1 Introduction

Phenotypic plasticity is defined as the variation on the phenotype of a single genotype caused by environmental conditions (Bradshaw 1965) and is essential to plant adaptation to changing environments (Fenollosa and Munné-Bosch 2019). Plant species with higher phenotypic plasticity present, in general, less specialized structures and mechanisms which allow them to be generalist in terms of environmental niches. On the other hand, plant species with specialized structures and mechanisms are believed to have a small phenotypic plasticity and so to be more niche-specific, or specialists (Sage and McKown 2006).

Sage and McKown (2006) addressed acclimation potential of C3/generalist and C4/specialist plants and pointed that although C4 plants evolved a more efficient carbon fixation pathway than C3 (Bräutigam and Gowik 2016) they have a slower acclimation to environmental constraints than C3 (Sage and McKown 2006). In their studies, they discuss a number of reasons why C4 plants might not have the same ability to acclimate as C3. However, there are few efforts to investigate the phenotypic plasticity of anatomical aspects related to gas exchange patterns.

Another underexplored point is the analysis of the phenotypic plasticity of intermediate plants. In the transitory evolution process from C3 to C4 photosynthetic pathways (Sage 2004), plants with distinct anatomical arrangements and metabolic pathways have emerged with generally called C3–C4 intermediates, usually grouped in different types as: C3 proto-Kranz, C2 and C4-like (Sage et al. 2014). The development of the carbon concentration mechanisms, in the C3–C4 intermediates and C4, required a structural and biochemical rearrangement of components present on C3, turning those groups of plants quite different from morphological to ultrastructural scales (Gowik and Westhoff 2011).

C4 photosynthesis evolved in 63 independent lineages, which are predominately distributed in monocot plants (Sage et al. 2011, 2018). Inside the Poaceae family, which has more than 20 transitions to the C4 pathway, grasses are divided into two clades: the BEP clade (Bambusoideae, Ehrhartoideae and Pooideae), where C4 pathway has not evolved; and the PACMAD clade (Panicoideae, Arundinoidea, Chloridoideae, Micrairoideae, Aristidoideae and Danthonioideae), which includes plants with the diverse photosynthetic types: C3, C3–C4 intermediates and C4 (Christin et al. 2012, GPWG II 2012). This fact turns the PACMAD clade to an important research target for studies of intermediate species due to the capacity to track facilitating characteristics of the C4 evolution (Christin et al. 2010a, b). They are also important in the current research efforts in adapting C3 plants, as rice, by the insertion of the C4 pathway (Karki et al. 2013). The genotypes used in this research, Oryza sativa (Ehrhartoideae subfamily), Homolepis isocalycia (Panicoideae subfamily) and Andropogon gayanus (Panicoideae subfamily) are species with C3, C3 proto-Kranz and C4 photosynthetic types, respectively.

Short-term water shortage is one of the most powerful environmental drivers inducing phenotypic plasticity for many terrestrial plant species (Nicotra and Davidson 2009) and can differently influence C3 and C4 responses, mainly when strengthened to increases in air (Sage and Kubien 2007) and soil (Nóia Júnior et al. 2018) temperatures. Water scarcity affects plants at different levels of its metabolism, anatomy and morphology and can greatly impact in carbon uptake and consequently in growth and production. At the metabolic levels, drought limits carbon uptake by decreasing photosynthesis, mainly, due to: a direct decrease in stomatal conductance and intercellular carbon concentration (Flexas et al. 2004); a decrease in Rubisco (Ribulose-1,5-Bisphosphate carboxylase oxygenase) activity and content (Parry et al. 2002); a decrease in the Ribulose-1,5-Bisphosphate (RuBP) regeneration (Bota et al. 2004); and a decrease in the ATP synthesis through the light reactions (Farooq et al. 2009).

Leaf anatomical structure can respond to environmental changes and influence the above-mentioned parameters, creating intricate and complex changes in metabolic cycles (Moreira et al. 2019). For instance, stomatal conductance can change accordingly to stomata number, density and shape and also epidermis area (de Boer et al. 2016); mesophyll conductance can be altered by cell organization, intercellular space patterns and leaf thickness (Terashima et al. 2011); CO2 availability to Rubisco can be affected by the same factors as mesophyll conductance and by the physical distribution of carbon assimilation specialized cells (Gago et al. 2016); and ATP synthesis through light reactions may be altered, for instance, by chloroplast arrangement (Xiao et al. 2016). Furthermore, for C4 species under drought, photosynthesis may be limited by Rubisco efficiency and not by the primary CO2 fixating enzyme, the phosphoenolpyruvate carboxylase (PEPcase) (Ghannoum 2009). This fact is supported by reports of increased CO2 leakiness from bundle sheath cells to the mesophyll, highlighting a decrease in the activity ratio of C3/C4 cycle enzymes under drought (Saliendra et al. 1996; Lopes et al. 2011).

The carbon concentrating mechanisms in C4 can greatly attenuate the impact of drought in carbon metabolism, mainly due to an improved stomata control and consequent increase in water use efficiency and maintenance of lower photorespiratory rates (Ghannoum 2009). However, it is somehow paradoxical that, geographically, decreases in rainfall appear to cause a decline in the abundance of C4 relative to C3 species (Ripley et al. 2007), and that severe drought conditions may lead to the loss of the above-mentioned advantages (Taylor et al. 2014). This emergent pattern of C4 species under severe drought may occur due to low phenotypic plasticity, in general or for specific traits, when compared to C3 species.

Thus, the objective of the present research was to evaluate the influence of water shortage in the phenotypic plasticity of anatomical characteristics of grasses with C3, C3 proto-Kranz and C4 photosynthetic types, and its impact in carbon fixation through photosynthesis.

2 Materials and methods

2.1 Plant material and growing environment

The study was performed in a glasshouse in the Plant Physiology sector of the Department of Biology of the Federal University of Lavras, Lavras, MG, Brazil. The experiment was carried out from May to October, 2018. Four genotypes from three species of the Poaceae family, with contrasting photosynthetic metabolism, were chosen: two Oriza sativa (C3) cultivars, Soberano and Douradão, Homolepis isocalycia (G. Mey) (C3 Proto-Kranz) and Andropogon gayanus Kunth var. bisquamulatus (Hochst) Hack. (C4). From the two O. sativa cultivars, Soberano is qualified as a drought-sensitive genotype, while Douradão is recognized as drought-tolerant (Guimarães et al. 2016).

Andropogon gayanus and H. isocalycia were grown from rhizomes obtained, respectively, from the Department of Animal Sciences of the Federal University of Lavras and collected from the Serra do Cipó region (latitude 19.17 S and longitude 43.71 W, altitude of 1194 m above sea level). Seeds for the rice cultivars were obtained from the Department of Agriculture of the Federal University of Lavras. Rhizomes/seeds were planted/sown to 4 L pots (15 cm diameter and 35 cm height), in a sand substrate and without pre-germination treatment. The plant fertigation was performed fortnightly with 111.42 g L−1 NPK (04-14-08) and 27.6 g L−1 CH4N2O (urea) with supplementation of 3.11 g L−1 Monoammonium phosphate (MAP), 8.87 g L−1 MgSO4, 11.48 g L−1 Ca(NO3)2 and 3.63 g L−1 KNO3.

The glasshouse conditions were monitored using a thermo hygrometer coupled to a datalogger (Extech Instruments, model RHT10): air temperatures and air relative humidity were measured hourly, during the experimental period. The average day/night air temperature during the experiment was 26.8 ± 4.5 °C and 15.7 ± 2.3 °C, respectively and the average day/night air relative humidity was 58.1 ± 14.4% and 87.4 ± 3.0%, respectively (average ± standard deviation).

2.2 Water treatments and experimental design

The substrate water holding capacity (SWHC) was calculated by the gravitational method and determined in 0.21 m3 m−3. The pots filled with substrate were irrigated until saturated, covered with foil and left to drain for 12 h. Then, they were weighted to determine the weight at maximum water holding capacity (100% of water content). After that, three water treatments were imposed and kept during the experimental period: high (~ 100% SWHC) intermediate (~ 75% SWHC) and low (~ 50% SWHC) water availability. The treatments were set by daily weighting of the pots using a scale (Elgin Excel S/A) and additional irrigation until completion of the specific weight for each treatment. The substrate moisture, presented as a percentage of water holding capacity, was calculated from the weighing of pots previous to irrigation is presented for specific days, randomly chosen, to compare the imposition of treatments to each species (Fig S1).

The experiment was assembled in a completely randomized design in a 4 × 3 factorial arrangement: four genotypes and three water treatments, with 10 replicates, totalizing 120 experimental units. Each experimental unit was composed by two plants for rice and one plant for the other two species, aiming to homogenize the number of tillers per pot, as rice usually presents half of the tillers per plant compared to the other two species (5 tillers plant−1 for O. sativa and 10–12 tillers plant−1 for H. isocalycia and A. gayanus, as defined in a pre-test).

2.3 Gas-exchange experiments

Fully expanded, young and free of injuries leaves of each treatment were analysed for their gas exchange, at 20–28 days after the imposition of the treatments. An Infra-Red Gas Analyzer (IRGA, LiCor LI-6400-XT) was used for performing A-Ci (response curves for net photosynthesis (A) at different levels of intercellular CO2 concentration (Ci)) and A-PAR (response curves for net photosynthesis (A) at different levels of photosynthetic active radiation (PAR)) curves. Previous to the gas-exchange evaluation, each plant was acclimated under controlled light conditions of 500 µmol m−2 s−1 for 45 min, for photosynthesis induction. A-Ci curves were performed at leaf temperature of 28 °C and PAR (measured in PPFD, photosynthetic photon flux density) of 1000 µmol m−2 s−1, going through the following Ca (atmospheric CO2 concentration) levels: 400, 300, 200, 100, 75, 50 and 25, 400, 600, 800 and 1000 µmol CO2 mol−1. The light intensity was defined in pre-tests aiming to deliver saturating, but non-inhibitory, light intensity at the leaf level. The saturation can be confirmed by the plateau reached for net photosynthesis, at 1000 µmol m−2 s−1, in A-PAR curves, for all the species and water conditions (Fig. 1). This method was used for the O. sativa and H. isocalycia species and, for the A. gayanus species, the level of 0 µmol CO2 mol−1 was also added to the curve. A-PAR curves were performed at leaf temperature of 28 °C and Ca of 400 µmol CO2 mol−1, going through the following PAR levels: 2000, 1000, 500, 250, 120, 60, 40,20, 10 and 0 µmol photons m−2 s−1, for all the species.

Curves were fitted and limiting points for compensation and saturation were calculated for each of the replicates in both curves (Dubois et al. 2007). For the A-Ci curves, the CO2 compensation point (CCP) was calculated as the interception of the Ci axes (at A = 0) for a linear regression of the points below 100 µmol CO2 mol−1; for A-PAR curves, the method was similar with the light compensation point (LCP) being calculated for the interception of the PAR axes (at A = 0) for a linear regression of the points below 120 µmol photons m−2 s−1. The saturation points for CO2 and light (CSP and LSP, respectively) were calculated based on the maximum photosynthesis and compensation points of each curve as per (Dubois et al. 2007; Lobo et al. 2013). Maximum photosynthesis for A-Ci and A-PAR curves (AmaxC and AmaxL, respectively) was the maximum A value reached during the gas-exchange experiment.

2.4 Anatomical evaluation

A fully expanded, young and free of injuries leaf was collected from three individuals of each genotype and water treatment in a total of 36 leaves, at the end of the experimental period, 32 days after the imposition of the water treatments. For the cross-sections, the central part of each leaf (5 mm in length) was fixed in 1% paraformaldehyde (v/v), 1% glutaraldehyde (v/v) and 0.05 M sodium cacodylate buffer, dehydrated through a graded series of ethanol and further blocked in London Resin White (Voznesenskaya et al. 2013). The blocked tissue was cross-sectioned at 2–3 μm on a rotative ultramicrotome UC7 coupled with a stereoscope Leica M80 and stained with 1% toluidine-blue. Three cross-sections of each sample (leaf) were photographed and five images were obtained, in a total of 15 images from each sample. From the images, the following parameters were obtained: interveinal distance (IVD), interveinal cell quantity (CQ), mesophyll width (MW), leaf total area (TA), upper epidermis width (UE), lower epidermis width (LE), epidermis area (EA), mesophyll cells area (MS), bundle sheath cells area (BS), MS:BS ratio (mesophyll:bundle sheath ratio), intercellular space area (ICS), bulliform cells area (BL), schlerenchyma area (SC), xylem area (XL) and phloem area (PH). Some parameters were calculated as a percentage of the total area (TA): percentage MS (pMS), percentage BS (pBS), percentage ICS (pICS), percentage BL (pBL) and percentage SC (pSC) (Dengler et al. 1994).

For the paradermal sections, in each leaf sample, two areas of 1 cm2 of the central area of the leaf were fixed in FAA solution (formalin (3.7%), acetic acid (5%) and alcohol (ethanol, 53%)). After that, the paradermal sections imprints were transferred to cyanoacrylate ester (Super Bonder®), obtaining semi-permanent laminas (Wilson et al. 1981). From each area, three images were obtained using a photomicroscope, in a total of six images per sample. The stomata number (SN) was calculated from the images.

The analysis of the images was performed using the ImageJ software (version 1.52 a). The values for each parameter were the averages of the total individual measurements, being 15 for cross-sectional and 6 for paradermal sections. Those analyses were performed in the Plant Anatomy and the Electronic Microscopy and Ultrastructural Analysis Laboratories at Federal University of Lavras.

2.5 Phenotypic plasticity analysis

Phenotypic plasticity was calculated for the anatomical and gas-exchange parameters using the Relative Distance Plasticity Index, as defined by Valladares et al. (2006). RDPI was calculated for two combination of water treatments: moderate water restriction (100–75% SWHC) and severe water restriction (100–50% SWHC) for each parameter and genotype. It is formed as the average of nine values, from the combination of three replicates in each water level (Eq. 1).

$${\rm{RDPI}}_{\rm{p}}=\frac{\left(\sum \left(\frac{\left|{\rm{p}}_{\rm{n}}-{\rm{p}}_{\rm{m}}\right|}{{\rm{p}}_{\rm{n}}+{\rm{p}}_{\rm{m}}}\right)\right)}{(\rm{n}.\rm{m})},$$
(1)

where RDPIp is the plasticity index for the parameter p, pn and pm are the values of the parameter for the observations at the higher and lower water level, respectively with n and m varying from 1 to 3 (number of replicates).

The trend of increase or decrease in each parameter (Trp), between the water levels, was calculated as the percentage of differences between the averages of the parameters in the two specific water treatments. The percentage of differences was calculated as the difference between the average of the parameter at the lower and the higher water level divided by the parameter at the higher water level (Eq. 2).

$${\rm{Tr}}_{\rm{p}}\left(\rm{\%}\right)= \sum \left(\frac{\overline{{{\rm{p}}_{\rm{m}}}}- \overline{{{\rm{p}}_{\rm{n}}}}}{\overline{{{\rm{p}}_{\rm{n}}}}}\right)\cdot 100,$$
(2)

where \(\overline{{{\rm{p}}_{\rm{m}}}}\) and \(\overline{{{\rm{p}}_{\rm{n}}}}\) are the averages of the parameter for the lower and higher water levels, respectively. Positive values for Trp means that the plants at lower water levels presented higher values for the specific parameter than the plants at higher water level, and the other way around for negative values.

2.6 Statistical analysis

For the gas-exchange parameters, the average values for each treatment combination (species/water level) for net photosynthesis (A), intercellular CO2 concentration (Ci) and radiation (PAR) levels were used to build the graphs of the A-Ci and A-PAR curves. The compensation and saturation points and maximum net photosynthesis for the mentioned curves were analysed for normality through the Shapiro–Wilk test. Analysis of variance was performed and the significant results were compared by the post-hoc Tukey test. Boxplots were generated for the parameters. The analysis was performed in the R software (version 3.6.2) using the packages ggpubr, agricolae and stats. For the anatomical parameters, RDPI and Trends, the same approach was used for normality, analysis of variance and post-hoc tests.

A principal component analysis (PC) was performed combining the anatomical traits in each treatment combination (species/water level). For the PCs, the data was centred and scaled to account for differences in magnitude of the parameters. The PCs and the graph were performed using the R software (version 3.6.2) and the packages FactoMineR, stats, devtools, lifecycle, ggplot2 and ggfortify.

Different significance levels (p) were considered for the multiple tests: for the Shapiro–Wilk normality test and Tukey post-hoc tests, statistical differences were considered significant under 0.05 and for the ANOVAS, the significance level was 0.1. This decision is based on the character of the genotypes and experimental conditions and on the fact that significance levels may be chosen, inside considered limits, by the scientific context of the research, as recommended by Fisher (1950).

3 Results

3.1 Gas-exchange analysis

For the A-Ci curves at 100% of SWHC, the A. gayanus presented a typical C4 pattern with lower compensation and saturation CO2 points, although the values of maximum photosynthesis were not significantly higher when compared to the other photosynthetic types (Table 1). The two O. sativa cultivars and the H. isocalycia performed very similarly for the curves and analysed parameters (Fig. 1a, Table 1). At 75% of SWHC, the shapes of the curves did not greatly differ from the 100% treatment (Fig. 1b), with A. gayanus presenting lower compensation and saturation CO2 points than the others (Table 1). However, in this treatment, the O. sativa cultivars presented lower saturation points than the H. isocalycia species (Table 1). In this case, the water treatment has impacted in the maximum photosynthesis with higher values for A. gayanus than to O. sativa cv. Soberano and H. isocalycia. For the 50% treatment, the shape of the curves changed drastically in relation to the other treatments (Fig. 1c), although the pattern of lower CO2 compensation and saturation points for A. gayanus was kept. No significant differences were observed for maximum photosynthesis (Table 1).

Table 1 Analysis of variance and post-hoc tests’ results for anatomical and gas-exchange parameters of four genotypes, O. sativa cv. Soberano (OsS), O. sativa cv. Douradão (OsD), H. isocalycia (Hi) and A. gayanus (Ag) at 100, 75 and 50% of substrate water holding capacity (SWHC)
Full size table
Fig. 1
figure 1

Net photosynthesis response to intercellular carbon (A-Ci; a–c) and light (A-PAR; d–f) at 100, 75 and 50% of substrate water holding capacity (SWHC) for four genotypes, O. sativa cv. Soberano (light blue circles/dotted line), O. sativa cv. Douradão (dark blue triangles/short dashed line), H. isocalycia (green diamonds/long dashed line) and A. gayanus (orange squares/solid line). (Color figure online)

Full size image

The behaviour of the genotypes under contrasting water treatments was less affected for the A-PAR curves. H. isocalycia presented lower light compensation point than A. gayanus at the 100% treatment, but no statistical difference was observed for the other two water levels (Fig. 1d–f and Table 1). Also, O. sativa cv. Douradão presented lower light saturation point than A. gayanus (Fig. 1-d–f and Table 1). For the maximum photosynthesis, different patterns were observed for each of the water levels. At 100%, A. gayanus reached the higher maximum photosynthesis levels, followed by O. sativa cv. Soberano (Fig. 1d and Table 1). For the 75% and 50%, O. sativa cv. Douradão presented higher values than H. isocalycia (Fig. 1e, f and Table 1).

3.2 Anatomical parameters

For the anatomical parameters, not all the measured traits are presented in the results. MS, BS, ICS, BL and SC presented very similar behaviour to their percentages to total area (TA) and only the percentages are presented (pMS, pBS, pICS, pBL and pSC), together with the other measured parameters IVD, CQ, MW, LE, UE, XL, PH and SN.

The PC approach was used firstly, allowing a global view of the genotypes’ behaviour and can be analysed in combination to the ANOVA results (Table 1). The first two principal components explained around 80% of the total variance. The PC has split the genotypes in clear groups: PC1 has differentiated the O. sativa cultivars from the H. isocalycia and A. gayanus, forming a group at the left-hand side of the PC1 axes. The two last genotypes were differentiated, mainly, by the PC2 (Fig. 2). PC1 was strongly negatively correlated to CQ and pMS and positively correlated to pICS. This represents that the O. sativa plants tend to present higher percentage of mesophyll cells and more cells in the interveinal space than the H. isocalycia and A. gayanus, while the two lasts tend to present bigger intercellular spaces. The pBS vector is directed to the H. isocalycia grouping and in opposition to O. sativa cultivars, meaning that the first shows consistent higher values for bundle sheath cells area than the second (Fig. 2 and Table 1). H. isocalycia also presented higher values for pBS than A. gayanus at low water availability (75 and 50% SWHC). The previous reported behaviours also explain the strong positive relation between MS:BS and the O. sativa cultivars. IVD vector is positioned against the A. gayanus grouping showing that A. gayanus tend to present lower IVD than all the others. In opposition, pBL vector positioning toward the bottom right quadrant shows that A. gayanus presented higher values for this trait than the others. Generally, it also happened for LE and UE, but those differences were not significant for all the water treatments (Table 1).

Fig. 2
figure 2

Principal component analysis (PC) of anatomical parameters of four genotypes, O. sativa cv. Soberano (OsS), O. sativa cv. Douradão (OsD), H. isocalycia (Hi) and A. gayanus (Ag) at 100, 75 and 50% of substrate water holding capacity (SWHC). IVD interveinal distance; CQ interveinal cell quantity; MW mesophyll width; pMS percentage mesophyll area; pBS percentage bundle sheath area; MS:BS mesophyll–bundle sheath areas ratio; pICS percentage intercellular spaces area; pBL percentage bulliform cells area, pSC percentage sclerenchyma cells area; LE lower epidermis area; UE upper epidermis area; XL xylem area; PH phloem area; SN stomata number

Full size image

The grouping of O. sativa cultivars in the PC is explained by the lower differences in specific anatomical parameters between them: at 100% of SWHC they did not differ for any anatomical parameters, at 75% they only differ for SN and for 50% only for LE area.

Both O. sativa cultivars tended to present, in general, higher areas of xylem and phloem than H. isocalycia and, in some cases, than A. gayanus, at water shortage 75 and 50%. O. sativa cv. Douradão presented higher SN than O. sativa cv. Soberano and H. isocalycia at 75 and 50% SWHC, respectively (Table 1).

3.3 Phenotypic plasticity analysis

The RDPI was calculated to better understand the short-term impact of the water regimes in the gas-exchange and phenotypic arrangement of leaf structural parameters. Plasticity was analysed for each parameter in two scenarios: moderate water restriction (from 100 to 75% SWHC) and severe water restriction (from 100 to 50% SWHC), with the results presented in Table 2.

Table 2 Analysis of variance and post-hoc tests’ results of the RDPI of anatomical and gas-exchange parameters of four genotypes, O. sativa cv. Soberano (OsS), O. sativa cv. Douradão (OsD), H. isocalycia (Hi) and A. gayanus (Ag) between environments of 100–75 and 100–50% of substrate water holding capacity (SWHC)
Full size table

For gas-exchange parameters calculated from A-Ci curves, significant plasticity was only observed for the CCP, with higher plasticity for A. gayanus than the others, in the moderate and severe water restriction scenarios (Table 2). For the parameters originated from A-PAR curves, plasticity was observed for the LCP at the two water scenarios. At moderate water restriction, A. gayanus was the most plastic, followed by H. isocalycia, O sativa cv. Soberano and O. sativa cv. Douradão. At severe water restriction, O. sativa cv. Douradão presented higher plasticity than the others. The same pattern of higher plasticity of O. sativa cv. Douradão was also observed for LSP at the two water scenarios and for AmaxL at the moderate water restriction (Table 2).

Regarding the anatomical parameters, in a scenarios of moderate water restriction (from 100 to 75% SWHC), H. isocalycia presented consistent higher plasticity levels than the others for multiple parameters, as pMS. It was more plastic than O. sativa cv. Douradão e A. gayanus for CQ and PH and than O. sativa cultivars for pBS and MS:BS. It was also more plastic than O. sativa cv. Douradão for IVD. On the other hand, A. gayanus was more plastic than the O. sativa cultivars for pSC and than O. sativa cv. Soberano for UE. Finally, O. sativa cv. Douradão was more plastic than O. Sativa cv. Soberano for pICS (Table 2).

Under severe water restriction (from 100 to 50% SWHC), some of the plasticity patterns were kept when compared to moderate water restriction as for pMS, pBS and MS:BS. H. isocalycia maintained higher plasticity than others for multiple parameters but with different behaviours: it was more plastic than all the other for XA and than A. gayanus and O. sativa cv. Douradão for SN and CQ. Also, H. isocalycia and A. gayanus were more plastic than O. sativa cv. Soberano for pSC. A. gayanus presented higher plasticity than H. isocalycia and O. sativa cv. Douradão for MW and than H. isocalycia and O. sativa cv. Soberano for UE. It was also more plastic than O. sativa cv. Douradão for IVD and than O. sativa cv. Soberano for pSC. Finally, O. sativa cv. Douradão was more plastic than O. sativa cv. Soberano and H. isocalycia for pICS and than all the others for LE. O. sativa cv. Soberano presented higher plasticity than Douradão and A. gayanus for pBL (Table 2).

4 Discussion

The present research focus on four plant genotypes from three species of grasses, with contrasting photosynthetic types, under restricting water treatments. Different phenotypic plasticity patterns for leaf anatomical traits as a response to water constraints and its impact in carbon fixation are presented. In overview, C4 plants are recognized to be more tolerant to drought than C3, mainly due to the compensation of stomatal limitation by carbon concentration mechanism and re-fixation of photorespired CO2 (Ghannoum 2009). However, this behaviour is not as clear in long-term or severe drought events (Taylor et al. 2011). In general, the gas-exchange results of this study are in agreement with both above-mentioned patterns: C4 advantages at moderate water deficit and lost of the advantage in more severe limitation. This is confirmed by A. gayanus gas-exchange trends of higher photosynthesis at 100 and 75% SWHC with a considerable reduction at 50% SWHC (Fig. 1). Those results are discussed in terms of the specific impacts of single anatomical traits and their combined arrangement to influence photosynthesis in the different plant genotypes and water availability levels.

Considering the arrangement of vascular tissues and mesophyll number of cells, A. gayanus consistently presented lower values for IVD and CQ in all the water levels (Table 1 and Fig. 3), what goes in accordance to Sage and McKown (2006). This is a typical C4 leaf cell arrangement aiming to potentialize the transport of metabolites between carbon assimilation and reduction tissues, as part of the carbon concentrating mechanism (Dengler et al. 1994; Lundgren et al. 2014). Considering that bundle sheath cells, which surround the vascular tissues, are the main location of carbon reduction in C4 plants, the increase of numbers of vessels and consequent reduction of mesophyll cells is normally observed, explaining the decreased CQ and IVD (Sage et al. 2011). The C4 pattern of decreasing IVD and CQ is normally related to decreases in MS:BS (Carmo-Silva et al. 2009) what can also be observed in the present study (Table 1). Higher plasticity was observed for CQ than for IVD, meaning that plants tend to change cell number instead of distance between vessels under water constraint (Table 2 and Fig. 4). However, the plasticity and trends of changes for these parameters did not help to explain the gas-exchange patterns, what may represent that they may not be a target trait for the understanding of C3 and C4 short term acclimation to drought. This may be due to the fact that only for C4 species, plasticity of CQ and IVD may be related to changes in gas-exchange patterns, as they are the only species, among the studied, that perform carbon fixation on bundle sheath cells.

Fig. 3
figure 3

Transverse leaf sections O. sativa cv. Soberano (a, e, i) and cv. Douradão (b, f, j), H. isocalycia (c, g, k) and A. gayanus (d, h, l) at 100% (a–d), 75% (e–h) and 50% (i–l) of substrate water holding capacity (SWHC). Bars = 50 µm

Full size image
Fig. 4
figure 4

Trends of traits for multiple anatomical parameters at 100, 75 and 50% of substrate water holding capacity (SWHC) for four genotypes, O. sativa cv. Soberano (light blue circles/dotted line), O. sativa cv. Douradão (dark blue triangles/short dashed line), H. isocalycia (green diamonds/long dashed line) and A. gayanus (orange squares/solid line). pMS percentage mesophyll area (pMS, a); pBS percentage bundle sheath area (pBS, b); mesophyll width (MW, c); percentage intercellular spaces area (pICS, d); percentage bulliform cells area (pBL, e); percentage sclerenchyma cells area (pSC, f); lower epidermis area (LE, g); upper epidermis area (UE, h); stomata number (SN, i). On tables, p values are relative to the ANOVA of RDPI comparing species in each water treatment. The values are the trends calculate for moderate and severe water limitations (Eq. 2). (Color figure online)

Full size image

Other parameters phenotyped in this study are normally related to physiological adjustments to water shortage and impact in gas-exchange. For instance, the balance between MS and BS is linked to photosynthetic capacity as those tissues are the place of carbon fixation by Rubisco in C3 and C4 species respectively (Carmo-Silva et al. 2009). Presence of BL and SC cells may also impact in mechanical sustentation of leaves and the leaf rolling phenotyping, impacting in light interception (Monteith 1977) and photosynthesis (Xu et al. 2018; Cal et al. 2019). Moreover, higher ICS are related to a strategy to increase the exposition of chloroplasts in the mesophyll cells to intercellular spaces (Terashima et al. 2011), where CO2 is transported, increasing mesophyll conductance (Ouyang et al. 2017). This is a highly wanted trait, mainly under water stresses, where stomata conductance may be limited, and is correlated to higher photosynthetic rates per leaf area (Lawlor 2002). Epidermic cells are natural barriers avoiding excessive transpiration and desiccation of inner leaf tissues and their widths impact in the capacity of plants to support drought (Bolhar-Nordenkampf 1987). SN is a key parameter considering the control of gas-exchanges and water use efficiency in plants, with greater impact under water limitation (Flexas et al. 2004). Finally, XL and PH areas impact in the possibility of cavitation and failure, respectively, which could lead to plant death under drought (Tyree and Sperry 1989; Sevanto 2014). Thus, the plasticity and trends of the anatomical parameters, when analysed in combination, helps to explain the gas-exchange patterns observed in this study. However, not all the mentioned traits have impacted directly to every plant species and their impacts are specifically discussed for each species.

Homolepis isocalycia presented a pattern of maintaining maximum photosynthesis in moderate and severe water limitation (Fig. 1). This can be confirmed by its low plasticity for AmaxL (Table 2). It has consistently decreased pMB and increased pBS (consequently decreasing MS:BS), without major changes in MW. It also has decreased SN, potentially limiting CO2 transportation to the inner parts of the leaves, but controlling transpiration (Fig. 4). Photosynthetic levels at lower water availability may have been kept in H. isocalycia by improving its leaf water status, by decreasing stomata number and investing in BS cells, which recognizably act in water regulation in C3 and C3-intermediate plants (Leegood 2008). Avoiding a sharp decrease in leaf water status may have impact in maintaining ATP synthesis, Rubisco activity and not limiting stomatal conductance (Farooq et al. 2009). Although BS cells are not photosynthetically active in C3 proto-Kranz, the pattern of increased area of this tissue is a crucial preparatory adaptation for the C4-types emergence, mainly considering the role of glycine metabolization, which is fundamental (Sage et al. 2014; Lundgren et al. 2016; Schulze et al. 2016).

On the other hand, A. gayanus presented a clear decrease in AmaxC and AmaxL at 50% SWHC treatment (Fig. 1). At moderate water limitation, this C4 species decreased the MW and pBS, but had lower impact in pMS. The decreases in pBS, the final carbon fixation tissue in C4 plants, could have greatly impacted in photosynthesis, what was not observed. This may be a result of a combined maintenance of pMS and pICS areas and increases in SN, ensuring ideal conditions to PEPcase through non-limiting stomatal and mesophyll conductances. Also, increases in pBL and pSC may be related to mechanical sustentation of plants and to avoiding leaf rolling, mainly when water potential is decreased (Xu et al. 2018; Cal et al. 2019), playing an important role in light capture and photosynthesis potential (Fig. 4). At severe water limitation, a similar decrease in pBS and maintenance in pMS was observed. However, the SN was not as big as at 75% SWHC, probably to avoid excess transpiration and pICS was also decreased. These last phenotypes may have limited the availability of CO2 to PEPcase by diminishing stomata conductance, if compared to 75% condition, and by decreasing the exposition of chloroplasts in the MS cells to intercellular spaces, where CO2 is present (Terashima et al. 2011). It may also have reduced efficiency of carboxylation of Rubisco and its relative content due to reduction in pBS (Ghannoum 2009). This combination of traits may have caused a phenotype of lower photosynthetic rates by combined limitation of carbon fixation by PEPcase in the mesophyll and further lower capacity of carbon reduction in the decreased area of BS cells.

When comparing O. sativa cultivars Soberana and Douradão and their respective susceptibility and resiliency to drought, the differences in photosynthetic performance under water limitation (Fig. 1) cannot be easily explained by single anatomical parameters itself, as they only statistically differ on stomata characteristics at 75% SWHC and on lower epidermis area at 50% SWHC. However, when analysing plasticity, there are many more characteristics that could potentially explain their behaviour at lower water treatments. The drought-sensitive O. sativa cv. Soberano did not present a drastic reduction in carbon uptake under moderate water limitation, with lower plasticity for maximum photosynthesis (Fig. 1 and Table 2). Apparently, it is not related the MW and the balance of BS and MS areas. This may be related to a better water use efficiency due to reduced SN. However, at severe water limitation, the reduction in maximum photosynthesis is more pronounced (Fig. 1). At lower water level, the stomata conductance may have suffered a quick and sharp decrease, being compensated by the increase in SN. The increased SN in combination to increase in pBL, avoiding leaf roll, may have caused a high transpiratory demand, closing stomata due to feedback control. With reduced gas-exchange rate and lower pICS, higher photorespiration rates may have been induced, decreasing net photosynthesis (Fig. 4).

Inversely, the drought-tolerant O. sativa cv. Douradão presented a group of characteristics which could explain its gas-exchange behaviour. At moderate water limitation it has greatly increased MW, a very common phenotype for plants under drought. The increase in MW may have powered the number of MS cells and carbon capture (Fig. 4). On the other hand, at severe water limitation, this MW changes were not observed and other strategies may have been assumed. For instance, increases in LE and UE may have avoided excessive transpiration. Also, the increases in ICS may have increased mesophyll conductance and CO2 levels to the Rubisco site (Gago et al. 2020), avoiding photosrespiration to outstand, mainly during low water availability (Tholen et al. 2012). The potential increase in water use efficiency could have allowed the cv. Douradão to escape the necessity of leaf rolling mechanism, as it presents increases in pSC, to keep leaf structure under water shortage. This may have allowed a better exposition to sunlight and atmospheric air, improving photosynthetic rates, as observed in this research.

These different combinations of traits for the two O. sativa cultivars have greatly impacted in their photosynthetic levels, mainly under limited water. This behaviour may reflect in growth and potential productivity and can help to explain their qualification as sensitive/tolerant to drought, as yield under water constraints is a key component of most of drought-tolerance indices (Svoboda et al. 2016), including the rice-specific ones (Raman et al. 2012). These responses may also be related to specific combination of genes of interest, related to abiotic stress responses, in the drought-tolerant cultivar, as also reported for other rice materials under environmental constrains (Paul and Roychoudhury 2019).

We could observe that the C4, A. gayanus, was greatly affected at higher water limitation, with decrease in photosynthesis at A-Ci and A-PAR curves. Although C4 plants evolved a more efficient carbon pathway than C3 (Bräutigam and Gowik 2016) they are recognized to have a slower acclimation to environmental changes than C3 (Sage and McKown 2006). These impacts in the photosynthetic capacity and efficiency (Leegood and Walker 1999), mainly due to the difficulty to coordinate the mesophyll-bundle sheath interaction, what can cause substantial CO2 leakage from the bundle sheath cells (von Caemmerer and Furbank 1999). In general terms, C4 photosynthesis tends to be less plastic than C3 photosynthesis, under environmental constraints (Sage and McKown 2006) what explains their qualification in C4/specialist and C3/generalist. This behaviour is not fully supported in this study as A. gayanus presented high plasticity for a number of traits. However, by analysing A. gayanus plasticity at moderate and severe water limitations, we can observe that most of the changes observed at 75% are similar to the ones observed at 50% SWHC (Fig. 4). This may represent a limitation of the C4 species to be further plastic in higher water limitation, what partially supports its qualification as specialist. This may also help to explain the fact that C4 species may lose their evolutive-acquired advantages in long-term or severe drought events (Taylor et al. 2011).

For the C3 proto-Kranz, H. isocalycia, the photosynthetic response in A-Ci curves and carbon saturation and compensation and maximum photosynthesis was very similar to the O. sativa cv. Douradão at all water levels (Fig. 1 and Table 1). However, for the A-PAR curves, H. isocalycia presented lower photosynthetic rate than the other genotypes (Fig. 1). Anatomically, C3 proto-Kranz presented a clear intermediate phenotype between C3 and C4 (Figs. 2, 3). The C3 proto-Kranz photosynthetic type is much closer to the C3 type than to the C4 type. The biggest difference from C3 to C3 proto-Kranz is the presence of mitochondria in the bundle sheath cells of the latest (Sage et al. 2014) which is a crucial step for the further C2 photosynthetic type. We observed that H. isocalycia tended to present higher BS and lower MS than the O. sativa cultivars, and consequently a lower MS:BS. Increases in the bundle sheath area in detriment of mesophyll area, may be related to the necessity to accommodate more cell components, as the mitochondria, in the bundle sheath. However, those bundle sheath cells are not yet functional as photosynthesizing tissues, as in C4 plants, explaining the reduced photosynthetic patterns in all water conditions in A-PAR curves for H. isocalycia. In this specific scenario, light increases would not boost photosynthesis as the main limitation could be related to the carbon fixation, instead.

Studying plasticity of anatomical parameters of C3–C4 intermediates may shed light to new strategies for adapting to water constrains. This is due to the fact that those plants have limitation to be plastic in specific anatomical parameters due to its evolutionary adaptations. For instance, H. isocalycia invests in BS cells that are not photosynthetically functional but find other ways to keep lower, but reasonably constant photosynthesis under water limitation, as reported in this study. This can be observed by the high plasticity of H. isocalycia for many of the studied traits (Table 2).

Finally, focusing in the plasticity of plants with different photosynthetic pathways may unravel possible adaptations to C3 species from the understanding of the intermediates to approach their metabolism to C4 plants mainly under drought stress (Yu 2020). This may require integrated studies in genetics, plant physiology and evolution (Pita-Barbosa et al. 2019), possibly shedding light to other potential approaches for important researches as the ones involving the insertion of C4 metabolism in C3 plants, as, for instance the C4 Rice project (von Caemmerer et al. 2012).