. Efforts should be made to sequence the genomes of F. bidentis and C. gynandra and to develop congeneric C3 species as genetic models for comparative studies. In combination, such resources would facilitate discovery of common and unique C4 regulatory mechanisms across genera.

Phenotypic landscape inference reveals multiple evolutionary paths to C4 photosynthesis.

C4 photosynthesis has independently evolved from the ancestral C3 pathway in at least 60 plant lineages, but, as with other complex traits, how it evolved is unclear. Here we show that the polyphyletic appearance of C4 photosynthesis is associated with diverse and flexible evolutionary paths that group into four major trajectories. We conducted a meta-analysis of 18 lineages containing species that use C3, C4, or intermediate C3-C4 forms of photosynthesis to parameterise a 16-dimensional phenotypic landscape. We then developed and experimentally verified a novel Bayesian approach based on a hidden Markov model that predicts how the C4 phenotype evolved. The alternative evolutionary histories underlying the appearance of C4 photosynthesis were determined by ancestral lineage and initial phenotypic alterations unrelated to photosynthesis. We conclude that the order of C4 trait acquisition is flexible and driven by non-photosynthetic drivers. This flexibility will have facilitated the convergent evolution of this complex trait. DOI:http://dx.doi.org/10.7554/eLife.00961.001.