Mesophyll conductance in cotton bracts: anatomically determined internal CO2 diffusion constraints on photosynthesis.

Mesophyll conductance (gm) has been shown to affect photosynthetic capacity and thus the estimates of terrestrial carbon balance. While there have been some attempts to model gm at the leaf and larger scales, the potential contribution of gm to the photosynthesis of non-leaf green organs has not been studied. Here, we investigated the influence of gm on photosynthesis of cotton bracts and how it in turn is influenced by anatomical structures, by comparing leaf palisade and spongy mesophyll with bract tissue. Our results showed that photosynthetic capacity in bracts is much lower than in leaves, and that gm is a limiting factor for bract photosynthesis to a similar extent to stomatal conductance. Bract and the spongy tissue of leaves have lower mesophyll conductance than leaf palisade tissue due to the greater volume fraction of intercellular air spaces, smaller chloroplasts, lower surface area of mesophyll cells and chloroplasts exposed to leaf intercellular air spaces and, perhaps, lower membrane permeability. Comparing bracts with leaf spongy tissue, although bracts have a larger cell wall thickness, they have a similar gm estimated from anatomical characteristics, likely due to the cumulative compensatory effects of subtle differences in each subcellular component, especially chloroplast traits. These results provide the first evidence for anatomical constraints on gm and photosynthesis in non-leaf green organs.

Introduction

To ren class="Chemical">ach the sites of carboxylation within chloroplasts of leaves of C3 plants, CO2 must diffuse through stomata and mesophyll. Stomatal CO2 diffusion occurs from the ambient air just outside the leaf to the substomatal cavities, while mesophyll CO2 diffusion occurs from the substomatal cavities to just outside the mesophyll cell wall (i.e. gas phase resistance) and to all the cell structures (cell wall, plasma membrane, cytoplasm, chloroplast envelope membranes, and stroma) that CO2 must necessarily pass through to reach the carboxylation center (i.e. liquid phase resistances; Evans , 2009). Although CO2 diffusion through the leaf has been widely studied, this fairly complex process is not fully understood yet (Evans ; Flexas ; Tosens ). Many studies have shown that mesophyll conductance (gm) significantly limits photosynthesis and often can be the main limitation to photosynthesis (Flexas ; Tosens Galmés ; Peguero-Pina ). In the gas phase conductance, CO2 diffusion through intercellular air spaces may be hindered by leaf thickness, mesophyll cell shape, relative distribution of palisade and spongy tissue (Evans ), and the volume fraction of intercellular air spaces (fias) (Syvertsen ; Terashima ). Regarding the liquid phase conductance, it is mainly constrained by the cell wall thickness (Tcw), the chloroplast dimensions, and the mesophyll and chloroplast surface area exposed to leaf intercellular air spaces (Sm/S and Sc/S) (Evans , 2009; Tosens ; Tomás ). These anatomical structures have been observed to strongly differ between different species (Tomás ; Peguero-Pina ) or even within the same species growing under complex and variable growth environments (Terashima ; Tosens ).

Mesophyll conductance (n class="Chemical">gm) is importanpan>t inpan> settinpan>g the planpan>t photosynthetic capn class="Chemical">acity. However, the neglect of CO2 drawdown from the substomatal cavities to chloroplasts in the photosynthetic model at the leaf level (Niinemets, 2007) and global carbon cycle model (Sun ), by using intercellular CO2 concentration (Ci) instead of chloroplastic CO2 concentration (Cc), results in an underestimation of the biochemical parameters, particularly the maximum carboxylation rate (Vcmax) and maximum electron transport rate (Jmax). To avoid such underestimation, some modeling studies have focused on gm and estimated photosynthetic parameters using Cc at the leaf (Ethier and Livingston, 2004; Sharkey ; Gu ; Sharkey, 2016) and whole canopy scales (Sun ).

It has pan> class="Chemical">been shown that non-leaf green organs are also an important source of assimilated carbon at the ecological and agricultural scales (Tambussi ; Redondo-Gómez ; Pengelly ; Hu ; Jia ; Zhang ), and thus make a considerable contribution to the terrestrial carbon exchange. However, the importance of the mesophyll diffusion limitation for photosynthesis has yet to be studied in the non-leaf green organs.

Currently, in agricultural production, it has pan> class="Chemical">been demonstrated that cotton bracts, non-leaf green organs that cover cotton fruits, make a significant contribution to cotton carbon gain especially in the later growth stages (Hu ). A higher water use efficiency (Hu ) and drought tolerance (Zhang ) in bracts than in leaves was also reported. However, no research has focused on the relationship between these photosynthetic characteristics and the property of CO2 diffusion, especially mesophyll CO2 diffusion in the bracts. Without explicit consideration of gm, some photosynthetic parameters in bracts would have been underestimated just like in the leaves. Generally, the difference in morphological and anatomical structures between leaves and bracts is obvious (Fig. 1; Hu ). But the difference in internal mesophyll structure is still not clear. Research has shown that in the cotton bracts there is only one type of photosynthetic tissue, which is similar to the spongy tissue of the leaf. However, there are smaller and less numerous chloroplasts and more loose tissue in the spongy tissue, from which we speculate there is a larger mesophyll limitation in the bract than in the leaf. To the best of our knowledge, no previous study has analysed the effect of internal structures of palisade and spongy tissues on mesophyll diffusion of CO2. To fill this gap, cotton leaves and bracts were studied and we compared the anatomy of palisade and spongy tissue structures with that of bracts. The aims of the study were (i) to determine if bracts are constitutively more limited than leaves for CO2 diffusion; (ii) to reveal if the different types of tissues lead to a difference in mesophyll diffusion between the leaf and the bract; and (iii) to quantify the contribution of mesophyll structures to setting differences in gm and photosynthesis between leaves and bracts.

Fig. 1.

A bract and a leaf from the cotton plant.

A brpan> class="Chemical">act and a leaf from the cotton plant.

Materials and methods

Plant material

Cotton (n class="Species">Gossypium hirsutum L. ‘Xinpan>luzao 45’) planpan>ts were grownpan> at anpan> experimental field of Shihezi Agricultural College, Shihezi University, Xinpan>jianpan>g, Chinpan>a (45°19′N, 86°03′E). pan> class="Chemical">Before sowing, drip irrigation tubes were installed beneath the plastic film, which supplied water for the cotton. Seeds were sown on 21 April 2015 in rows 12 cm apart at a plant density of 1.8 × 105 ha−1. The plots were fertilized before sowing with 240 kg N ha−1 (urea), 170 kg P2O5 ha−1 [(NH4)3PO4], and 1500 kg ha−1 organic fertilizer (235 g kg−1 organic matter, 18 g kg−1 total N, 14 g kg−1 total P, and 22 g kg−1 total K). An additional 120 kg N ha−1 (urea) was applied by drip irrigation during the growing seasons. Weeds and pests were controlled in the field using standard management practices. At peak bolling stage (100–110 days after sowing), the topmost fully expanded leaf on the main stem and bract on the fruit branch of the cotton were selected for the experiment. Meteorological data during the growing season are shown in Supplementary Fig. S1 at JXB online.

Gas exchange and chlorophyll fluorescence

Gas excn class="Chemical">hange anpan>d pan> class="Chemical">chlorophyll fluorescence were measured simultaneously on the main leaves and bracts, using an open gas-exchange system (Li-6400; Li-Cor, Inc., Lincoln, NE, USA) connected to a leaf fluorometer chamber (Li-6400-40; Li-Cor Inc.). The bolls were detached from the bracts so as to be able to clamp the bract to obtain the CO2 and light response curves. Leaf temperature was set to 30 °C. The vapor pressure deficit (VPD) was between 2 and 3 kPa and the flow rate was set at 300 μmol s−1. The ratio of red:blue light was set to 90:10% PPFD to maximize stomatal aperture. CO2 concentration in the Li-6400 leaf chamber was provided by a CO2 cylinder and maintained constant at 400 μmol CO2 mol−1. Light-response curves were obtained under the light intensities 2000, 1800, 1500, 1200, 1000, 800, 500, 300, 200, 150, 100, 50, and 0 μmol m−2 s−1 for leaves and 1500, 1200, 1000, 800, 500, 300, 200, 150, 100, 50, and 0 μmol m−2 s−1 for bracts. CO2-response curves in light saturating conditions were obtained by first determining the parameters at 2000 μmol m−2 s−1 photosynthetically active photon flux density (PPFD) for leaves and at 1000 μmol m−2 s−1 for bracts (see Fig. 2B for AN–PAR curves confirming light saturating conditions for both leaves and bracts). Photosynthesis was induced with an ambient CO2 concentrations (Ca) of 400 μmol mol−1 and 21% O2 surrounding the leaf. Once steady state was reached (usually 20 min after clamping the leaf), data were recorded. Immediately after, the air inlet pipe was connected to a 2% O2 and 98% N2 medical gas bag, and a CO2-response curve (net assimilation rate (AN)–Ci curve) was obtained. After that, the Li-COR inlet was disconnected from N2 medical gas bag (i.e. air with 21% O2 was supplied again to the plant). After reaching steady state, another AN–Ci curve was obtained. In regard to the AN–Ci curve, gas exchange and chlorophyll fluorescence were first measured at Ca of 400 μmol mol−1; then Ca was decreased stepwise to 50 μmol mol−1. Upon completion of measurements at low Ca, Ca was returned to 400 μmol mol−1 to restore the original AN. Then Ca was increased stepwise to complete the curve. The number of different Ca values used for the curves was 12, and the time interval between two consecutive measurements at different Ca was restricted to 2–4 min, so that each curve was completed in 30–50 min. The actual photochemical efficiency of photosystem II (ΦPSII) was determined by measuring steady state fluorescence (Fs) and maximum fluorescence during a light-saturating pulse of ca. 8000 μmol m−2 s−1 (Fm′):

Fig. 2.

Net CO2 assimilation rate (AN) expressed on the basis of leaf area as a function of intercellular CO2 concentration (Ci) (A) and photosynthetically active photon flux density (PPFD) (B) in cotton leaves and bracts. Bracts data are also shown in the insets (C, D). Values are means±SE.

Net n class="Chemical">CO2 assimilation rate (AN) expressed on the pan> class="Chemical">basis of leaf area as a function of intercellular CO2 concentration (Ci) (A) and photosynthetically active photon flux density (PPFD) (B) in cotton leaves and bracts. Bracts data are also shown in the insets (C, D). Values are means±SE.

The electron transport rate (Jflu) was then calculated as:

Where PPFD is the photosynthetically n class="Chemical">active photon flux density, α is leaf absorptance and β reflects the partitioning of absorbed quanta between photosystems II and I (PSI and PSII). α was assumed to be 0.85 and β to be 0.5. Because numerous studies have shown that the estimation of Jflu is affected by PSI and the signal-to-noise ratio in the determination of Fm′ at high light, the electron transport rate from gas exchange under 2% O2 conditions (JA) was used to calibrate Jflu (see Pons for details).

g m was estimated by the variapan> class="Chemical">ble J method (Harley ) as:

where Γ* is the n class="Chemical">CO2 compensation poinpan>t inpan> the absence of mitochondrial respiration and Rd is day respiration. AN and Ci were taken from gas-exchange measurements at saturating light and the value of Γ* (44.04) at 30 °C from Bernacchi used for the variable J methods of calculating gm:

where TL is the leaf temperature (°C). Rd was assumed to n class="Chemical">be half of the measured dark respiration (Rn, Rd=Rn/2) (Villar ; Niinemets ). Rn was determined by gas exchange (Li-6400), after plants had been dark-adapted for more than half an hour in the evening. CO2 leakage of the leaf cuvette was determined by performing AN–Ci response curves with photosynthetically inactive leaves and bracts enclosed in the leaf chamber (obtained by heating the leaves until no variable chlorophyll fluorescence was observed), and used to correct measured leaf fluxes (Flexas ).

Estimation of gm by AN–Ci curve fitting

The curve-fitting method introduced n class="Chemical">by Sharkey (2016) was used to obtain an alternative estimate of gm. This method is based on changes in the curvature of AN–Ci response curves due to a finite gm. By non-linear curve fitting minimizing the sum of squared model deviations from the data, gm can be estimated from observed data. The same data were used for estimation of gm by the methods of Sharkey (2016) and Harley ).

Estimation of Vcmax and Jmax

The AN–Cc curves were fitted n class="Chemical">based upon the model of Farqupan> class="Chemical">har , which was later modified and developed by Harley ,b). According to the biochemical model, AN can be expressed as:

With

where n class="Chemical">Ac, Aj, anpan>d Ap are the net pan> class="Chemical">CO2 assimilation rate limited by Rubisco, ribulose 1,5-bisphosphate (RuBP), and triose phosphate use (TPU), respectively. Vc and Vo are rates of carboxylation and oxygenation of Rubisco. O is the O2 concentration at the sites of carboxylation within chloroplasts. Kc and Ko are Michaelis–Menten constants for carboxylation and oxygenation, respectively (Bernacchi ). Best-fit values of the parameters Jmax and Vcmax were obtained using the whole curve data points (i.e. Eqn 4) rather than a portion of the curve according to ‘method I’ of Miao .

Electron microscopy

Leaf and n class="Chemical">brpan> class="Chemical">act samples (4 mm×1.5 mm) were fixed by infiltration of 2.5% glutaraldehyde and 3% paraformaldehyde in 0.1 mol l−1 phosphate buffer (pH 7.2) under vacuum. Leaf samples were fixed again in 1% osmium tetroxide overnight and dehydrated in a graded acetone series and embedded in Spurr’s resin. Semi-thin leaf cross-sections of 4 μm for light microscopy and ultra-thin (80 nm) cross-sections were prepared with an ultramicrotome (Leica Ultracut, Germany). The sections for light microscopy were stained with toluidine blue. Ultra-thin cross-sections for transmission electron microscopy were stained with uranyl acetate and lead citrate double staining, observed under an electron microscope (TEM HT7700, Japan), and electron micrographs were taken with a digital camera (BH-2, Olympus). Each anatomical trait per replicate was measured 6–10 times. It should be noted that electron micrographs of palisade and spongy tissues were taken and then quantified according to the below methods and formulas.

The surface of mesophyll cells anpan>d chloroplasts exposed to leaf inpan>tercellular air sppan> class="Chemical">aces (Sm/S and Sc/S) were calculated following the method of Syvertsen as:

where Lmes is the total length of mesophyll cells fn class="Chemical">acinpan>g the inpan>tercellular air space in the palisade tissue or spongy tissue section, F is the curvature correction factor that depends on the shape of the cells (Thain, 1983; Evans ), and W is the width of the section measured.

where Lc is the total length of chloroplast surface area fpan> class="Chemical">acing the intercellular air space in the palisade tissue or spongy tissue sections.

The volume fractionpan> of inpan>tercellular air sppan> class="Chemical">ace (fias) was determined as:

where tmes is the mesophyll thickness n class="Chemical">betweenpan> the two epidermal layers anpan>d ∑Ss is the sum of the cross-sectional areas of mesophyll cells.

The volume fractionpan> of inpan>tercellular air sppan> class="Chemical">ace of palisade tissue and spongy tissue was determined respectively as:

Where ∑Spal is the sum of the cross-sectional areas of n class="Chemical">palisade tissue cells, tpal is the pan> class="Chemical">palisade tissue thickness, ∑Sspo is the sum of the cross-sectional areas of spongy tissue cells, tspo is the spongy tissue thickness.

Chloroplast length (Lchl), chloroplast thickness (Tchl) and n class="Chemical">Tcw were obtained at different positions in each sample at ×30000 magnifications. For a given section, all characteristics were determined in at least three different fields of view, and at least three different sections were analysed.

The cross-section of a chloroplast is assumed to n class="Chemical">be oval. Therefore, the cross-section area of chloroplast (Arepan> class="Chemical">achl) was calculated in the palisade tissue or spongy tissue section as:

where π is the ratio of the circumference of a circle to its diameter.

g m modeled from anatomical characteristics

n class="Chemical">Accordinpan>g to the quanpan>titative one-dimensional gas diffusion model of Niinpan>emets anpan>d Reichsteinpan> (2003) further used pan> class="Chemical">by Tosens , mesophyll conductance of total leaf, palisade, spongy tissue and bract was estimated using the leaf anatomical characteristics (i.e. gm(anatomy)). In the model, gm(anatomy) is separated into gas phase conductance and liquid phase conductance (Evans ):

where gn class="Disease">ias is conductanpan>ce from substomatal cavities to outer surface of cell walls and gliq is the conductance from outer surface of cell walls to chloroplasts; R is the gas constant (Pa m3 K−1 mol−1), H is the Henry’s law constant (Pa m3 mol−1), and Tk is the absolute temperature (K). H/(R×Tk) is needed to convert gliq to a gas phase equivalent conductance (Niinemets and Reichstein, 2003).

The gas pn class="Chemical">hase conductanpan>ce (gpan> class="Disease">ias) was calculated as described in Niinemets and Reichstein (2003):

where ΔLias was takenpan> as pan> class="Chemical">half the mesophyll thickness (Niinemets and Reichstein, 2003), Da (m−2s−1) is the diffusion coefficient for CO2 in the gas phase (1.51 × 10–5 m−2 s−1 at 25 °C), and ς is the diffusion path tortuosity (m m−1) for which we used a default value of 1.57 m m−1 (Syvertsen ; Niinemets and Reichstein, 2003).

The total liquid pn class="Chemical">hase conductanpan>ce is provided pan> class="Chemical">by the sum of the inverse of serial conductances (Tosens ):

where the partial conductances are those for cell wall (n class="Chemical">gcw), plasmalemma (gpl), cytosol (gct), chloroplast envelope (gen), and chloroplast stroma (gst). gcw, gct, and gst were calculated as described in Tomás . Cell wall porosity (pcw) varied with Tcw according to Tosens (pcw=−0.3733×Tcw+0.3378). We used an estimate of 0.0035 m s−1 for gpl and gen (Tosens ). Conductance in units of m s−1 can be converted into molar units considering that:

where TL is the leaf temperature (°C) and P (Pa) is the air pressure.

Quantitative analysis of partial limitation of gm modeled

n class="Chemical">Accordinpan>g to Tosenpan>s , the limitations derived from differenpan>t componenpan>ts were calculated as:

where n class="Chemical">gm(anpan>atomy) is mesophyll conductanpan>ce estimated from anpan>atomical characteristics applying the model of Niinemets and Reichstein (2003) as modified by Tosens , Lias is the limitation derived from the gas phase component, Li is the component limitation in the cell wall, plasmalemma, cytoplasm, chloroplast envelope, and stroma, gi refers to the component diffusion conductance of the corresponding diffusion pathways.

Relative limitation analyses on AN

The relative limitation on AN was analysed in cotton leaves and n class="Chemical">brpan> class="Chemical">acts. According to Grassi and Magnani (2005), relative stomatal limitation (ls), mesophyll limitation (lm), and biochemical limitation (lb) were investigated in the cotton leaves and bracts. lm was calculated using gm calculated from gas-exchange plus fluorescence measurements following Harley ) (gm(Harley)), from anatomical characteristics applying the model of Niinemets and Reichstein (2003) as modified by Tosens (gm(anatomy)) and from the average value between the anatomy and Harley methods. The relative changes in light- saturated assimilation can be expressed in terms of parallel relative changes in stomatal and mesophyll conductance and in biochemical capacity as follows:

where gtot is total conductance to CO2 pan> class="Chemical">between the leaf surface and the sites of carboxylation (1/gtot=1/gs+1/gm); ls, lm, and lb are the corresponding relative limitation (0b). Here, was calculated as the slope of AN–Cc response curves over a Cc range of 50–100 µmol mol–1 (Tomás )

Chlorophyll content, mass per area and nitrogen content

The n class="Chemical">chlorophyll content of leaves anpan>d pan> class="Chemical">bracts was determined in eight leaf discs (0.186 cm2 each). Discs of the green organs were extracted in 80% (v/v) acetone for 24 h at room temperature in the dark. The absorbance of an extract was measured with a spectrophotometer, and the chlorophyll content was calculated according to Lichtenthaler (1987).

Leaf mass per unit area (LMA) is the ratio of dry weight anpan>d leaf area. Dry weight was determinpan>ed from ovenpan>-dried certainpan> area of leaf dipan> class="Chemical">scs after 48 h at ca. 80 °C. Leaf density was defined as LMA divided by leaf thickness.

For the measurement of n class="Chemical">nitrogen contenpan>t, leaves anpan>d pan> class="Chemical">bracts were harvested on the same day. Total nitrogen content of the dried tissues was determined according to the micro-Kjeldahl method (Schuman ).

Statistical analysis

Statistical analysis was performed with SPSS 17.0 for Windows (SPSS Inc., Chicago, IL, USA). All data were tested n class="Chemical">by anpan>alysis of varianpan>ce (ANOVA). The significanpan>ce of differences pan> class="Chemical">between treatment means was determined by the Student–Newman–Keuls (S-N-K) test at the 0.05 probability level. Data are presented as the means±standard error (SE) of three replicates.

Results

Difference in photosynthetic properties between cotton leaves and bracts

The net n class="Chemical">CO2 assimilation rate (AN), stomatal conductanpan>ce (gs), anpan>d mesophyll conductanpan>ce (pan> class="Chemical">gm) were significantly higher in leaves than in cotton bracts (Table 1). In cotton leaves the AN response to increasing Ci initially increased, then peaked, and finally remained stable above 600 μmol mol−1Ci. AN as a function of Ci in cotton bracts was lower than that in leaves (Fig. 2A). Relative to leaves, AN response to increasing PPFD in the bracts was minor and saturated at a lower irradiance (Fig. 2B). Both chlorophyll (a+b) content and the ratio between chlorophyll a and chlorophyll b (Chla/b) of cotton bracts were much lower than those in the cotton leaves (Table 2). The nitrogen content of bracts was 21% lower than that of leaves (Table 2). Larger Vcmax and Jmax derived from AN–Cc curves were observed in the leaf than in the bract. The electron transport rate from chlorophyll fluorescence (Jflu) calibrated by electron transport from gas exchange (JA) under 2% O2 conditions was close to Jmax based on the Cc. There was a difference in AN, AN-chl, and AN-N as a function of Cc between leaves and bracts (Fig. 3A–C).

Table 1.

Net assimilation rate (AN), stomatal conductance (gs), and mesophyll conductance (gm) estimated by three independent methods: using gas-exchange plus fluorescence measurements following Harley ), the curve-fitting method of Sharkey (2016), and using gm estimated from anatomical characteristics applying the model of Niinemets and Reichstein (2003) as modified by Tosens in total leaves, palisade and spongy tissue of leaves, and bracts

		A N (µmol CO2 m−2 s−1)	g s (mol H2O m−2 s−1)	g m(Harley) (mol CO2 m−2 s−1)	g m(Sharkey) (mol CO2 m−2 s−1)	g m(anatomy) (mol CO2 m−2 s−1)
Leaf	Total	37.26 ± 1.93a	0.62 ± 0.07a	0.37 ± 0.01a	0.48 ± 0.10a	0.33 ± 0.03a
	Palisade tissue	—	—	—	—	0.25 ± 0.02b
	Spongy tissue	—	—	—	—	0.14 ± 0.02c
Bract	Total	3.58 ± 0.28b	0.06 ± 0.00b	0.03 ± 0.00b	0.05 ± 0.02b	0.11 ± 0.01c

Values are means±SE. Different letters indicate significant differences at the 0.05 probability level.

Table 2.

Chlorophyll a+b (chl(a+b)), the ratio between chlorophyll a and chlorophyll b (Chla/b), nitrogen (N) content (%), maximum carboxylation rate (Vcmax), and maximum electron transport rate (Jmax) based on the chloroplastic CO2 concentration (Cc) and electron transport rate from chlorophyll fluorescence (Jflu) calibrated by electron transport from gas exchange (JA) under 2% O2 conditions

	Chl(a+b)(mg dm−2)	Chla/b(%)	N content(%)	V cmax-Cc (μmol m−2 s−1)	J max-Cc (μmol m−2 s−1)	J flu (μmol m−2 s−1)
Leaf	5.73 ± 0.25a	3.14 ± 0.21a	3.62 ± 0.44a	526.7 ± 65.0a	456.0 ± 68.0a	345.4 ± 24.3a
Bract	2.15 ± 0.17b	2.51 ± 0.08b	2.86 ± 0.07b	39.3 ± 9.0b	52.0 ± 7.9b	55.0 ± 0.5b

Values are means±SE. Different letters indicate significant differences at the 0.05 probability level.

Fig. 3.

Net CO2 assimilation rate expressed on the basis of area (AN) (A), chlorophyll (a+b) (AN-chl) (B), and nitrogen content (AN-N) (C) as a function of chloroplastic CO2 concentration (Cc) in cotton leaves and bracts. Values are means±SE.

Net assimilation rate (AN), stomatal conductance (gs), and mesophyll conductance (n class="Chemical">gm) estimated pan> class="Chemical">by three independent methods: using gas-exchange plus fluorescence measurements following Harley ), the curve-fitting method of Sharkey (2016), and using gm estimated from anatomical characteristics applying the model of Niinemets and Reichstein (2003) as modified by Tosens in total leaves, palisade and spongy tissue of leaves, and bracts

Values are means±SE. Different letters indicate significant differences at the 0.05 pron class="Chemical">bapan> class="Chemical">bility level.

n class="Chemical">Chlorophyll a+pan> class="Chemical">b (chl(a+b)), the ratio between chlorophyll a and chlorophyll b (Chla/b), nitrogen (N) content (%), maximum carboxylation rate (Vcmax), and maximum electron transport rate (Jmax) based on the chloroplastic CO2 concentration (Cc) and electron transport rate from chlorophyll fluorescence (Jflu) calibrated by electron transport from gas exchange (JA) under 2% O2 conditions

Values are means±SE. Different letters indicate significant differences at the 0.05 pron class="Chemical">bapan> class="Chemical">bility level.

Net n class="Chemical">CO2 assimilation rate expressed on the pan> class="Chemical">basis of area (AN) (A), chlorophyll (a+b) (AN-chl) (B), and nitrogen content (AN-N) (C) as a function of chloroplastic CO2 concentration (Cc) in cotton leaves and bracts. Values are means±SE.

At the low values found in n class="Chemical">brpan> class="Chemical">acts, the accuracy of the estimates of gm is low and the photosynthesis limitation analysis is very sensitive to small variation in any of its input parameters. Consequently, the results of the limitation analysis were completely different depending on which estimate we used (Fig. 4; Supplementary Fig. S2). Still, the Harley and the anatomy methods rely on completely independent assumptions (they have no single assumption in common), and yet both indicated low gm in bracts (see the anatomy method results in the next section). Because of the aforementioned accuracy problems, the absolute values, however, have to be taken with caution. The ‘real’ values would very likely be somewhere in between the two extremes represented by the Harley method on the one hand and the anatomy-based estimates on the other. For this reason we used the average of both methods to run the photosynthesis limitation analysis (Fig. 4). There was no significant difference between lm based on the average gm between the anatomy and Harley methods and ls in bracts (Fig. 4). However, lb was higher than ls and lm in bracts (Fig. 4). Leaves had the same level of ls, lm, and lb (Fig. 4).

Fig. 4.

Relative limitation analysis of photosynthesis in the leaves and bracts of cotton under normal ambient conditions. The total relative photosynthetic limitation was composed of stomatal (ls), mesophyll conductance (lm), and biochemical limitation (lb). lm was calculated using the average gm between the anatomy and Harley methods. Values are means±SE. Different letters indicate significant differences between ls, lm, and lb at the 0.05 probability level.

Relative limitation analysis of photosynthesis in the leaves and n class="Chemical">brpan> class="Chemical">acts of cotton under normal ambient conditions. The total relative photosynthetic limitation was composed of stomatal (ls), mesophyll conductance (lm), and biochemical limitation (lb). lm was calculated using the average gm between the anatomy and Harley methods. Values are means±SE. Different letters indicate significant differences between ls, lm, and lb at the 0.05 probability level.

Anatomical measurements of cotton leaves and bracts

In the C3 cotton leaves, two types of chlorenchyma are found, n class="Chemical">palisade tissue anpan>d spongy tissue. In order to compare the structural differences pan> class="Chemical">between leaf and bract, palisade and spongy tissue of leaf and bract were quantified separately. In leaves, mesophyll tissue is differentiated into palisade and spongy mesophyll, the palisade tissue being more compact with a lower porosity (fias; see Supplementary Fig. S3A, B). However, in the cotton bract, only one type of tissue was found, which was similar to the leaf spongy mesophyll (Fig. S3A, B). Although the LMA, leaf thickness (T) and density (D) in the cotton leaf were significantly higher than those in the bract (Table 3), we observed that there were no differences in the Sm/S, Sc/S, chloroplast thickness (Tchl) and Areachl between spongy tissue of leaves and bracts (Table 3). Bracts had higher fias and cell wall thickness (Tcw) than spongy tissue of leaves (Table 3). The spongy tissue of leaves and bracts also showed similar anatomical structure (Fig. S3). Quantitative limitations of gm modeled by anatomy were estimated according to the component diffusion conductance of the corresponding diffusion pathways (Fig. 5). The limitation derived from the gas phase components (Lias) (5–34%) was lower than the total limitation from liquid phase components. In the liquid phase, the palisade and spongy tissue of leaves revealed the highest limitation by the stroma (Ls) of around 50%. There was no significant difference in the Ls between palisade and spongy tissue of leaves and bracts. In bracts, cell walls accounted for up to 50% of the limitations, compared with only 11% and 16% in palisade tissue and spongy tissue, respectively. The limitations derived from plasmalemma (Lp) and chloroplast envelop (Le) in bracts were lower than those in spongy tissues, but were similar to those of palisade tissues of leaves.

Table 3

Leaf mass per unit area (LMA), leaf thickness (T), density (D), mesophyll thickness (Tmes), the surface of mesophyll cells and chloroplasts exposed to leaf intercellular air spaces (Sm/S and Sc/S; µm2 µm−2), the chloroplast thickness, the thickness of the cytoplasm between the cell membrane and the chloroplast (Tcyt), the cross-section area of chloroplast (Areachl), the volume fraction of intercellular air space (fias) and cell wall thickness (Tcw) in the cotton leaves and bracts

		LMA (g m−2)	T (μm)	D (g cm−3)	T mes (μm)	S m/S	S c/S	T chl (μm)	T cyt (μm)	Areachl (μm2)	f ias (%)	T cw (μm)
Leaf	Total	128.2 ± 0.5a	411 ± 10.2a	0.32 ± 0.00a	375 ± 6.6a	39.96 ± 2.3a	29.66 ± 2.5a	1.79 ± 0.28a	0.24 ± 0.07a	26.45 ± 4.1b	0.47 ± 0.02c	0.18 ± 0.01b
	Palisade tissue	—	—	—	195 ± 2.5c	28.9 ± 0.3b	22.6 ± 0.9b	2.22 ± 0.47a	0.34 ± 0.11a	37.81 ± 3.4a	0.33 ± 0.03d	0.20 ± 0.01b
	Spongy tissue	—	—	—	173 ± 4.2c	11.1 ± 2.1c	7.1 ± 0.9c	1.35 ± 0.15a	0.14 ± 0.09a	18.51 ± 2.1c	0.60 ± 0.01b	0.17 ± 0.01b
Bract	Total	53.1 ± 2.4b	350 ± 18.8b	0.15 ± 0.01b	300 ± 18.0b	14.6 ± 0.3c	7.4 ± 0.6c	1.12 ± 0.12a	0.13 ± 0.02a	15.65 ± 1.3c	0.75 ± 0.02a	0.45 ± 0.02a

Values were means±SE. Different letters indicate significant differences at the 0.05 probability level.

Fig. 5.

Quantitative analysis of partial limitation of mesophyll conductance modeled (gm(anatomy)) in the palisade and spongy tissue of leaves and the bracts. Lias, the limitation derived from the gas phase components; Lcw, the limitation derived from the cell wall; Lp, the limitation derived from the plasmalemma; Ls, the limitation derived from the stroma; Le, the limitation derived from the chloroplast envelope; Lc, the limitation derived from the cytoplast. Different letters indicate significant differences between palisade, spongy and bracts at the 0.05 probability level.

Leaf mass per unit area (n class="Chemical">LMA), leaf thickness (T), density (D), mesophyll thickness (Tmes), the surface of mesophyll cells and chloroplasts exposed to leaf intercellular air spaces (Sm/S and Sc/S; µm2 µm−2), the chloroplast thickness, the thickness of the cytoplasm between the cell membrane and the chloroplast (Tcyt), the cross-section area of chloroplast (Areachl), the volume fraction of intercellular air space (fias) and cell wall thickness (Tcw) in the cotton leaves and bracts

Values were means±SE. Different letters indicate significant differences at the 0.05 pron class="Chemical">bapan> class="Chemical">bility level.

Quantitative analysis of partial limitation of mesophyll conductance modeled (n class="Chemical">gm(anpan>atomy)) inpan> the pan> class="Chemical">palisade and spongy tissue of leaves and the bracts. Lias, the limitation derived from the gas phase components; Lcw, the limitation derived from the cell wall; Lp, the limitation derived from the plasmalemma; Ls, the limitation derived from the stroma; Le, the limitation derived from the chloroplast envelope; Lc, the limitation derived from the cytoplast. Different letters indicate significant differences between palisade, spongy and bracts at the 0.05 probability level.

Discussion

Lower AN in bracts than in leaves is due to co-limiting CO2 diffusion and biochemistry

It is well known tn class="Chemical">hat leaves are the mainpan> photosynthetic organpan>s inpan> planpan>t species, but numerous researchers have shown that non-leaf green organs are also an important source of assimilated carbon (Tambussi ; Redondo-Gómez ; Pengelly ; Hu ; Jia ; Zhang ) and make a considerable contribution to terrestrial carbon exchange. In the case of cotton, bracts also have a photosynthetic function and contribute to carbon gain (Zhang ; Hu , 2013). Moreover, it has been shown that some non-leaf green organs also have a strong stress tolerance, such as salt tolerance of rosette bracts (Redondo-Gómez ) and drought tolerance of cotton bracts (Zhang ) and wheat ears (Jia ). Hence, it is possible that, under abiotic stress conditions, non-leaf green organs make a considerable contribution to the carbon cycle.

Despite their importance, no previous study n class="Chemical">has focused on photosynthetic limitations anpan>d their anpan>atomical pan> class="Chemical">basis in cotton bracts. In our study, AN expressed on an area basis in bracts saturated at low irradiance (Fig. 2B; 3.58 μmol CO2 m−2 s−1), suggesting that light intensity was not the most important limiting factor for bract photosynthesis. It is likely that bracts can tolerate and thrive in low light intensity due to their growth in a shaded position over an evolutionary time of at least 1.1 million years since the appearance of tetraploid cotton (Hu ). A lower Chla/b (Table 1) in the bract may be also a long-term adaption to capture more light. Björkman (1981) also suggested controlling the Chla/b is one way to adapt the photosynthetic function to light.

The high values of photosynthesis observed inpan> cotton leaves inpan> this study (Table 1) were similar to those already reported for this species (Ephrath ; Faver and Gerik, 1996). Consequently, net CO2 assimilation rate (AN) values in cotton bracts were 90% lower than those obtained for leaves (Table 1; Fig. 2), which is accompanied by lower values of all the parameters related to photosynthesis. gs in bracts was only about 10% that of leaves, gm 8–36% (depending on which gm estimate was used), Vcmax 7.5% and J 11–16% (depending on whether considering Jmax or Jflu). In this study, gm was estimated by three independent methods that gave similar results for leaves, but this similarity was not found in bracts, with a significantly higher gm(anatomy) than gm(Harley) (Table 1). This may be partly due to the estimation biases of the currently available techniques. For instance, the variable J method was influenced by accuracy of Ci estimation (Gu and Sun, 2014) and (photo) respiratory CO2 recycling (Tholen ). We did our best to ensure the accuracy of Ci through calibration. Although the variable J method cannot rule out the effect of (photo) respiratory CO2 recycling, it is unlikely that this alone causes such a big difference in gm(Harley) between leaf and bract and between gm(anatomy) and gm(Harley). In addition, the estimation of gm(anatomy) is also subject to uncertainties. For instance, variable cell wall porosity was considered as a function of cell wall thickness (see ‘Materials and methods’ and Tosens ). But there was still a highly apparent discrepancy or inconsistency between gm(anatomy) and gm(Harley), and the difference in gm(anatomy) between leaf and bract could not account for the observed difference in photosynthesis (Table 1). The overestimation of gm(anatomy) in the bract could be because the anatomical model does not account for variations in some biochemical properties (e.g. the expression of aquaporins and carbonic anhydrase) that might be involved in the CO2 diffusion. In our experiment, a conservative constant value (0.0035 m s−1) was used to estimate the gpl and gen as suggested in Evans and Tosens ), but membrane permeability is affected by the expression of aquaporins and varies among different organs, species, and environments. In order to test the role of membrane permeability, we tried different values that have been reported in the literature (see Table 1 in Evans ). When gpl and gen in the bract were replaced by 0.0008 m s−1 (which is the permeability reported for yeast cells), gm(anatomy) was 0.069 mol CO2 m−2 s−1, i.e. much closer to gm(Harley). Therefore, membrane permeability can be another potential cause of (i) the huge difference in photosynthesis between leaf and bract, and (ii) the discrepancy between gm(anatomy) and gm(Harley). This is why, considering that the actual value may be somewhere in between the extremes of the estimations, we have used the average gm between anatomy and Harley values for the limitation analyses on photosynthesis. Both the similarity in the reductions of all the parameters related to photosynthesis (Table 1) and the relative limitation analysis (Fig. 4) confirmed that CO2 diffusion and biochemistry co-limit bract photosynthesis in a similar way. High biochemical limitation in bracts could be caused by a low Rubisco activity, as Bota proved its good agreement with Vcmax derived from AN–Cc curves. Bracts had very low gs, and high stomatal limitation would be likely due to the limited hydraulic capacity caused by the low main vein density.

Subcellular anatomical traits play important roles in setting gm of bracts

Leaf mass per unit area (n class="Chemical">LMA) is anpan> inpan>tegrative trait of leaf structural characteristics affecting gm. It is mainly dependent on leaf thickness and density (Niinemets, 2015). John reported that the number of cell layers and cell volume that is associated with leaf thickness and density are among the most important intrinsic drivers of LMA. Because leaf thickness and density are closely related to the Sc/S and Tcw, theoretically LMA has an important role in setting gm. Leaf thickness was 1.25 times larger than bract thickness and leaf density was 1.89 times larger than bract density (Table 3). These results suggest that higher density in the leaf mainly contributed to larger LMA. A lower proportion of mesophyll and a higher fias due to random cell arrangement and lower cell numbers led to lower density in bracts (Table 3; Supplementary Fig. S3). While early studies have shown that there is a negative relationship between LMA and gm across broad functional groups and within species (Flexas ; Niinemets ; Galmés ; Tosens ), this is not consistent with our results, which show that bracts have lower LMA than leaves despite having a higher Tcw (Table 3). Recently, Onoda highlighted that subcellular anatomical traits such as Tcw, Sm/S, and Sc/S are much more important than LMA in setting gm. Similarly, Peguero-Pina showed that these parameters as well as fias can mask the effects of LMA on gm.

n class="Chemical">CO2 diffuses from the inpan>tercellular air spaces to the sites of carboxylation within chloroplasts in gas phases largely affected by leaf porosity, reflected by fias (Hanba ), and liquid phases largely affected by Tcw, Sm/S, and Sc/S (Evans ; Hanba ; Terashima ; Tomás ; Peguero-Pina ). Evans concluded that gias is so large that it is not a major determinant of gm in leaves. Instead, Tcw, Sm/S, and Sc/S, which affect gliq, are considered the main determinants of differences in gm among species (Terashima ; Tomás ; Peguero-Pina , 2017). Bracts with thin mesophyll thickness had smaller Sm/S and Sc/S than leaves (Table 2). Several studies have indicated larger Sm/S and Sc/S in thicker leaves (Hanba ; Terashima ; Peguero-Pina ), likely reflecting the more developed palisade tissues in the thicker leaves. The smaller Sm/S and Sc/S were also likely due to higher fias that was caused by the fewer and smaller cells in bracts. Based on this, we quantified separately the anatomical structure of palisade and spongy tissues in leaves (Table 3). The quantitative results indicated that Sm/S and Sc/S of bracts were similar to the spongy tissue of leaves, which contributed to lower gm in bracts and the spongy tissue of the leaves. In addition, chloroplast size and thickness are also an important factor limiting CO2 diffusion to Rubisco (Tomás ; Tosens ; Veromann-Jürgenson ), and thus smaller Areachl in bracts and the spongy tissue of the leaves was also a cause of lower gm. Although numerous studies reported that Tcw is generally higher in woody species with thick leaves (Evans ; Terashima ; Tosens Veromann-Jürgenson ), higher Tcw was observed in bracts than in either the palisade or spongy tissues of leaves. This is possibly due to a larger nitrogen investment in structural construction (i.e. the cell wall construction) in bracts than in leaves, which is supported by lower AN-N expressed on the basis of nitrogen content in bracts (Fig. 3C). In this sense, relative to leaves, a larger nitrogen investment in cell wall construction led to lower gm in bracts. Instead, mesophyll conductance of bracts was similar to that of spongy tissues, the highest values being those of palisade tissue. The further anatomical limitation analysis (Fig. 5) showed that the limitation derived from the cell wall (Lcw) is higher in bracts than in spongy tissue, but the cumulative effects of subtle differences in each subcellular structure, especially the chloroplast traits, compensate Lcw to yield similar gm in bracts and spongy tissue. However, there was smaller gm in bracts than in palisade tissue because of the larger Tcw, greater fias, smaller chloroplasts, and lower Sm/S and Sc/S.

Conclusion

In summary, net n class="Chemical">CO2 assimilation rate (AN) was lower inpan> cotton pan> class="Chemical">bracts than in leaves, and this was due to concomitant and similar limitations by biochemistry, stomatal conductance (gs), and mesophyll diffusion conductance (gm). Concerning gm, we provide the first report showing that anatomical traits setting the limits for gm in leaves also operate in non-leaf photosynthetic tissues like bracts. Specifically, larger Tcw and fias, smaller and fewer chloroplasts, lower Sm/S and Sc/S, and, perhaps, smaller membrane permeability in bracts than in leaves led to lower gm. It has been shown that in leaves of angiosperm species (Flexas, 2016), but especially in crops (Nadal and Flexas, submitted), stomatal, mesophyll conductance and biochemical limitations to photosynthesis are of similar magnitude, for which significantly improving leaf photosynthetic capacity in crops cannot be achieved unless all three factors are improved (Flexas, 2016). Here we show that this might be similar in bracts. Since bracts contribute significantly to the photosynthetic carbon gain of plants (Zhang , Hu , 2013), the present results should be considered in future attempts to improve crop productivity by means of manipulating photosynthesis.

Supplementary data

Supplementary data are availan class="Chemical">ble at JXpan> class="Chemical">B online.

Fig. S1. Daily maximum and minimum air temperature and precipitation during the growing season at the experimental field.

Fig. S2. Relative limitation analysis of photosynthesis for leaves and brpan> class="Chemical">acts of cotton under ambient conditions.

Fig. S3. Light and electron micron class="Chemical">scopy images of cotton leaves anpan>d pan> class="Chemical">bracts.

Click here for additional data file.