Coordination of Leaf Photosynthesis, Transpiration, and Structural Traits in Rice and Wild Relatives (Genus Oryza)^,,

Linkages of leaf and mesophyll cell traits to CO₂ diffusion, photosynthesis, transpiration, and water use efficiency were identified across accessions of the genus Oryza.

Abstract

The genus Oryza, which includes rice (Oryza sativa and Oryza glaberrima) and wild relatives, is a useful genus to study leaf properties in order to identify structural features that control CO₂ access to chloroplasts, photosynthesis, water use efficiency, and drought tolerance. Traits, 26 structural and 17 functional, associated with photosynthesis and transpiration were quantified on 24 accessions (representatives of 17 species and eight genomes). Hypotheses of associations within, and between, structure, photosynthesis, and transpiration were tested. Two main clusters of positively interrelated leaf traits were identified: in the first cluster were structural features, leaf thickness (Thick_leaf), mesophyll (M) cell surface area exposed to intercellular air space per unit of leaf surface area (S_mes), and M cell size; a second group included functional traits, net photosynthetic rate, transpiration rate, M conductance to CO₂ diffusion (g_m), stomatal conductance to gas diffusion (g_s), and the g_m/g_s ratio. While net photosynthetic rate was positively correlated with g_m, neither was significantly linked with any individual structural traits. The results suggest that changes in g_m depend on covariations of multiple leaf (S_mes) and M cell (including cell wall thickness) structural traits. There was an inverse relationship between Thick_leaf and transpiration rate and a significant positive association between Thick_leaf and leaf transpiration efficiency. Interestingly, high g_m together with high g_m/g_s and a low S_mes/g_m ratio (M resistance to CO₂ diffusion per unit of cell surface area exposed to intercellular air space) appear to be ideal for supporting leaf photosynthesis while preserving water; in addition, thick M cell walls may be beneficial for plant drought tolerance.

Leaves have evolved in different environments into a multitude of sizes and shapes, showing great variation in morphology and anatomy (Evans et al., 2004). However, all leaf typologies share common functions associated with chloroplasts, namely to intercept sunlight, take up CO₂ and inorganic nitrogen, and perform photosynthesis as a primary process for growth and reproduction.

Investigating relationships between leaf anatomy and photosynthetic features (CO₂ fixation, which involves physical and biochemical processes and loss of water by transpiration) could lead to the identification of structural features for enhancing crop productivity and improve our understanding of plant evolution and adaptation (Evans et al., 2004).

Stomata, through which CO₂ and water vapor diffuse into and out of the leaf, are involved in the regulation and control of photosynthetic and transpiration responses (Jarvis and Morison, 1981; Farquhar and Sharkey, 1982). Besides stomata distribution patterns between the abaxial and adaxial lamina surfaces (Foster and Smith, 1986), stomatal density and size are leaf anatomical traits contributing to build the leaf stomatal conductance to gas diffusion (g_s). This is calculated as the reciprocal of the stomatal resistances to gas diffusion; stomatal control results in a lower concentration of CO₂ in the leaf mesophyll (M) intercellular air space (C_i) than in the atmosphere (C_a; Nobel, 2009).

Leaf M architecture greatly contributes to the pattern of light attenuation profiles within the lamina (Terashima and Saeki, 1983; Woolley, 1983; Vogelmann et al., 1989; Evans, 1999; Terashima et al., 2011) and affects CO₂ diffusion from the intercellular air space (IAS) to the chloroplast stroma. Therefore, it influences photosynthetic activity (Flexas et al., 2007, 2008) and can have effects on leaf hydrology and transpiration (Sack et al., 2003; Brodribb et al., 2010; Ocheltree et al., 2012). In addition, M architecture sets boundaries for leaf photosynthetic responses to changing environmental conditions (Nobel et al., 1975).

Fortunately, several methodologies are currently available (Flexas et al., 2008; Pons et al., 2009) to determine M conductance to CO₂ diffusion (g_m), expressed per unit of leaf surface area. It is calculated as the reciprocal of the cumulated partial resistances exerted by leaf structural traits and biochemical processes from the substomatal cavities to photosynthetic sites (Evans et al., 2009; Nobel, 2009). The resistance to CO₂ diffusion in the liquid phase is 4 orders of magnitude higher than in the gaseous phase (Nobel, 2009); therefore, the changes in CO₂ concentration in the leaf gas phase are small in comparison with the changes in the liquid phase (Niinemets, 1999; Aalto and Juurola, 2002; Nobel, 2009). In the liquid phase, the resistance to CO₂ transfer is built from contributions by the cell walls, the plasmalemma, cytoplasm, chloroplast membranes, and stroma (Tholen and Zhu, 2011; Tholen et al., 2012); in addition, it involves factors associated with the carboxylation reaction (Kiirats et al., 2002; Evans et al., 2009). Thus, the concentration of CO₂ in the chloroplasts (C_c) is lower than C_i and can limit photosynthesis.

At steady state, the relationships between the leaf net photosynthetic rate (A), the concentrations of CO₂, and the stomatal conductance to CO₂ diffusion (g_{s_CO2}) and g_m are modeled based on Fick’s first law of diffusion (Nobel, 2009) as:

(1)

where C_a, C_i, and C_c are as defined above (Flexas et al., 2008).

The magnitude of g_m has been found to correlate with certain leaf structural traits in some species, in particular with the M cell surface area exposed to IAS per (one side) unit of leaf surface area (S_mes) and its extent covered by chloroplasts (S_chl; Evans and Loreto, 2000; Slaton and Smith, 2002; Tholen et al., 2012). From a physical modeling perspective, increasing S_mes provides more pathways acting in parallel for CO₂ diffusion (to and from the chloroplasts) per unit of leaf surface area; thus, it tends to reduce the resistance to CO₂ movement into the M cells and to increase g_m (Evans et al., 2009; Nobel, 2009). A number of leaf structural traits affect S_mes, including leaf thickness, cell density, cell volume and shape, and the fraction of the M cell walls in contact with the IAS (Terashima et al., 2001, 2011), and the degree they are linked to S_mes can vary between species (Slaton and Smith, 2002; Terashima et al., 2006). In particular, the presence of lobes on M cells, which are prominent in some Oryza species, may contribute to g_m through increasing S_mes (Sage and Sage, 2009; Terashima et al., 2011; Tosens et al., 2012). The M cell wall can provide resistance in series for M CO₂ diffusion (Nobel, 2009); thicker cell walls may increase resistance to CO₂ movement into the M cells and decrease g_m (Terashima et al., 2006, 2011; Evans et al., 2009).

Other leaf traits, such as M porosity (the fraction of M volume occupied by air spaces [Vol_IAS]), has been shown to have a positive correlation with g_m in some species (Peña-Rojas et al., 2005), but the association may be mediated by light availability (Slaton and Smith, 2002). Leaf thickness (Thick_leaf) tends to be negatively linked to g_m, and it may set an upper limit for the maximum g_m, according to Terashima et al. (2006), Flexas et al. (2008), and Niinemets et al. (2009).

With respect to leaf structural traits and water relations, Thick_leaf may increase the apoplast path length (resistances in series; Nobel, 2009) in the extra-xylem M (Sack and Holbrook, 2006; Brodribb et al., 2007) for water to reach the evaporation sites, which could decrease the conductance of water through the M and lower the transpiration rate. Interestingly, while thicker M cell walls may reduce g_m, they can enable the development of higher water potential gradients between the soil and leaves, which can be decisive for plant survival and longevity under drought conditions (Steppe et al., 2011).

The purpose of this study was to provide insight into how the diversity of leaf structure relates to photosynthesis and transpiration among representative cultivated species and wild relatives in the genus Oryza. This includes, in particular, identifying leaf structural features associated with the diffusion of CO₂ from the atmosphere to the chloroplasts, photosynthesis, transpiration efficiency (A/E), and drought tolerance. The genus consists of 10 genomic groups and is composed of approximately 24 species (the number depending on taxonomic preferences; Kellogg, 2009; Brar and Singh, 2011), including the cultivated species Oryza sativa and Oryza glaberrima. Oryza species are distributed around the world, and they exhibit a wide range of phenotypes, with annual versus perennial life cycles and sun- versus shade-adapted species (Vaughan, 1994; Vaughan et al., 2008; Brar and Singh, 2011; Jagadish et al., 2011). This diversity in the genus is an important resource, which is being studied to improve rice yield, especially under unfavorable environmental conditions. In particular, O. glaberrima, Oryza australiensis, and Oryza meridionalis are of interest as drought-tolerant species (Henry et al., 2010; Ndjiondjop et al., 2010; Scafaro et al., 2011, 2012), while Oryza coarctata is salt tolerant (Sengupta and Majumder, 2010). In this study, a total of 43 leaf functional and structural parameters were collected on 24 accessions corresponding to 17 species within eight genomes (Table I) to represent the spectrum of the leaf diversity in the genus Oryza.

Table I. Oryza accessions used in this study and identified by the assigned numbers.

Leaf functional measurements were performed on all accessions; structural measurements were made on accessions labeled with asterisks (for O. sativa cv IR72 and O. australiensis 22, structural measurements were only made on stomata but not other traits). Genomes for Oryza species are according to Brar and Singh (2011). Life cycle is as follows: A = annual; B = biennial; P = poliennial. Habitat is as follows: S = shade; S-Sh = sun-shade.

Genome	Species	Life Cycle	Habitat	Accession	No.
AA	O. barthii	A	S	PI 590400*	1
AA	O. glaberrima	A	S	PI 450430*	2
AA	O. glumaepatula	P	S	PI 527362*	3
AA	O. longistaminata	P	S	IRGC 101207*	4
AA	O. longistaminata	P	S	IRGC 101754	5
AA	O. meridionalis	A/P	S	IRGC 93265*	6
AA	O. nivara	A/B	S	PI 590405*	7
AA	O. rufipogon	P	S	PI 104640	8
AA	O. rufipogon		S	PI 590421*	9
AA	O. sativa	A	S	IR64*	10
AA	O. sativa	A	S	IR72	11
BB	O. punctata	A	S-Sh	IRGC 105690*	12
BBCC	O. minuta	P	S-Sh	IRGC 101141*	13
CC	O. officinalis	P	S-Sh	PI 59412*	14
CC	O. rhizomatis	P	S	IRGC 101609	15
CC	O. rhizomatis	P	S	IRGC 105950*	16
CCDD	O. alta	P	S-Sh	PI 590398*	17
CCDD	O. latifolia	P	S-Sh	IRGC 100959*	18
CCDD	O. latifolia	P	S-Sh	IRGC 105173	19
EE	O. australiensis	P	S	IRGC 101397*	20
EE	O. australiensis	P	S	IRGC 105277*	21
EE	O. australiensis	P	S	IRGC 86527	22
FF	O. brachyantha	B	S	IRGC 101232*	23
HHKK	O. coarctata	P	S	IRGC 104502*	24

For evaluating aspects of photosynthesis, the model in Equation 1 was considered, and all the listed functional variables, A, g_{s_CO2}, (C_a − C_i), g_m, and (C_i − C_c), were determined. In addition, among the leaf functional traits, the M resistance to CO₂ diffusion per unit of cell surface area exposed to IAS (reciprocal of g_m/S_mes) was calculated as described by Evans et al. (2009): it represents the resistance to CO₂ diffusion from IAS to chloroplasts in a liquid solution through cell wall and membranes (Nobel, 2009). Leaf transpiration rate (E), A/E, the intrinsic A/E (ratio between A and stomatal conductance to water vapor diffusion [g_{s_H2O}]), g_m/g_{s_CO2} (representing the coordination between g_m and g_s), and the carbon isotope composition of leaf biomass (δ¹³C; calculated as ¹³C/¹²C) were determined. The value of δ¹³C has been recognized as a potential indicator of leaf A/E: increased limitations on photosynthesis by decreased g_s can lead to higher A/g_{s_H2O} ratios and less discrimination against assimilation of ¹³CO₂ (for review, see Farquhar et al., 1989); the leaf A/E may also be positively linked to the g_m/g_s ratio (Flexas et al., 2008, 2013; Barbour et al., 2010). With respect to leaf structure, the stomatal density, stomatal pore length, and indices of stomatal pore area on both lamina sides (according to Sack et al., 2003), the Thick_leaf, Vol_IAS, S_mes, S_chl, area of M cell section (a_cell) in leaf cross sections, cell wall thickness (Thick_cw), and M cell surface lobing (Lob_cell) were the principal traits estimated. A statistical multivariate analysis (Child, 2006) was employed to identify clusters of highly interrelated leaf traits; trait-to-trait correlation analysis was carried out to further examine leaf structural, functional, and structural-functional relationships.

The following are the main hypotheses examined in this study. (1) Leaf thickness will be associated with certain M structural features. (2) g_m will be coordinated with M structural traits. (3) A will be correlated with g_s, g_m, and E. (4) Leaf structural traits will be involved in the relationship between A and E, which will affect leaf A/E. (5) The g_m/g_s ratio will be positively correlated with leaf A/E; associations with high Thick_cw could have implications for plant drought tolerance.

RESULTS

Significance of ANOVA for Structural and Functional Leaf Traits

The accessions analyzed in this study, and the corresponding species and genomes, are listed in Table I; and, the list of leaf traits determined are shown in Table II. The overall significance for differences among Oryza genomes, species nested within genomes, and accessions nested within species and genomes, for each leaf structural and functional trait, are reported in Table III. Among genomes, there were significant differences in most structural and functional traits. Among species nested within genomes, there were significant differences in about half of the leaf traits, including stomatal pore area index in the abaxial lamina surface (I_stab), mean Thick_leaf, total S_mes, Thick_cw, maximum carboxylation efficiency (CE), g_m, and A. Among accessions nested within species and genomes, there were significant differences in only a few leaf traits, notably I_stab, Thick_cw, g_m, A, E, and A/E.

Table II. List of leaf traits estimated in this study, and symbols and units adopted.

Trait Category/Trait	Symbol	Unit
Leaf structural traits
Stomatal density in the abaxial lamina surface	Dst ab	mm−2
Stomatal density in the adaxial lamina surface	Dst ad	mm−2
Stomatal pore length in the abaxial lamina surface	Lst ab	μm
Stomatal pore length in the adaxial lamina surface	Lst ad	μm
Stomatal pore area index in the abaxial lamina surface	Ist ab	mm2 mm−2
Stomatal pore area index in the adaxial lamina surface	Ist ad	mm2 mm−2
Stomatal pore area index per unit (one side) of lamina surface area	Ist	mm2 mm−2
Leaf thickness	Thickleaf	μm
Fraction of leaf mesophyll volume occupied by intercellular air space	VolIAS	%
Total mesophyll cell surface area exposed to intercellular air space per unit (one side) of leaf surface area (M index)	Smes	μm2 μm−2
Total mesophyll cell surface area occupied by chloroplasts exposed to intercellular air space per unit (one side) of leaf surface area (chloroplast index)	Schl	μm2 μm−2
M cell structural traits
Area of cell section (in a leaf cross section)	acell	μm2
Ratio between perimeter and area of a cell section (in a leaf cross section)	P/acell	μm μm−2
Cell surface lobing (corresponding to cell perimeter tortuosity)	Lobcell	μm μm−1
Cell wall thickness	Thickcw	μm
Length of a single cell wall exposed to intercellular air space	Lcw_IAS	μm
Fraction of cell wall exposed to intercellular air space	CWIAS	%
Length of a single cell wall exposed to intercellular air space covered by chloroplasts	Lcw_IAS_chl	μm
Fraction of cell wall exposed to intercellular air space covered by chloroplasts	CWIAS_chl	%
Length of a single cell wall adjacent to other cells	LACW	μm
Length of a single cell wall adjacent to other cells covered by chloroplasts	LACW_chl	μm
Fraction of cell wall adjacent to other cells covered by chloroplasts	ACWchl	%
Fraction of cell wall covered by chloroplasts	CWchl	%
Fraction of cell volume occupied by chloroplasts	Volchl	%
Fraction of cell volume occupied by vacuole	Volvac	%
Mitochondria minor axis	dmit	μm
Leaf functional traits
13C/12C biomass isotopic signature	δ13C	‰
Maximum net photosynthetic rate per unit (one side) of leaf surface area	Amax	μmol CO2 m−2 s−1
Maximum carboxylation efficiency	CE	mol CO2 m−2 s−1 bar−1 CO2
Mesophyll conductance to CO2 diffusion	gma	mol CO2 m−2 s−1 bar−1 CO2
Mesophyll resistance to CO2 diffusion per unit of cell surface area exposed to intercellular air space	Smes/gm	m2 s bar CO2 mol−1 CO2
CO2 compensation point at 20% oxygen	Γ	μbar
Intercellular partial pressure of CO2	Cia	μbar
Difference between atmospheric and intercellular partial pressure of CO2	Ca − Cia	μbar
Chloroplastic partial pressure of CO2	Cca	μbar
Difference between intercellular and chloroplastic partial pressure of CO2	Ci − Cca	μbar
Stomatal conductance to water vapor diffusion	gs_H2Oa	mol water m−2 s−1
Stomatal conductance to CO2 diffusion	gs_CO2a	mol CO2 m−2 s−1
Ratio between mesophyll and stomatal conductance to CO2 diffusion	gm/gs_CO2	bar−1
Net photosynthetic rate per unit (one side) of leaf surface area	Aa	μmol CO2 m−2 s−1
Transpiration rate per unit (one side) of leaf surface area	Ea	mmol water m−2 s−1
Transpiration efficiency	A/E	μmol CO2 mmol−1 water
Intrinsic transpiration efficiency	A/gs_ H2O	μmol CO2 mol−1 water

^aDetermined at 350 μbar air partial pressure of CO₂.

Table III. Overall significance of three-stage (genomes, species, accessions) nested ANOVA for all considered leaf structural and functional traits in the genus Oryza.

Trait symbols correspond to the descriptions in Table II. Asterisks represent the significance for differences among genomes, species nested within genomes, and accessions nested within species and genomes. *P < 0.05, **0.01 ≥ P ≥ 0.001, ***P < 0.001. Min and Max correspond to the minimum and maximum means. F = F test statistic; Num df and Den df = numerator and denominator degrees of freedom, respectively.

Trait Category/Trait	Unit		Min	Max	Genomes						Species						Accessions
					F	Num df	Den df	P	Min	Max	F	Num df	Den df	P	Min	Max	F	Num df	Den df	P
Leaf and cell structural
Dst ab	mm−2	156.8	391.2	8.63	7	39	***	156.8	450.0	4.25	9	39	***	156.8	450.0	1.10	3	39
Dst ad	mm−2	167.3	501.0	11.61	7	38	***	147.3	501.0	2.09	9	38		147.3	501.0	0.67	3	38
Lst ab	μm	9.67	17.32	14.67	7	39	***	8.87	18.07	9.54	9	39	***	8.87	19.03	0.91	3	39	**
Lst ad	μm	10.50	17.67	6.42	7	39	***	10.1	17.67	2.11	9	39		10.1	18.57	0.99	3	39
Ist ab	mm2 mm−2	0.028	0.062	5.47	7	39	***	0.017	0.091	4.82	9	39	***	0.017	0.099	5.87	3	39	**
Ist ad	mm2 mm−2	0.009	0.012	8.89	7	38	***	0.026	0.140	1.49	9	38		0.026	0.140	0.75	3	38
Ist	mm2 mm−2	0.183	0.062	7.03	7	38	***	0.043	0.183	3.53	9	38	**	0.043	0.183	3.10	3	38	*
Thickleaf	μm	57.8	120.4	26.02	7	28	***	57.8	120.4	6.98	9	28	***	57.8	125.3	1.44	1	28
VolIAS	%	15.75	25.70	4.59	7	28	**	15.75	25.97	2.20	9	28		15.75	27.95	2.13	1	28
Smes	μm μm−2	10.17	24.43	15.71	7	28	***	8.77	24.43	4.25	9	28	**	8.77	24.43	0.04	1	28
Schl	μm μm−2	9.05	23.40	22.98	7	28	***	6.27	23.40	6.15	9	28	***	7.90	23.40	3.42	1	28
acell	μm2	58.8	208.2	7.29	7	19	***	58.8	242.6	2.14	9	19		58.8	242.6	0.11	1	19
P/acell	μm μm−2	0.380	0.580	4.62	7	19	**	0.380	0.590	1.58	9	19		0.370	0.590	0.12	1	19
Lobcell	μm μm−1	1.218	1.403	8.83	7	19	***	1.060	1.445	1.32	9	19		1.060	1.445	0.01	1	19
Thickcw	μm	0.150	0.190	3.43	7	20	*	0.125	0.190	5.55	9	20	***	0.125	0.190	6.87	1	20	*
Lcw_IAS	μm	14.60	40.78	7.41	7	20	***	14.60	41.20	2.24	9	20		14.60	41.20	0.00	1	20
CWIAS	%	38.0	54.8	8.21	7	20	***	35.0	54.8	2.50	9	20	*	35.0	55.5	0.16	1	20
Lcw_IAS_chl	μm	13.15	39.58	9.53	7	20	***	13.15	39.58	2.00	9	20		13.15	39.80	0.01	1	20
CWIAS_chl	%	86.55	97.13	1.69	7	20		63.75	97.95	3.63	9	20	**	63.75	97.95	0.01	1	20
LACW	μm	16.60	50.63	5.95	7	20	***	16.60	52.30	1.24	9	20		16.60	52.30	0.09	1	20
LACW_chl	μm	14.15	36.10	3.82	7	20	**	14.15	38.20	0.95	9	20		14.15	38.20	0.01	1	20
ACWchl	%	72.23	88.56	4.54	7	20	**	68.85	93.05	1.95	9	20		68.85	93.05	0.60	1	20
CWchl	%	78.37	91.56	3.07	7	20	*	73.95	94.30	2.53	9	20	*	73.95	94.30	0.05	1	20
Volchl	%	49.85	61.68	2.09	7	20		39.00	64.45	3.84	9	20		39.00	64.45	0.28	1	20
Volvac	%	15.30	27.35	2.23	7	20		8.20	43.25	6.63	9	20	***	8.20	43.25	0.11	1	20
dmit	μm	0.475	0.745	6.49	7	20	***	0.385	0.745	4.38	9	20	**	0.385	0.755	0.10	1	20
Leaf functional
δ13C	‰	−26.86	−24.03	13.18	7	43	***	−27.37	−24.03	4.92	9	43	***	−27.42	−24.03	2.38	6	43	*
Amax	μmol CO2 m−2 s−1	29.40	45.27					24.47	45.27					24.22	46.56
CE	mol CO2 m−2 s−1 bar−1 CO2	0.044	0.189	11.64	7	47	***	0.044	0.189	5.61	9	46	***	0.044	0.202	3.00	7	46	**
gm	mol CO2 m−2 s−1 bar−1 CO2	0.051	0.467	19.94	7	46	***	0.051	0.467	5.05	9	46	***	0.051	0.555	9.38	7	46	***
Smes/gm	m2 s bar CO2 mol−1 CO2	21.9	337.2	20.08	7	27	***	21.9	337.2	1.52	9	27		21.9	337.2	0.55	1	27
Γ	μbar	40.55	52.92	11.58	7	48	***	37.17	52.92	2.38	9	48	*	37.17	52.92	2.58	7	48	*
Ci	μbar	219.7	278.4	4.65	7	47	**	219.7	278.4	1.78	9	47		208.7	278.4	0.71	7	47
Ca − Ci	μbar	75.6	134.3	4.65	7	47	**	75.6	134.3	1.75	9	47		75.6	145.3	0.79	7	47
Cc	μbar	89.1	178.4	7.32	7	41	***	89.1	185.4	2.28	9	41	*	89.1	188.0	1.91	7	41
Ci − Cc	μbar	45.5	145.4	13.74	7	41	***	45.5	145.4	3.37	9	41	*	40.5	145.4	2.05	7	41
gs_H2O	mol water m−2 s−1	0.13	0.417	4.39	7	45	**	0.13	0.520	1.38	9	45		0.13	0.572	1.67	7	45
gm/gs_CO2	bar−1	0.619	2.566	5.87	7	43	***	0.619	2.566	0.93	9	43		0.619	3.369	2.84	7	43
A	μmol CO2 m−2 s−1	9.65	22.87	11.67	7	47	***	9.65	27.09	3.39	9	47	**	9.65	27.09	2.76	7	47	*
E	mmol water m−2 s−1	2.40	6.08	4.00	7	45	**	2.40	7.74	1.88	9	45		2.40	7.81	3.06	7	45	*
A/E	μmol CO2 mmol−1 water	2.85	4.49	1.42	7	45		2.85	4.69	1.97	9	45		2.85	5.40	4.05	7	45	**
A/gs_ H2O	μmol CO2 mol−1 water	51.45	80.64	2.68	7	45	*	41.10	80.64	2.69	9	45	*	41.10	80.64	0.85	7	45

For each leaf trait, mean values (±se) per Oryza genome, species, and accession are reported in Supplemental Tables S1, S2, and S3, respectively. In addition, for each trait, corresponding letters of statistical significance (P < 0.05) between genomes, between Oryza species nested within genomes, and between Oryza accessions nested within species and genomes are reported in Supplemental Tables S1, S2, and S3, respectively.

Differences in Leaf and Cell Structural Traits across Accessions

Light and electron micrographs of leaf cross sections for five representative species of Oryza are shown in Figure 1. These images illustrate variation in leaf thickness and shape, the structure of M cells including lobe development, the positions of organelles, and cell area occupied by the vacuole. The adaxial surface of leaves ranges from being relatively flat (Fig. 1, A, D, and J) to having deep invaginations with protrusions above the veins (by extension of bundle sheath and adjacent M cells), as observed, in particular, in Oryza brachyantha (Fig. 1G) and O. coarctata (Fig. 1M). In all species studied, the mitochondria and peroxisomes are preferentially located internal to the chloroplasts, with the latter being appressed to the plasma membrane (Fig. 1, F, I, L, and O).

Figure 1.

Light (left panels) and electron (middle and right panels) microscopy images of leaf anatomy in representative Oryza species (A–C, O. sativa cv IR64; D–F, O. rufipogon 9; G–I, O. brachyantha; J–l, O. australiensis 21; M–O, O. coarctata). A, D, G, J, and M, Leaf cross-sections. B, C, E, H, K, and N, M cell shape, vacuole development, distribution of chloroplasts, and development of lobes. F, I, L, and O, Positions of mitochondria and peroxisomes internal to the chloroplasts. bc, Bulliform cells; bs, bundle sheath; ch, chloroplast; e, epiderm; l, lobes; m, mitochondria; mc, mesophyll cell; n, nucleus; p, peroxisomes; v, vacuole; vb, vascular bundle.

The minimum and maximum means within Oryza accessions for each leaf trait quantified by microscopy, are shown in Table III (see mean trait values for each accession in Supplemental Table S3). There is an approximately 2-fold difference in Thick_leaf across accessions (from approximately 58 to 125 μm). Oryza rufipogon 9 (Fig. 1D) and O. brachyantha (Fig. 1G) have the thinnest leaves; O. australiensis 21 (Fig. 1J), O. australiensis 20, Oryza rhizomatis 16, and O. coarctata (Fig. 1M) have the thickest leaves. The Vol_IAS is lowest in O. brachyantha and highest (approximately 1.8 times) in O. australiensis 20.

Across accessions, 89% (overall mean) of S_mes is covered by chloroplasts; the lowest values were 63% for O. glaberrima and 67% for O. meridionalis; all other values are above 83% (calculations are based on S_mes and S_chl data in Supplemental Table S3). There is an approximately 3-fold difference across accessions in S_mes (and S_chl); Oryza glumaepatula 3 had the lowest (8.8 μm² μm⁻²) and O. coarctata had the highest (24.4 μm² μm⁻²) S_mes values. The a_cell varied approximately 4-fold across accessions (lowest in O. brachyantha to highest in Oryza latifolia 18). There is a 1.5-fold difference in the extent of Lob_cell, which was lowest in O. brachyantha (1.06 μm μm⁻¹), high in O. rufipogon 9 and O. sativa cv IR64, and highest in Oryza barthii (1.45 μm μm⁻¹). Thick_cw varied approximately 1.5-fold (from 0.125 μm in O. rufipogon 9, to 0.190 μm in Oryza minuta).

The fraction of M cell wall covered by chloroplasts (CW_chl) was lowest in O. glaberrima (approximately 74%) and highest in Oryza nivara 7 (approximately 94%); most accessions have CW_chl greater than 85% (Supplemental Table S3). Among accessions, the fraction of M cell wall exposed to the IAS (CW_IAS) ranged from 35% to 56%, and the fraction of M cell wall exposed to IAS covered by chloroplasts (CW_{IAS_chl}) ranged from 64% to 98%. The fraction of cell wall adjacent to other cells (ACW) and ACW covered by chloroplasts (ACW_chl) varied from 69% in O. latifolia 18, to 93% in O. nivara. There was a 5.3-fold difference in the fraction of cell volume occupied by vacuoles; it was lowest in O. rufipogon 9 (8.2%; Fig. 1, E and F), medium high in O. coarctata (Fig. 1, N and O), and highest in O. glaberrima (43.3%).

There is an approximately 4-fold difference in the leaf cumulative stomatal pore area index (I_st), which is lowest in Oryza alta (having low stomatal density and size) and highest in O. barthii (having high stomatal density and size). In general, except for O. coarctata and O. brachyantha, the stomatal density is higher on the abaxial than on the adaxial lamina surface.

Differences in Leaf Functional Traits across Accessions

The minimum and maximum means within Oryza accessions for each leaf functional trait, are shown in Table III (for mean trait values for each accession, see Supplemental Table S3). The difference in A was approximately 2.9-fold (minimum of 9.7 μmol CO₂ m⁻² s⁻¹ in O. brachyantha and maximum of 27.1 μmol CO₂ m⁻² s⁻¹ in O. glaberrima), that in E was approximately 3-fold, and that in A/E and intrinsic transpiration efficiency (A/g_{s_H2O} ratio) was approximately 2-fold. The difference in CE was approximately 4.6 fold (minimum in O. brachyantha and maximum in O. australiensis 22), in g_m was approximately 10.9 fold (maximum in O. australiensis 21, minimum in O. coarctata), and in M resistance to CO₂ per unit of cell surface area exposed to IAS (S_mes/g_m) was approximately 15-fold. In addition, across accessions, there was a range of approximately 1.4-fold in CO₂ compensation point (Γ), approximately 1.3-fold in C_i, and approximately 2-fold in C_c. There was an approximately 4.4-fold difference in g_{s_H2O} and an approximately 5.4-fold difference in the ratio between g_m and g_{s_CO2}.

Leaf Structural-Functional Principal Factor Analysis

From principal factor analysis (PFA) performed on the leaf structural and functional traits of Oryza accessions four main axes of covariation (factors) were considered (for trait loadings on the factors, see Fig. 2, left side). The cluster of interrelated leaf traits with high positive association with factor 1 included Thick_leaf (along with other structural features, Vol_IAS, S_mes, S_chl, and a_cell), maximum net photosynthetic rate (A_max), and A/E; A/g_{s_H2O} had a lower positive association. In contrast, E and the ratio between M cell perimeter and cell section area in a leaf cross section (P/a_cell) had negative associations. Factor 1 was interpreted to represent M architecture. The cluster of interrelated leaf traits with high positive associations with factor 2 included A, CE, g_m, C_c, g_{s_H2O}, and E, while g_m/g_{s_CO2} had a lower positive linkage; conversely, the difference between intercellular and chloroplastic partial pressure of CO₂ (C_i − C_c) and S_mes/g_m had negative associations with factor 2 (Fig. 2, left side). Factor 2 was interpreted to represent photosynthetic-transpiration activity. Factors 1 and 2 accounted for 27.1% and 19.6% of the total variability, respectively.

Figure 2.

PFA of leaf structural and functional traits. At left are factor loadings for all leaf structural and functional traits determined on 18 Oryza accessions. Var_tot expl, Total variance explained. Factor loadings greater than 0.4 in absolute value are shown in boldface. The graph associated with the data are in Supplemental Table S4, left side. Trait symbols are as in Table II. At right are loadings of the Oryza accessions (identified by number as in Table I) on factor 1 (M architecture) and factor 2 (photosynthetic-transpiration activity). Genome symbols are associated with each accession.

A third cluster of interrelated leaf traits positively associated with factor 3 included stomatal density in both the abaxial (D_stab) and adaxial (D_stad) lamina surfaces, I_stab and stomatal pore area index in the adaxial lamina surface (I_stad), leaf cumulative I_st, and ACW_chl, while Vol_IAS had a negative linkage. Factor 3 was interpreted to represent leaf stomatal composition; it accounted for 13.0% of total variability (Fig. 2, left side; data not shown). The fourth cluster of interrelated leaf traits positively associated with factor 4 included CW_{IAS_chl}, ACW_chl, CW_chl, and the fraction of cell volume occupied by chloroplasts (Vol_chl), while the fraction of cell volume occupied by vacuole (Vol_vac) had a negative loading. Factor 4 was taken to represent M cell chloroplast display; it accounted for 12.0% of the total variability (Fig. 2, left side; data not shown). In total, the main four factors accounted for 71.7% of the total variability in the data set.

The Oryza accessions loaded on M architecture (factor 1) and photosynthetic-transpiration activity (factor 2) are shown in Figure 2, right side (the corresponding scores are displayed in Supplemental Table S4, right side). In general, the accessions tended to group in three clusters based on M architecture, while they showed an overall higher variability on photosynthetic-transpiration activity. In particular, O. australiensis 20 and 21 and O. rhizomatis 16 had high scores on factor 1, while O. rufipogon 9 and O. glumaepatula 3 had low scores on this factor. However, all these species had high scores on factor 2. The rice accessions O. sativa cv IR64 (10) and O. glaberrima 2 had the highest scores on factor 2; vice versa, O. brachyantha 23 and O. coarctata 24 had the lowest scores (but O. coarctata had high scores on factor 1 compared with O. brachyantha 23). The accession scores for factor 3 and 4 are not graphed but are displayed in Supplemental Table S4, right side.

Pearson Correlation Matrices for Structural, Functional, and Structural-Functional Leaf Traits

To complement the PFA, trait-to-trait correlations were further analyzed, based on the data set. The Pearson correlation matrix for leaf structural traits is shown in Supplemental Table S5. In particular, Thick_leaf showed positive correlations with several leaf structural features, including a_cell, S_mes, S_chl (P < 0.01), and Vol_IAS (P < 0.05). S_mes and S_chl are highly correlated (r = 0.97; P < 0.01). S_mes was also closely associated with I_stad (r = 0.67; P < 0.01). Lob_cell showed a positive correlation with the length of M cell walls adjacent to other cells covered by chloroplasts (L_{ACW_chl}; P < 0.01), and also with a_cell, although not significant (r = 0.30). There was a close negative association between Vol_chl and Vol_vac (P < 0.01).

The Pearson correlation matrix for leaf functional traits is shown in Supplemental Table S6. There is a positive correlation between CE and both A and A_max (P < 0.01). g_m has a positive association with A and C_c (P < 0.01) and with g_{s_H2O} (P < 0.05). g_m was negatively correlated (P < 0.01) with S_mes/g_m, which is linked to the large difference in g_m across accessions compared with S_mes. S_mes/g_m exhibited a negative association with C_c, g_{s_H2O}, A, and E (P < 0.01) and with CE (P < 0.05). C_i was negatively associated with A/g_{s_H2O}, A/E, and g_m/g_{s_CO2} (P < 0.01). There was a tight positive correlation between A and E (P < 0.01). In addition, g_{s_H2O} was positively correlated with both A and E and negatively associated with C_i − C_c (P < 0.01).

The portion of the Pearson correlation matrix showing the leaf structural-functional trait-to-trait associations is in Supplemental Table S7. Thick_leaf had a close negative correlation with C_i (P < 0.01) and a positive association with A_max (P < 0.01); it had a negative relationship with E (P > 0.05) and significant positive associations with A/E (P < 0.01) and A/g_{s_H2O} (P < 0.05). In addition, S_mes had a tight negative association with E (P < 0.01) and close positive associations with A/E (P < 0.01) and A/g_{s_H2O} (P < 0.05).

With respect to stomatal traits on the abaxial leaf surface, D_stab has a significant positive association with g_{s_H2O} (P < 0.05) and with A and E (P < 0.01); in addition, I_stab has a positive association with CE (P < 0.05) and a negative strong relationship with δ¹³C (P < 0.01). In contrast, on the adaxial leaf surface, there is a significant positive correlation between I_stad and S_mes/g_m (P < 0.01). In addition, I_stad has a negative association with E (r = −0.41; P > 0.05) and a positive association with A/g_{s_H2O} (P < 0.05). Lob_cell has a negative correlation with Γ (P < 0.05). Although not significant (P > 0.05), the results suggest that Thick_cw has a negative relationship with g_m (r = −0.42) and a positive association with S_mes/g_m (r = 0.37).

Trait-to-Trait Correlations

Considering the clusters of interrelated leaf traits closely associated with PFA factors 1 and 2, and of parameters of most interest with respect to relationships between leaf structure, photosynthesis, and transpiration (see introduction), trait-to-trait correlations were analyzed in plots of mean values for each Oryza accession (Figs. 3–10; note that in most panels, the Pearson correlation coefficients were significant; Supplemental Tables S5–S7).

Figure 3.

Correlation of leaf thickness with other leaf structural traits in Oryza accessions. A, Vol_IAS. B, S_mes. C, a_cell. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Figure 10.

Correlations between Thick_cw and functional traits in Oryza accessions. A, S_mes/g_m. B, g_m. Accessions above the dashed line have g_m > 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂ and show a negative relationship between g_m and g_{s_CO2} (Fig. 5C). Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Structural Traits

First, with respect to structural traits, the significant positive associations between mean values of Vol_IAS, S_mes, and a_cell, and the corresponding mean values of Thick_leaf, are shown in Figure 3. In particular, O. australiensis 20 and 21, O. rhizomatis 16, and O. coarctata 24 showed the highest values for these M and cell structural features. In contrast, O. brachyantha 23 and O. sativa cv IR64 (10) had the lowest values. There is also a close positive association between Lob_cell and L_{ACW_chl} (Fig. 4), which is in part driven by O. brachyantha 23, which has the lowest L_{ACW_chl} value; O. barthii 1 has the highest L_{ACW_chl} value.

Figure 4.

Correlation, across Oryza accessions between Lob_cell and L_{ACW_chl}. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Functional Traits

Significant relationships for leaf functional features associated with the photosynthetic process are shown in Figures 5 and 6. In the close positive association between A and g_{s_CO2} across accessions (Fig. 5A), O. coarctata 24 and O. brachyantha 23 have the lowest values for both traits; in contrast, O. sativa cv IR64 (10) and O. latifolia 19 have the highest A (above 25 μmol CO₂ m⁻² s⁻¹) and g_{s_CO2} values. There is also a strong positive association between A and g_m (Fig. 5B); however, there were deviations; O. australiensis 21 and 22 have the highest g_m values (close to 0.55 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) but not the highest A. Also, very high g_m values (greater than 0.45 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) were found in O. rufipogon 9, with A approximately 25 μmol CO₂ m⁻² s⁻¹, and in Oryza punctata 12, with A approximately 22 μmol CO₂ m⁻² s⁻¹. In the plot of g_m versus g_{s_CO2} (Fig. 5C), there is no apparent linear relationship; rather, two pools of data are recognized. In particular, there is a close positive relationship between g_m and g_{s_CO2} up to g_m values of 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂ (r = 0.78; P < 0.01). At the higher g_m values, from 0.35 up to 0.56 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂, there is a negative relationship between g_m and g_{s_CO2} (r = −0.64; P > 0.05), notably in O. australiensis 21 and 22, O. rufipogon 9, O. punctata 12, O. latifolia 19, O. glaberrima 2, Oryza longistaminata 4, and O. meridionalis 6. Figure 6 shows the significant negative correlation between C_c and S_mes/g_m, which is partially driven by the high S_mes/g_m values of O. coarctata 24 and O. brachyantha 23 (above 180 m² s bar CO₂ mol⁻¹ CO₂); most accessions have mean S_mes/g_m values in the range of 25 to 80 m² s bar CO₂ mol⁻¹ CO₂.

Figure 5.

Correlations between leaf functional traits associated with the photosynthetic process in accessions of Oryza species. A, A versus stomatal conductance to CO₂ diffusion. B, A versus M conductance to CO₂ diffusion. C, g_m versus stomatal conductance to CO₂ diffusion. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Figure 6.

Correlation between C_c and S_mes/g_m across Oryza accessions. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Plots of important relationships between leaf functional features associated with transpiration are shown in Figure 7. In Figure 7A, the significant negative correlation between g_{s_H2O} and C_i − C_c (which is partially driven by the highest C_i − C_c and the lowest g_{s_H2O} values of O. coarctata 24 and O. brachyantha 23) reflects a significant negative association between E and C_i − C_c (data not shown). O. australiensis 21 and 22, O. rufipogon 9, and O. sativa cv IR64 (10) had the lowest C_i − C_c values; among these accessions, O. australiensis 21 also had a medium-low g_{s_H2O}. O. glaberrima 2, O. latifolia 19, and O. australiensis 20 also had relatively low C_i − C_c values. The tendency for a general positive association between δ¹³C and A/g_{s_H2O} (i.e. less negative δ¹³C tends to correspond to high A/g_{s_H2O}) is displayed in Figure 7B. In particular, O. nivara 7 has the lowest δ¹³C (the most negative value) and the lowest A/g_{s_H2O}, while O. punctata 12 has the highest δ¹³C and a medium A/g_{s_H2O} value. O. australiensis 21 and O. glaberrima 2 were outliers, as they have low δ¹³C values (more negative than −26.7‰) and the highest A/g_{s_H2O} values.

Figure 7.

Correlations between leaf functional traits associated with the transpiration process in Oryza accessions. A, g_{s_H2O} and C_i − C_c. B, δ¹³C and A/g_{s_H2O}. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

Structural Versus Functional Traits

Important relationships between leaf structural and functional traits related to both photosynthesis and transpiration are shown in Figures 8–10. There is a significant positive association between Thick_leaf and A_max (Fig. 8A), with O. australiensis 20 and 21 and O. rhizomatis 16 having the highest values for both traits. There is a close negative association between Thick_leaf and C_i (Fig. 8B), with O. australiensis 21 and O. coarctata 24 having the lowest C_i values. Results suggestive of a negative association between Thick_leaf and E are displayed in Figure 8C; O. australiensis 20 is an outlier, having high Thick_leaf and a relatively high E, while O. brachyantha 23 is an outlier that has the lowest Thick_leaf along with low E. Figure 8D shows the close positive association between Thick_leaf and A/E; among accessions having the thickest leaves, O. rhizomatis 16 and O. australiensis 21 have the highest A/E values, while O. australiensis 20 is an outlier with a medium A/E value. Also, there is a significant positive association (r = 0.59; P < 0.05) between Thick_leaf and A/g_{s_H2O} (data not shown).

Figure 8.

Correlations between Thick_leaf and leaf functional traits in Oryza accessions. A, A_max. B, C_i. C, E. D, A/E. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

The significant positive relationship between A and D_stab is shown in Figure 9. O. coarctata 24 has the lowest D_stab, while O. rufipogon 9 has the highest value. O. sativa cv IR64 (10) and O. glaberrima 2 have medium to high D_stab and the highest A values.

Figure 9.

Correlation in Oryza accessions between leaf A and D_st ab. Points are mean values per accession (se values are given in Supplemental Table S3). Numbers correspond to the accessions listed in Table I.

The data suggestive of a positive association between Thick_cw and M resistance to CO₂ diffusion per unit of cell surface area exposed to IAS are shown in Figure 10A. O. coarctata 24 and O. brachyantha 23 are outliers, with high S_mes/g_m values. Results suggestive of a negative correlation between Thick_cw and g_m are shown in Figure 10B. In particular, accessions having g_m values greater than 0.45 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂ (including O. rufipogon 9, O. punctata 12, and O. australiensis 21) had Thick_cw of 0.150 μm or less. All the accessions above the dashed line in Figure 10B correspond to those in Figure 10A having the lowest S_mes/g_m values and to those in Figure 5C having a negative association between g_m (for values above 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) and g_{s_CO2}. In contrast, the accessions below the dashed line have the highest S_mes/g_m (Fig. 10A) and a positive correlation between g_m (below 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) and g_{s_CO2} (Fig. 5C).

There is a negative association between Γ and Lob_cell (data not shown; r = −0.53; P < 0.05); this is driven in part by O. brachyantha 23, which has the highest Γ (52.9 μbar) and the lowest Lob_cell (1.060 μm μm⁻¹), and by O. minuta 13, which also has a high Γ (51.7 μbar) and Lob_cell of 1.225 μm μm⁻¹ (Supplemental Table S3).

DISCUSSION

Interrelations among Leaf Thickness and Other Leaf Structural Traits

The hypothesis that in Oryza species some M structural traits will be associated with leaf thickness was confirmed by the PFA results for factor 1 (M architecture) and by trait-to trait correlation analysis; in particular, Thick_leaf had significant positive correlations with S_mes and S_chl, Vol_IAS, and a_cell (Terashima et al., 2006).

The positive correlation between Thick_leaf and S_mes (and S_chl) suggests that the higher number of chloroplasts in thicker leaves is coordinated with increases in S_mes. In the absence of an association between Thick_leaf and S_mes, more chloroplasts would become located away from the cell surface area facing the IAS: this would increase the M diffusive resistance to CO₂ (Nobel et al., 1975; Evans et al., 2009), and light availability for many chloroplasts could also become limited (Oguchi et al., 2003). In this regard, Hanba et al. (1999) found that the association between S_mes and Thick_leaf holds across different leaf types among evergreen species.

A positive association was found across Oryza accessions between Thick_leaf and Vol_IAS, with the mean Thick_leaf ranging from approximately 58 to 125 μm and Vol_IAS from approximately 16% to 28%. Niinemets (1999), in a study on woody shrubs and grass species, reported a huge variation in leaf anatomy, with Thick_leaf ranging from approximately 55 to 1,960 μm and M porosity (corresponding to Vol_IAS in this study) ranging from 10% to 36%. It was suggested that the variation in Thick_leaf and in the corresponding leaf structural profile affects the pattern of light absorption through the layers of M cells and the distribution of chlorophyll and Rubisco, while the consequences for the diffusion of CO₂ in the IAS are considerably smaller.

In addition, in Oryza species, there was a tendency for a positive association between a_cell and Lob_cell (r = 0.30; P > 0.05); this suggests that Lob_cell may increase with cell volume, which could reduce the resistance to CO₂ diffusion into the larger M cells. Also, among Oryza accessions there is a tendency toward a positive association between Thick_cw and both S_mes and Thick_leaf (r = 0.42 and 0.24, respectively; P > 0.05); Thick_cw, acting as resistance in series for M CO₂ diffusion, may reduce g_m, as observed by Terashima et al. (2006).

Coordination between g_m and M Structural Traits

Another premise of this study was that g_m in Oryza accessions would be correlated with leaf M structural traits. Based on trait-to-trait analysis, g_m was not significantly correlated with any single M structural trait, including S_mes, Lob_cell, or Thick_cw; however, there was a tendency toward a lower g_m and higher S_mes/g_m with increasing M cell wall thickness (P > 0.05; Fig. 10). Lack of close correlations of g_m with single traits might be expected if there are covariations of multiple leaf and M cell structural traits that tend to have both positive (e.g. S_mes) and negative (e.g. Thick_cw) effects on g_m (Supplemental Materials and Methods S3; Terashima et al., 2011).

Based on the analysis conducted on leaf data from multiple species by Evans et al. (2009), there was an apparent positive correlation between g_m and S_chl, but with a low slope when Thick_cw was greater than 0.140 µm. Since Thick_cw in Oryza accessions is also relatively high (approximately 0.125–0.190 µm), this may account for the lack of a positive correlation between g_m and S_mes (or S_chl). Evans et al. (2009) estimated that the cell wall may contribute from 25% to 30% or more of the total M resistance to CO₂, even if changes in one component of the conduction pathway may be countered by pleiotropic compensatory changes; these findings are in agreement with Barbour et al. (2010).

Interestingly, lobes in M cells can increase the M cell surface to cell volume (Parker and Ford, 1982; Burundukova et al., 2003), which is convenient for CO₂ diffusion into the cell (Nobel, 2009), and they may contribute to S_mes (Nobel, 2009) and g_m (Evans and Loreto, 2000; Evans et al., 2009). M cell lobes are prominent in some Oryza species, and most lobes of contiguous cells do not interlock. Approximately 50% of the M cell wall is adjacent to other cells, with a high percentage of this area also occupied by chloroplasts. In contrast, C₃ species with high photosynthetic activity generally have few chloroplasts located where the cell wall is in contact with other cells (Evans et al., 2004). In Oryza accessions Lob_cell was positively correlated with the length of the cell wall adjacent to another cell (L_ACW) and with L_{ACW_chl}. These tight associations suggest that cell lobing may allow a larger proportion of the cell wall to be adjacent to other cells, while allowing for CO₂ and light access to the chloroplasts via the cell walls between the lobes. These M cell structural features may contribute to an increase in g_m and enhance leaf photosynthetic activity, with a positive feedback on the refixation of photorespired CO₂ (Nobel, 2009); this could lower Γ by decreasing the CO₂ chloroplastic photocompensation point. In this study, in fact, there was a significant negative association between Lob_cell and Γ (P < 0.05), and the results are also suggestive of a positive association between Lob_cell and A (P > 0.05). M cell lobes may be particularly beneficial in hot environments, where rates of photorespiration are high (Peterhansel and Maurino, 2011). Also, the relative positions of mitochondria and peroxisomes compared with chloroplasts in lobed M cells (Fig. 1; Sage and Sage, 2009) may enhance CO₂ refixation (Busch et al., 2013).

In addition to these leaf and M cell structural traits, the composition of cell walls, plasma membrane, and chloroplast envelope (Kogami et al., 2001; Nobel, 2009) as well as biochemical factors increasing membrane permeability to CO₂ are recognized to affect the CO₂ diffusion from IAS to chloroplast stroma (Bernacchi et al., 2002; Terashima and Ono, 2002; Terashima et al., 2006; Uehlein et al., 2008; Evans et al., 2009; Tosens et al., 2012). According to Flexas et al. (2007), the implication of active transport of CO₂ by cooporins in the regulation of g_m (Hanba et al., 2004; Flexas et al., 2006) may explain the weak relationship between g_m and leaf M structure. In this study, intrinsic differences in biochemical factors might partially explain the variation in S_mes/g_m among Oryza accessions.

No correlation was found across Oryza accessions between Vol_IAS and g_m, which might be explained by the fact that changes in the gaseous CO₂ diffusion resistances to variation in Vol_IAS tend to be relatively small (Niinemets, 1999; Nobel, 2009). Both negative and positive correlations between M porosity (corresponding to Vol_IAS) and g_m have been observed in previous studies (Flexas et al., 2008); this might depend on the covariation of some other structural trait that affects the resistance in the liquid phase and, consequently, g_m (Evans et al., 2009; Nobel, 2009). Several studies have shown that leaf density (g m⁻³), leaf thickness, and their product leaf dry mass per unit (one side) of the leaf surface area (g m⁻²) tend to be negatively linked to g_m and may set an upper limit for the maximum value of g_m (Terashima et al., 2006; Flexas et al., 2008; Niinemets et al., 2009). In particular, high leaf dry mass may be associated with thick leaves having thick M cell walls, which could offset the positive effect of S_mes on g_m and could contribute to an increase in the M resistance to CO₂ diffusion (Evans and von Caemmerer, 1996; Niinemets, 1999). Controversial results were found among evergreen, deciduous, and annual species in these relationships (Hanba et al., 1999).

Coordination between Functional Traits: A, g_s and g_m, and E

The hypothesis of correlations between A, g_s and g_m, and E in Oryza species was supported by the PFA results for factor 2 (photosynthesis-transpiration activity) and by trait-to-trait correlation analysis. The data indicate a coupling between photosynthesis and transpiration processes and are aligned with evidence that both g_s and g_m are important physical determinants of the CO₂ supply from the atmosphere to the chloroplasts (Evans and von Caemmerer, 1996; Evans, 1999; Evans and Loreto, 2000; Barbour et al., 2010).

However, across Oryza accessions despite an overall positive correlation between g_{s_CO2} and g_m (P < 0.05), a strong positive relationship (P < 0.01) between g_{s_CO2} and g_m up to g_m values of approximately 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂ was observed, corresponding to g_m/g_{s_CO2} ratios from 0.62 to approximately 1.4 bar⁻¹. Above this g_m threshold, there was a negative correlation (P > 0.05) between g_m and g_{s_CO2}, with g_m/g_{s_CO2} ratios between approximately 1.0 and 3.4 bar⁻¹. The contrasting relationships between g_m and g_s in the two g_m ranges have important implications for A/E in the different accessions. A meta-analysis of data from a multitude of grass, shrub, and tree species (Niinemets et al. (2009) revealed how photosynthesis tends to become more sensitive to fluctuations in stomatal conductance than in M conductance at high g_m values (or under conditions where g_m is higher than g_s). Similarly, in Oryza accessions the low sensitivity of A to changes in g_m higher than 0.35 mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂ could be associated with a redistribution of overall CO₂ diffusion limitation between stomata (g_s) and M cells, which affects the coordination between photosynthetic and transpiration processes.

In Oryza species, there is a close negative relationship between leaf A and C_i − C_c and a strong positive correlation between A and C_c. A positive correlation between S_mes (or S_chl) and A, which was not found in this study, would have kept C_i − C_c independent of A (Evans, 1999). In addition, a tendency for a negative correlation between C_a − C_i and C_i − C_c was observed (P > 0.05), and based on Equation 1, g_m together with g_s should be involved in the tradeoff between C_a − C_i and C_i − C_c. In fact, there was a significant negative correlation between g_m and C_i − C_c (P < 0.01); also, a tight positive association between S_mes/g_m and C_i − C_c was found (P < 0.01). In this regard, Niinemets et al. (2009) described how a drawdown of CO₂ partial pressure from substomatal cavities to chloroplasts (C_i − C_c), rather than g_m per se, characterizes the M diffusion limitation of photosynthesis.

Across Oryza accessions the data suggest that a negative correlation exists between g_m and C_i (P > 0.05); a higher g_m when C_i is low could prevent C_c from becoming too low with loss of efficiency of Rubisco activity, while a lower g_m associated with a higher C_i tends to avoid excess C_c compared with Rubisco capacity. There is evidence for a strong inverse relationship between g_m and C_i when photosynthesis is not limited by CO₂ but by regeneration of Rubisco or triose phosphate utilization (Flexas et al., 2007). In addition, there is a weak negative association between g_s and C_a − C_i and a strong negative correlation between g_s and C_i − C_c (P < 0.01), which is partially driven by O. brachyantha and O. coarctata (data not shown). This is plausible, since C_i − C_c is affected by the coordination between carboxylation activity and g_m, which can adjust faster than g_s, so that changes in C_i − C_c may occur quicker than in C_a − C_i (Flexas et al., 2007).

Relationships between Leaf Structural Traits and A/E

The results are consistent with the hypothesis that leaf structural traits in Oryza accessions are involved in the relationship between A and E, which affects A/E. Based on PFA, A/E was among the interrelated leaf traits (besides Thick_leaf and a number of structural factors) that were positively associated with M architecture; vice versa, E showed a negative association. Trait-to-trait analysis also suggested a negative relationship between Thick_leaf and E (P > 0.05), which reflects the negative correlation found between Thick_leaf and g_s (P > 0.05). When excluding two outlier accessions (O. brachyantha 23 and O. australiensis 20), a significant negative association between Thick_leaf and E was observed (P < 0.01; data not shown).

When considering features associated with photosynthesis, there was a close negative correlation (P < 0.01) between Thick_leaf and C_i, which was mirrored by a corresponding positive correlation between Thick_leaf and C_a − C_i. This suggests that C_c could potentially become more limiting for photosynthesis in thick leaves. With respect to leaf thickness, Parkhurst et al. (1988) reported that, in general, the activity of Rubisco varies through the M in response to a light gradient, making it possible to have a balance between local CO₂ pressures and local enzyme activities in a way that may increase the overall assimilation rate of the leaf. However, in the Oryza accessions Thick_leaf was not positively associated with A. This result, which is in agreement with Niinemets (1999), can be explained by the lack of correlation between Thick_leaf and g_m discussed earlier, by the tight negative relationship between Thick_leaf and C_i (and a weak negative correlation between Thick_leaf and g_s), and by the weak negative correlation between Thick_leaf and C_i − C_c. In contrast, there is a significant positive correlation between Thick_leaf and A_max; this may occur considering that saturating CO₂ will increase C_i and the C_i − C_c gradient, promoting CO₂ diffusion to the chloroplasts.

With respect to stomata, there were positive associations between the stomatal density on the abaxial side of the lamina and features of both CO₂ and water exchange, but not on the adaxial side. The results suggest that the stomata on the abaxial side of the lamina are mainly conducting CO₂ and water vapor between the leaf and atmosphere; that is, the amphistomatous leaves of genus Oryza could functionally behave as hypostomatous, by closing or nearly closing the adaxial stomata (Reich, 1984). While a strong stomatal regulation on the adaxial side of leaves to limit water loss tends to occur naturally in leaves (Smith et al., 1997), it may have been intensified in this study by light exposure of the lamina in the cuvette.

The tendency for a negative correlation (P > 0.05) between Thick_leaf and D_stab, and a significant positive linkage between D_stab and both g_s and E, may explain the tendency for a negative association observed across Oryza accessions between Thick_leaf and E. In addition, the leaf hydraulic architecture might contribute to the relationships between leaf structure and transpiration (Brodribb et al., 2007, 2010; Guyot et al., 2012). In this regard, bundle sheath extensions are known to facilitate leaf water conductivity through the M to the stomata (Sack and Holbrook, 2006). However, O. coarctata and O. brachyantha, which have prominent bundle sheath extensions toward the adaxial surface, have the majority of stomata located deep in the furrows between bundle sheaths. This leaf stomatal arrangement may provide an additional boundary layer resistance that limits E while maintaining a high leaf water potential. The structural design of O. coarctata appears to restrict the loss of water at the expense of photosynthesis (low A), by having the lowest g_s and g_m values and a low g_m/g_{s_CO2} ratio, with a high A/E (among the highest A/E and the highest A/g_{s_H2O} across Oryza accessions).

In summary, across Oryza accessions under current ambient CO₂ partial pressures and at 30°C leaf temperature, there is a positive significant association between Thick_leaf and leaf A/E (this occurs with Thick_leaf having a negative correlation with E and without a correlation with A across accessions). Higher levels of CO₂ can increase plant A/E by enhancing photosynthesis and decreasing water loss (due to a decrease in g_s with increasing CO₂; Long et al., 2004; Lammertsma et al., 2011). In Oryza accessions under CO₂ saturating conditions, there is a positive correlation of A_max with Thick_leaf; thus, accessions having thicker leaves could especially benefit from increased photosynthesis and A/E above ambient levels of CO₂ (i.e. when CO₂ is not limiting).

Correlation between g_m/g_s Ratio and Leaf A/E

A high g_m/g_s ratio is considered favorable for increasing A/E (Flexas et al., 2008, 2013; Barbour et al., 2010). The hypothesis that the g_m/g_s ratio would positively correlate with leaf A/E in Oryza species was supported. Several accessions with g_m/g_{s_CO2} greater than 1.0 and high g_m values had medium to high A and A/g_{s_H2O} (and A/E) ratios. They were O. australiensis 21 and 22, O. glaberrima 2, O. longistaminata 4, O. rufipogon 9, O. punctata 12, and O. latifolia 19; in addition, they had relatively low S_mes/g_m. These accessions have leaf features that may be of value for maintaining carbon gain while conserving water. O. rufipogon (Zhao et al., 2008, 2010) and O. australiensis (Zhao et al., 2010) are two of the wild relative resources for crop improvement of O. sativa, and there is interest in O. australiensis and O. glaberrima for attributes associated with drought tolerance (see introduction).

The δ¹³C has been used as an indicator of leaf A/E in several species (Flexas et al., 2008; Barbour et al., 2010), including rice (Dingkuhn et al., 1991; Kondo et al., 2004; Centritto et al., 2009; Xu et al., 2009). In this study on Oryza accessions a weak positive association between δ¹³C and A/g_{s_H2O} was observed (P > 0.05). Interestingly, O. glaberrima 2 and O. australiensis 21 were outliers, having high A/g_{s_H2O} but among the most negative δ¹³C values. In these two accessions, the high g_m with medium to high g_{s_H2O} values could maintain high C_c, A, and A/g_{s_H2O} with lower δ¹³C values.

Leaf Traits Contributing to Plant Drought Tolerance

Plants have diverse leaf morphoanatomical and architectural adaptations to face the challenge of coordinating structural requirements for strength and durability, controlling rates of water loss, and functioning photosynthetically (light capture, facilitating the diffusion of CO₂ to the chloroplasts, and the development of metabolic capacity for carbon assimilation; Evans, 1999; Evans et al., 2004; Brodribb et al., 2010; Waite and Sack, 2010).

In this study, there was support for the hypothesis that a high g_m/g_s ratio is coordinated with M leaf traits to confer plant drought tolerance. In particular, the results indicate that a high g_m and g_m/g_s ratio, a low S_mes/g_m, along with a high M Thick_cw are features that could be beneficial in warm climates and limited water availability. O. glaberrima, O. meridionalis, and O. australiensis 20, which are recognized as species having tolerance to drought, have relatively high Thick_cw (greater than 0.165 μm) compared with most Oryza species. In addition, they have relatively high g_m and g_m/g_s ratio and low S_mes/g_m ratio despite having high Thick_cw; also, O. longistaminata 4 has similar features. Further study of these species is needed to determine the contribution of nonstructural factors to the relationships between high g_m, low S_mes/g_m ratio, and high cell wall thickness.

CONCLUSION

This study provides insight into the diversity of leaf structure and how it relates to photosynthesis and transpiration between representatives of cultivated rice (O. sativa and O. glaberrima) and a number of wild relatives in the genus Oryza. There are significant correlations between certain structural traits, between certain functional traits associated with photosynthesis and transpiration, and fewer correlations of function with individual structural traits. There is a close positive relationship between A and g_m, but neither of these two fundamental leaf functional traits is significantly correlated, at ambient CO₂ levels, with Thick_leaf or any other single M structural feature. It is recognized that the value of g_m in plants depends on multiple structural traits that can covary between species with Thick leaf (which may include leaf S_mes and M cell features such as Thick_cw, cell volume, and Lob_cell); in addition, g_m may be affected by cell wall composition, along with biochemical factors controlling the levels and transport of inorganic carbon in the liquid phase from IAS to chloroplast stroma. Large differences were observed in leaf A/E, which is suggested to be dependent on a tradeoff between g_m and g_s, and leaf structural features (e.g. a positive correlation with leaf thickness). Interestingly, species were identified having high A/E while sustaining photosynthesis. They have high g_m values, along with high g_m/g_s ratios, low S_mes/g_m, and, in some species, relatively thick M cell walls, which could be beneficial traits for rice cultivated in warm climates and limiting availability of water.

MATERIALS AND METHODS

Plant Material

Accessions of Oryza species (listed in Table I) were grown in 3-L free-drainage pots in a controlled-environment growth chamber. The photoperiod was 12 h (8 am to 8 pm standard time) with air temperature of 28°C, and the dark period was with air temperature of 25°C. Light was emitted by 400-W metal halide and high-pressure sodium lamps and supplied to the canopy in a bell-shaped pattern (by incremental changes in photosynthetically active photon flux density [PPFD] every 2 h) with maximum PPFD of 1,100 µmol photons m⁻² s⁻¹ for 4 h. Air relative humidity was kept at approximately 70%, which corresponded to a maximum air vapor pressure deficit of approximately 1.8 kPa.

Three plants per accession were grown (one per pot) in a Sunshine Mix LC-1 soil (Sun Gro Horticulture) mixed with turface (ratio of 3:1 in volume). They were irrigated daily and fertilized twice per week to pot saturation with a nutrient solution including Sprint 330 iron chelate (1.3 g L⁻¹), magnesium sulfate (0.6 g L⁻¹), Scotts-Peters Professional 10-30-20 compound (2.8 g L⁻¹), and Scott-Peters Soluble Trace Element Mix (8.0 mg L⁻¹; Scotts).

Throughout, O. sativa accessions are written in full. For the wild relatives, when multiple accessions were analyzed per species, the species name is followed by the accession number (from Table I). Some figures identify accessions by the numbers in Table I (1–24); when referring to these figures, the accession number is always given.

Leaf and M Cell Structural Traits

Sample Preparation for Light and Electron Microscopy

The study of leaf anatomy was carried out on three 20- to 40-d-old plants per each Oryza accession marked with an asterisk in Table I (note that for O. sativa cv IR72 and O. australiensis 22, structural measurements were only made on stomata). On each plant, the midportion of fully expanded leaves (usually the third and fourth leaves from the apex on the central stem) was sampled. They were fixed at 4°C in 2% (v/v) paraformaldehyde and 2% (v/v) glutaraldehyde in 0.1 m phosphate buffer (pH 7.2), postfixed in 2% (w/v) OsO₄, and, after a standard acetone dehydration procedure, embedded in Spurr’s epoxy resin. Leaf cross sections were made using a Reichert Ultracut R ultramicrotome (Reichert-Jung). For light microscopy, semithin leaf sections were stained with 1% (w/v) toluidine blue O in 1% (w/v) Na₂B₄O₇ and observed at 100× magnification with the Olympus BH-2 light microscope (Olympus Optical) equipped with LM Digital Camera & Software (Jenoptik ProgRes Camera, C12plus). For transmission electron microscopy, ultrathin leaf cross sections were stained with 4% (w/v) uranyl acetate followed by 2% (w/v) lead citrate. Transmission electron microscopes, H-600 (Hitachi) and JEM-1200 EX (JEOL USA) equipped with a MegaView III Digital Camera, and Soft Imaging System software were used for observation and photography.

Analyses of Leaf Cross Sections

Leaf and cell structural traits, which were determined from light and electron microscopy on leaf cross sections, are listed in Table II. For each accession, the micrographs analyzed were on two to three leaves, depending on the technical difficulty, with each leaf taken from a different plant (n = 2 or n = 3). All measurements were performed on digital images using an image-analysis program (UTHSCSA Image Tool for Windows, version 3.00; University of Texas Health Science Center). For a consistent protocol, measurements on leaf M cells from light and electron microscopy were made on the first cell layer facing the adaxial side of leaves. The appearance of the size and structure of M cells throughout a leaf cross section was similar except for the few distinctive paraveinal M cells (which are very large in some species), which were excluded from analysis.

Light Microscopy

From light microscopy images of leaf cross sections, Thick_leaf (μm) was measured between the bundles and bulliform cells (Fig. 1A). The lumen a_cell (μm²) was determined (and taken as a surrogate for M cell volume; see Supplemental Materials and Methods S1). From micrographs made on each leaf, 15 to 25 Thick_leaf and a_cell measurements were taken, and the corresponding mean values per leaf were calculated.

The leaf M porosity (%), defined as Vol_IAS, was calculated as

(2)

where a_{tot_M} (μm²) is the total area between epidermal layers (including M, air space, and vascular tissue) from measurement of the leaf cross sections and a_IAS (μm²) is the area of the IAS in the leaf cross-section (Supplemental Fig. S2A).

The total length of M cell wall exposed to IAS (μm) and the width of the leaf cross section analyzed (μm) were measured, and the corresponding total S_mes (μm² μm⁻²) was inferred based on Evans et al. (1994) as:

(3)

where F is the curvature correction factor. A general prolate spheroid shape with the major axis being twice the length of the other two minor axes was assumed, as in Sage and Sage (2009) and Scafaro et al. (2011), and accordingly, based on Thain (1983), an F value of 1.42 was used.

Electron Microscopy

From electron micrographs of leaf cross sections, the total M cell wall length exposed to IAS covered by chloroplasts (μm) was determined; this was used as the numerator in Equation 3 (above) to calculate S_chl (μm² μm⁻²). The extent of Lob_cell was assumed to correspond to the cell perimeter tortuosity (μm μm⁻¹), which was determined from measurements on individual M cells (Supplemental Fig. S2B). In the absence of lobes, cell perimeter tortuosity = 1.0. Other structural measurements on M cells included Thick_cw (μm), L_{cw_IAS} (μm), CW_IAS (%), length of a single cell wall exposed to intercellular air space covered by chloroplasts (L_{cw_IAS_chl}; μm), CW_{IAS_chl} (%), L_ACW (μm), L_{ACW_chl} (μm), ACW_chl (%), CW_chl (%), Vol_chl (%), Vol_vac (%), and minor diameters of mitochondria (μm). A minimum of 25 measurements per each M cell trait were taken on each leaf, and a mean value (per leaf) was calculated.

In addition, the mean P/a_cell (μm μm⁻²) for each leaf analyzed was calculated, where P is the perimeter of the M cell section. P/a_cell was taken as a relative gauge of the ratio of M cell surface area/cell volume (Supplemental Materials and Methods S1).

Analysis of Leaf Surfaces

Images of the adaxial and abaxial epidermal surfaces were captured on three live leaves (each from different plants; n = 3) per Oryza accession under the low-vacuum mode with a Quanta 200F scanning electron microscope (FEI). For each leaf, stomatal pore length in the adaxial and abaxial lamina surface (L_stad and L_stab; μm) were calculated based on 10 to 25 measurements. The stomatal density on each leaf side was determined as D_stad and D_stab (mm⁻²). For each leaf, stomatal pore area index (I_st, mm² mm⁻²) was calculated as an index of maximum stomatal pore area per lamina area based on Sack et al. (2003) as

(4)

where D_st is the cumulative (adaxial plus abaxial) number of stomata per unit (one side) of lamina surface area and L_st is the mean stomatal pore length. In addition, the I_stad and I_stab were calculated.

Leaf Functional Measurements

Gas-Exchange Measurements: Equipment Setup and Protocol

Leaf-atmosphere CO₂ and H₂O exchange measurements were performed in Pullman, WA (mean atmospheric pressure of 92.1 kPa) using a LI-6400XT portable photosynthesis system equipped with a 6400-40 Leaf Chamber Fluorometer (LI-COR Biosciences). The Leaf Chamber Fluorometer is designed with a uniform, integrated red/blue light-emitted diode light source and a pulse-amplitude modulated (PAM) fluorometer enabling simultaneous measurement of chlorophyll fluorescence and gas exchange over the same leaf area of 2.0 cm². The fraction of blue light was set to 10% of the PPFD (μmol photons m⁻² s⁻¹) to maximize stomatal aperture. A CO₂ cartridge was used, and a 6400-01 CO₂ Injector System controlled the CO₂ partial pressure entering the cuvette (C_a, μbar). The oxygen partial pressure entering the leaf cuvette was set by mixing different partial pressures of nitrogen and oxygen in a CO₂-free air stream through two mass flow controllers (MKS Instruments).

For all accessions of Oryza species in Table II, measurements were made on one leaf on three different plants (n = 3). In general, the same leaves, or leaves in the same position as those sampled for leaf anatomical survey, were chosen. The mid to distal portion of each leaf blade was inserted in the leaf chamber for gas-exchange measurements. For almost all Oryza accessions (except for O. alta, O. latifolia 18, and O. longistaminata 5), two leaf blades were used to cover the cuvette luminal surface area.

The data were acquired between 9 am and 4 pm standard time in a room with air temperature around 30°C. Measurements were made after plants were acclimated for approximately 1 h. Leaf functional traits that were provided by leaf gas exchange are listed in Table II.

Leaf A-C_i Curves

A (µmol CO₂ m⁻² s⁻¹) and C_i (μbar) were determined at 184.2 mbar ambient oxygen partial pressure (near current ambient levels at the study site, and corresponding to 200 mmol oxygen mol⁻¹ air). PPFD was set at 1,500 µmol photons m⁻² s⁻¹, leaf temperature at 30°C, while leaf-to-air vapor pressure deficit was maintained between 1.0 and 1.5 kPa.

For each Oryza accession, an A-C_i response curve was acquired on three different plants at (decreasing) C_a from 350 to 18 µbar. In one of the replications, following measurements with decreasing CO₂, C_a was increased from 350 to 1,380 µbar to determine the A_max (µmol CO₂ m⁻² s⁻¹).

Kneading synthetic rubber (Terostat IX; Henkel Technologies) was put around the gaskets of the Leaf Chamber Fluorometer; this was to limit CO₂ diffusion leaks into and from the chamber when imposing C_a in the cuvette lumen below and above the CO₂ partial pressure in the outside air. In addition, the magnitude of the cuvette CO₂ diffusion leaks was estimated, and a CO₂ diffusion correction term was included in A and C_i computations (Supplemental Material and Methods S2).

For each A-C_i response curve, least-square regression analysis was applied to the initial slope (C_i ≤ 100 µbar) to determine the CE (mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) and the Γ (µbar). Additionally, A, C_i, C_a − C_i, E (mmol water m⁻² s⁻¹), and g_{s_H2O} (mol water m⁻² s⁻¹) at 350 µbar C_a (near current ambient levels at the study site) were extracted from the data set, and the corresponding leaf A/E (μmol CO₂ mmol⁻¹ water) and A/g_{s_H2O} (µmol CO₂ mol⁻¹ water) were calculated.

Also, g_{s_CO2} (mol CO₂ m⁻² s⁻¹) was determined from g_{s_H20} as

(5)

where 1.6 is the ratio between the molecular diffusivities of water and CO₂ in air (Massman, 1998).

The data set submitted to statistical analysis was composed of three repetitions of A, CE, Γ, C_i, C_a − C_i, E, g_{s_H2O}, A/g_{s_H2O}, and A/E (n = 3) and one value of A_max (n = 1) for each Oryza accession.

Leaf Chlorophyll Fluorescence Measurements

Leaf chlorophyll fluorescence was measured using the 6400-40 fluorometer (LI-COR Biosciences) simultaneously while acquiring the A-Ci response curves. Steady-state fluorescence (F_s) and maximum fluorescence (F_m′) were recorded after equilibration to a steady state. Maximum fluorescence was measured using a 0.8-s saturating pulse of light (approximately 8,000 µmol photons m⁻² s⁻¹ supplied by light-emitting diodes with peak emission at 630 nm).

Estimation of M Conductance to CO₂ Diffusion Based on Leaf Gas-Exchange and Chlorophyll Fluorescence Measurements

The combined leaf gas-exchange and chlorophyll fluorescence data from the A-C_i response curves were used to calculate g_m (mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) using the variable J method (Di Marco et al., 1990; Harley et al., 1992), applied as in Li et al. (2009). According to Genty et al. (1989), the total electron transport rate (J_T; µmol e⁻ m⁻² s⁻¹) was determined as

(6)

where α_leaf is the leaf absorbance of photosynthetic quanta, β is the fraction of photons absorbed by PSII, PPFD (µmol photons m⁻² s⁻¹) is the PPFD incident on the leaf, and ϕ_PSII is the photochemical yield of PSII. The default value taken for α_leaf was 0.85 (Li et al., 2009) and that for β was 0.45, assuming that approximately 45% of the photons are absorbed by PSII (Laisk et al., 2006). ϕ_PSII was computed as

(7)

Electron transport rates associated with carboxylation (J_c) and oxygenation (J_o) were calculated according to Epron et al. (1995) as

(8)

and

(9)

where A (μmol CO₂ m⁻² s⁻¹) is the net photosynthetic rate per unit of leaf surface area and R_d (μmol CO₂ m⁻² s⁻¹) is the corresponding mitochondrial respiration rate in the light (assumed to be 0.7 μmol CO₂ m⁻² s⁻¹ based on experimental measurements; data not shown). The C_c (µbar) was calculated according to Epron et al. (1995) as

(10)

where S_{in_vivo_VJ} is the in vivo apparent CO₂/oxygen specificity factor for Rubisco determined from the initial slope of the linear regression forced to the origin from plots of J_c/J_o versus C_i/oxygen partial pressure (at decreasing C_a from 350 μbar) and S_{in_vivo_Γ*} is the in vivo apparent CO₂/oxygen specificity factor for Rubisco associated with the photorespiratory compensation point (Γ*; μbar) estimated by gas-exchange measurements. In this study, Γ* was determined based on the method of Laisk (1977) at 184 mbar oxygen partial pressure and at 30°C leaf temperature. Four Oryza species were considered, two of them having among the lowest (O. sativa cv IR64 and O. rufipogon 8) and two among the highest (O. glumaepatula and O. minuta) Γ values; Γ* ranged from 42 to 48 μbar.

The equation developed by Laing et al. (1974) and Jordan and Ogren (1984) describes the relationship between S and Γ* as

(11)

where pO₂ (μbar) is air partial pressure of oxygen, k is the stoichiometry factor between carboxylase and oxygenase activity at the photorespiratory CO₂ compensation point, and S is the effective CO₂/oxygen specificity factor for Rubisco based on kinetic properties determined in vitro (bar bar⁻¹). k = 0.5 at the photorespiratory compensation point under the assumption that one CO₂ is generated in the glycolate pathway for every two oxygens reacting with ribulose-1,5-biphosphate.

The equation can be used to calculate the apparent Rubisco specificity factor S_{in_vivo_Γ*} from the measurement of Γ* as

(12)

g_m (mol CO₂ m⁻² s⁻¹ bar⁻¹ CO₂) was then calculated, based on Equation 1, as

(13)

where A, C_i, and C_c are as already defined. Per each leaf, the g_m values at C_a = 350 µbar were used in data analysis.

The ratio between g_m at C_a= 350 μbar and the corresponding g_{s_CO2} (g_m/g_{s_CO2}, bar⁻¹ CO₂), and the ratio between S_mes and g_m (S_mes/g_m, m² s bar CO₂ mol⁻¹ CO₂, which represents the M resistance to CO₂ diffusion per unit S_mes and is listed in Table II among the functional traits) were calculated. In particular, the terms included in g_m and the physical meaning of S_mes/g_m, based on Fick’s laws, are described in Supplemental Materials and Methods S3.

Based on the protocol followed, three replications (n = 3) of C_c, C_i − C_c, g_m, g_m/g_{s_CO2}, and S_mes/g_m for each Oryza accession were used in the statistical analysis.

For comparison with the chlorophyll fluorescence method, M conductance was estimated on a few Oryza accessions based on simultaneous measurements of leaf gas exchange and ¹³C isotope discrimination (Evans et al., 1986; Pons et al., 2009). Leaves of O. sativa cv IR64, O. glaberrima 2, O. glumaepatula 3, and O. rufipogon 9 were sampled according to the methodology described for leaf gas-exchange measurements. A LI-6400XT portable photosynthesis system equipped with a 6400-22L lighted opaque conifer chamber (LI-COR Biosciences) was used under the same environmental conditions described above for gas exchange measurements, except the C_a was 350 μbar. The LI-6400XT was coupled to a tunable-diode laser absorption spectroscope (TDLAS, model TGA 200A, Campbell Scientific) with a laser that measures CO₂ and ¹³C (Bowling et al., 2003; Ubierna et al., 2013). M conductance calculations were performed based on the “single-point method” (Lloyd et al., 1992); they provided g_m values similar to those estimated by chlorophyll and gas-exchange measurements. Precisely, O. sativa cv IR64, O. glaberrima 2, O. glumaepatula 3, and O. rufipogon 9 showed mean g_m values of 0.342, 0.358, 0.411, and 0.457 mol CO₂ m⁻² s⁻¹ bar⁻¹, respectively (for comparison, see corresponding g_m data in Supplemental Table S3).

Stable Carbon Isotope Analysis of Leaf Biomass

A portion of each leaf used for the A-C_i response curve was collected for analysis of δ¹³C (‰). Dried samples were combusted using a Eurovector elemental analyzer. The resulting nitrogen and CO₂ gases were separated by gas chromatography and admitted into the inlet of a Micromass Isoprime isotope ratio mass spectrometer for determination of ¹³C/¹²C ratio. δ¹³C (‰) was used to represent the ¹³C/¹²C ratio of the leaf sample relative to the isotopic reference material Vienna Pee Dee Belemnite (Bender et al., 1973), and it was determined as

(14)

where R_sample and R_standard are the ¹³C/¹²C ratios of the leaf samples and the Vienna Pee Dee Belemnite limestone, respectively. Three repetitions of δ¹³C (n = 3) were obtained for each Oryza accession.

Statistical Analyses

Statistical analyses were performed similar to Waite and Sack (2010) using SAS version 9.2 (SAS Institute). First, a three-stage (genomes, species, accessions) nested ANOVA was carried out to describe the variability of each structural and functional leaf trait, based on the entire data set for Oryza accessions. Significance for differences among genomes, species nested within genomes, and accessions nested within species and genomes was evaluated at P < 0.05, 0.01 ≥ P ≥ 0.001, and P < 0.001.

PFA (Child, 2006) was run to identify the most important axes of covariation based on a total of 42 structural plus functional leaf traits (Table II). The mean values of each trait for each of 18 accessions were used. Based on a Varimax orthogonal rotation, eight factors associated with eigenvalues greater than 1 were retained. For each trait, factor loadings greater than 0.4 in absolute value were considered important; based on the size and the sign of the important loadings, a meaningful name was suggested for each of the first four PFA factors. The eight factor scores for each Oryza accession were computed. Since the PFA was conducted on a large set of leaf traits in relation to the number of accessions examined, this can reduce the power to detect the underlying factors (Osborne and Costello, 2004).

Pearson correlation matrices were calculated based on mean values of each leaf structural and functional trait per each Oryza accession to evaluate the trait-to-trait associations. The significance of correlation coefficients was determined at P < 0.05 and P < 0.01.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Scanning electron microscope images illustrating leaf surfaces of two Oryza species having different stomatal density and size.

• Supplemental Figure S2. Example of a light microscopy image of a leaf cross section, and the scheme followed to estimate cell surface lobing.

• Supplemental Table S1. Mean values, se, and significance between Oryza genomes for each leaf structural and functional trait.

• Supplemental Table S2. Mean values, se, and significance between Oryza species (nested within genomes) for each leaf structural and functional trait.

• Supplemental Table S3. Mean values, se, and significance between Oryza accessions (nested within species and genomes) for each leaf structural and functional trait.

• Supplemental Table S4. Leaf structural-functional PFA: factor loadings and factor scores on the first four main factors.

• Supplemental Table S5. Pearson correlation matrix for leaf structural traits.

• Supplemental Table S6. Pearson correlation matrix for leaf functional traits.

• Supplemental Table S7. Portion of the Pearson correlation matrix for leaf structural to functional traits.

• Supplemental Materials and Methods S1. Estimate of cell volume from values of a_cell.

• Supplemental Materials and Methods S2. Magnitude of the CO₂ diffusion leaks into and from the Leaf Chamber Fluorometer and its effect on A and C_i calculations.

• Supplemental Materials and Methods S3. Fick’s laws applied to estimate g_m.

Glossary

g_s

stomatal conductance to gas diffusion

M

mesophyll

C_i

CO₂ concentration in the intercellular air space

C_a

CO₂ concentration in the atmosphere

IAS

intercellular air space

g_m

mesophyll conductance to CO₂ diffusion

C_c

CO₂ concentration in the chloroplasts

A

net photosynthetic rate

g_{s_CO2}

stomatal conductance to CO₂ diffusion

S_MES

mesophyll cell surface area exposed to intercellular air space per unit of leaf surface area

S_chl

total mesophyll cell surface area occupied by chloroplasts exposed to intercellular air space per unit of leaf surface area

Vol_IAS

fraction of mesophyll volume occupied by air spaces

Thick_leaf

leaf thickness

E

leaf transpiration rate

A/E

transpiration efficiency

δ¹³C

carbon isotope composition of leaf biomass

g_{s_H2O}

stomatal conductance to water vapor diffusion

Thick_cw

cell wall thickness

a_cell

area of mesophyll cell section

Lob_cell

mesophyll cell surface lobing

CE

carboxylation efficiency

I_st ab

stomatal pore area index in the abaxial lamina surface

I_st ad

stomatal pore area index in the adaxial lamina surface

CW_chl

fraction of mesophyll cell wall covered by chloroplasts

CW_IAS

fraction of mesophyll cell wall exposed to the intercellular air space

CW_{IAS_chl}

fraction of mesophyll cell wall exposed to intercellular air space covered by chloroplasts

ACW

fraction of cell wall adjacent to other cells

ACW_chl

fraction of cell wall adjacent to other cells covered by chloroplasts

I_st

stomatal pore area index

Γ

CO₂ compensation point

PFA

principal factor analysis

A_max

maximum net photosynthetic rate

P/a_cell

ratio between mesophyll cell perimeter and cell section area in a leaf cross section

C_i − C_c

difference between intercellular and chloroplastic partial pressure of CO₂

Ca − C_i

difference between atmospheric and intercellular partial pressure of CO₂

D_st ab

stomatal density in the abaxial lamina surface

D_st ad

stomatal density in the adaxial lamina surface

Vol_chl

fraction of cell volume occupied by chloroplasts

L_{ACW_chl}

length of mesophyll cell walls adjacent to other cells covered by chloroplasts

L_ACW

length of the cell wall adjacent to another cell

S_MES/g_m

mesophyll resistance to CO₂ diffusion per unit cell surface area exposed to intercellular air space

A/g_{s_H2O}

intrinsic transpiration efficiency

PPFD

photosynthetically active photon flux density

L_st ad

stomatal pore length in the adaxial lamina surface

L_st ab

stomatal pore length in the abaxial lamina surface