Anatomical measurements in maize and rice
The partial volume of the leaf represented by vasculature (midrib and vascular bundles in the blade), veins (both including and excluding the bundle sheath cells), and xylem were not significantly different across genotypes in either maize or rice, except for the partial volume of veins including the bundle sheath in rice (Table 1 and 2). In rice, the partial volume of veins including the bundle sheath was slightly greater in WT (0.22 ± 0.02) than cslf6-1 (0.18 ± 0.01) and cslf6-2(0.19 ± 0.01), but there was no difference in anatomy with growth light intensity (Table 2). The partial volume of total vasculature, veins (both including and excluding the bundle sheath), and xylem in maize (0.50 ± 0.01, 0.25 ± 0.01, 0.11 ± 0.01, 0.03 ± 0.01, respectively) and in rice (0.53 ± 0.04, 0.20 ± 0.03, 0.08 ± 0.01, 0.01 ± 0.02, respectively) showed that about half of the leaf was vasculature, and only a small fraction of the vasculature was conducting tissue (Table 2).
In maize and rice, stomatal density (Sd) differed with growth light intensity where plants under high light had greater Sd (117.2 ± 1.5 and 737 ± 35, respectively) than those grown under low light (88.7 ± 1.4 and 528 ± 31, respectively) (Table 1 and 2). The Sd was slightly higher in leaves of Zmasft double mutants (105.7 ± 3.4) than wildtype (99.8 ± 3.9), but there were no differences across rice genotypes. Sd on both adaxial and abaxial sides of the leaf was significantly higher under high light, but neither was significantly different across genotypes (Table 1, S1). In maize, abaxial stomata subsidiary cells length was greater in wildtype grown under low light intensity (34.5 µm ± 2.0) than all other conditions (25.7 to 26.6 µm; Table 1, S1).
Changes in Δ18OLW between plants grown in high and low light (HL-LL) co-varied with changes inL v, E , Sd, Sdabaxial, and partial volume of veins excluding the bundle sheath (P < 0.05) and partial volume of the total vasculature andg s (P = 0.05; Table S2). The mean genotypic difference (HL-LL) in Sd and E formed significant linear relationships with Δ18OLW (Fig. 5a, b), while E and Sd also formed a linear relationship (Fig. 5c). In contrast, the change in Δ18OLWbetween plants grown at high and low light did not co-vary with changes in Δ18Oe, mesophyll and bundle sheath cell wall thickness (bordering the IAS), leaf thickness, mesophyll cell layers, volume fraction of intercellular air spaces (IAS) per mesophyll area, mesophyll surface area exposed to IAS per unit leaf area (S mes), surface area of chloroplasts exposed to IAS per unit of leaf area (Sc), and leaf dry mass per area [Table S2, these data were previously published in Ellsworth et al. (2018)].
Mesophyll and xylem P éclet models in maize and rice
All genotypes of maize and rice formed significant positive relationships between f sw and transpiration rate (E ), suggesting a Péclet effect in these genotypes (Fig. 6a,b). When f sw and E were fitted to the mesophyll and xylem Péclet model (Eqn. 10) to solve for the following three variables: mesophyll L (L m), xylemL (L v), and the proportion of xylem water (ϕ x), the model in maize converged, but the estimated value for L m was essentially 0 (2 x 10-17). In rice the model failed to converge on a significant value for L m. Considering thatL m was estimated as ~0 in maize and did not converge on a significant value in rice, it was assumed that the mesophyll Peclet was nonexistent, and the location of the Peclet effect was in the xylem only (Eqn. 11) for maize and rice. In maize, the xylem Peclet model failed to converge on significant values ofL v and ϕ x when all data were used. However, when only data from 80 % RH were examined,Zmasft had a larger estimate of L v (186 ± 86 mm) than wildtype (113 ± 47 mm; Table 3). At 50 % RH, the xylem Péclet model did not converge. In rice, L vcalculated from the xylem Péclet only model was smaller incslf6-1 than wildtype and cslf6-2 . In each species, estimates of ϕ x were similar across genotypes, but ϕ x was twice as large in rice than maize (Table 3).
L v was estimated using the xylem Peclet model when the partial volumes of total vasculature, veins including and excluding BS were used as ϕ x in maize and only when the partial volume of total vasculature was used as ϕx in rice. When the partial volume of leaf vasculature or vascular bundles was used as ϕ x in the xylem Péclet model, estimates of L v were greater forZmasft double mutants than WT, and this difference was most apparent at low light intensity (Table 1, 3). In rice,L v was lower in cslf6-1 than WT andcslf6-2 . These results were similar to modelledL v (solving for both L vand ϕ x), except that L vwas considerably smaller when partial pressure of the total vasculature was used as ϕx in maize. The partial volume of veins, including the bundle sheath, was most similar tof sw (especially at high light intensity) andϕ x values estimated from the xylem Péclet model in maize (Table 3). Rice, having larger f swvalues than maize, the partial volume of the total vasculature of rice was most similar to f sw andϕ x.
A sensitivity analysis of the interaction betweenL v and ϕ x in the xylem Péclet model showed that ϕ x in rice was greater than in maize and that L v had a large effect onϕ x until it reached its minimum threshold in both maize and rice at approximately 75 mm (Fig. 7). Genotypic differences were only apparent at low light, where L v andϕ x were greater in Zmasft than in wildtype. In rice, L v andϕ x were lower in cslf6-1 than wildtype andcslf6-2 when L v was greater than approximately 50 mm. The results between maize at 50 % RH and rice were similar in that ϕ x was greater at high light than low light across the range of L v (Fig. 7).