2.3.2. Instantaneous Measurement of rPPFD

In addition to measuring the diurnal course of PPFD for the selected leaves, we measured instantaneous PPFDs for all 62 leaves to estimate the light environment at the two sites. The measurements were conducted around midday on an overcast day (20 June 2020). We used two quantum sensors (MIJ-14PAR) and simultaneously measured the PPFD incident on each leaf and on top of an approximately 2 m tripod placed at the open place. One sensor at the tripod was fixed horizontally. The other was placed at a tip of a hand-held measuring bar [4,72] and was inclined for each leaf to measure the PPFD at the bottom surface of the cone- or funnel-shaped leaf (Figure 4). Each sensor was connected to a voltage logger (LR5041), and the voltages of the two sensors were recorded simultaneously. We calculated the relative PPFD (rPPFD) as the ratio of the PPFD on each leaf to the PPFD at the top of the tripod.

**Figure 4.** A schematic diagram of instantaneous relative photosynthetic photon flux density (rPPFD) measurement with a hand-held measuring bar (illustration by Kohei Koyama).

#### *2.4. Leaf Thickness and LMA*

On 4 and 7 July 2020, we sampled pieces (ca. 40 cm2) from the lamina edges of 41 leaves (clearing, *n* = 20; understory, *n* = 21), measured their thickness with a digital caliper (CD-15PSX (resolution: 0.01 mm); Mitutoyo Corp, Kawasaki, Japan), and scanned them immediately after sampling with an A4 flatbed scanner (CanoScan LiDE 220; Canon, Tokyo, Japan). These samples included all 12 leaves for which photosynthetic rates were measured, and the lamina samples were taken at the same positions on the leaves as the photosynthetic measurements (Figure 3). The lamina parts were selected to avoid the thickest leaf veins. The laminae were then oven-dried at 70 ◦C for at least one week, and their dry mass was measured with a precision balance. The projected area of each piece of lamina was measured with the Image J software (NIH, Bethesda, MD, USA) [73]. We calculated leaf mass per area (LMA, g m−2) as the ratio between the dry mass and the area of one side of each sampled piece of lamina [74]. We calculated mass-based values by dividing the area-based values by the LMA of that leaf [2].

#### *2.5. Calculation of Daily Photosynthesis*

We calculated the instantaneous net photosynthetic rate for each target leaf every 10 min for 24 h using the estimated light response curves (Equation (1)) and the diurnal course of incident PPFD. At night, the dark respiration rate (*R*area) was used as the nighttime respiration rate. By integrating these, we calculated the daily net photosynthesis for each target leaf.

#### *2.6. Simulation 1: Exchanged Leaves*

We performed a simulation in which we hypothetically exchanged the photosynthetic light response curves between the two sites. We expected that if the difference in the photosynthetic traits between the sites was the acclimation to a local light environment, then the exchange of leaves would reduce the daily net photosynthesis for both sites. For each understory leaf, we replaced the photosynthetic light response parameters (*P*g\_max\_area, *R*area, Φ, and θ in Equation (1)) with the mean photosynthetic parameters obtained in the clearing and vice versa. Next, we calculated the daily net photosynthesis for each hypothetical leaf by using the actual incident PPFD on each leaf. We also calculated the critical PPFD for each understory leaf. This value was defined as the lowest PPFD, such that if instantaneous PPFD exceeded that value, the net photosynthetic rate at that moment would be higher for the hypothetically set clearing leaf than for the actual understory leaf in question. This value was used to investigate how often PPFD exceeded this value due to sunflecks.

#### *2.7. Simulation 2: Understory without Sunflecks*

To evaluate the significance of sunflecks in the understory, we performed another simulation in which sunflecks were hypothetically removed from the original dataset of the diurnal course of PPFD on the sunny day (July 3). If PPFD at a given moment exceeded 200 μmol m−<sup>2</sup> s<sup>−</sup>1, that PPFD value was replaced by a fixed value of 200 μmol m−<sup>2</sup> s<sup>−</sup>1. This value was approximately equal to the maximum PPFD (199.72) observed in the understory on the overcast day (June 24). We also observed that the diel cycle of PPFD in the understory did not exceed 200 μmol m−<sup>2</sup> s−<sup>1</sup> on the sunny day except during sunflecks (see Results), so that the background diel cycle of PPFD was retained by this simulation. We then calculated the daily net photosynthesis in the understory without sunflecks, either with (1) actual understory leaves, or (2) the hypothetically set clearing leaf (described above).

#### *2.8. Statistical Analysis*

All statistical analyses were performed with the statistical software R v4.0.3 (Vienna, Austria) [75] and packages ("cowplot" [76], "ggbeeswarm" [77], "ggplot2" [78], and "lme4" [79]). The results were compared between sites by fitting generalized linear models (GLM) using the function *glm* (family = Gamma (link = "log")), except for the simulation results. The differences in the simulation results

obtained under different scenarios were tested using a generalized linear mixed model (GLMM), treating individual leaves (i.e., diurnal courses of PPFD on different leaves) as random intercepts, and using the function *glmer* (family = Gamma (link = "log")) [79], except in one case (simulated clearing leaves in the understory on the overcast day), in which simulated net daily photosynthesis values contained a negative value. For that case, a linear mixed model (LMM) was fit with the function *lmer* [79]. In all cases, the significance of the fixed effect was tested using the likelihood ratio test with the function *anova* (test = "Chisq").

#### **3. Results**

#### *3.1. Leaf Shape*

The differences in the light environment were quantified by large differences in rPPFD and daily light integral between the two sites (Table 1). Lamina openness angle was significantly larger in the understory (U) than in the clearing (C) (*p* < 0.01, Table 1; Figure 5a,b), indicating that the three-dimensional arrangement of leaf laminae was flatter in the understory (see photographs in Figure 1a,b). Compared with the difference in shape, the difference in lamina size was relatively small: *d*high was slightly larger in the understory than in the clearing (*p* = 0.025), and neither *d*opp (*p* = 0.37) nor the petiole length (*pl*) (*p* = 0.28) significantly differed between the sites (Table 1; Figure 5c–e).


**Table 1.** Leaf traits in the clearing (C) and in the understory (U).

<sup>1</sup> Same values as *R*area.

**Figure 5.** (**a**–**e**) Leaf shape parameters in the understory (U) and in the clearing (C). Each closed circle indicates one leaf (i.e., bee swarm plot).

#### *3.2. Area-Based Photosynthetic Traits*

The area-based net photosynthetic rate at PPFD = 2000 μmol m−<sup>2</sup> s−<sup>1</sup> (*P*net\_2000), the maximum gross photosynthetic rate (*P*g\_max\_area), dark respiration rate (*R*area), and light compensation point (LCP) were all significantly higher in the clearing than in the understory (*p* < 0.01) (Table 1; Figure 6a–d). Neither the initial slope (Φ) nor the convexity (θ) significantly differed between the sites (*p* = 0.48–0.49) (Table 1; Figure 6e,f).

**Figure 6.** (**a**–**f**) Area-based photosynthetic parameters of the sun leaves in the clearing (C) and the shade leaves in the understory (U). Each closed circle indicates one leaf. LCP: light compensation point.

#### *3.3. LMA and Leaf Thickness*

The mean LMA of the leaves in the clearing leaf was 1.87 times larger than that of the leaves in the understory (*p* < 0.01) (Table 1; Figure 7a). Additionally, the mean leaf thickness was 1.23 times larger than that of the leaves in the understory (*p* < 0.01) (Table 1; Figure 7b).

**Figure 7.** (**a**) Leaf mass per unit area (LMA) and (**b**) lamina thickness of the sun leaves in the clearing (C) and the shade leaves in the understory (U). Each closed circle indicates one leaf. The thicknesses were measured with a resolution of 0.01 mm.

#### *3.4. Mass-Based Photosynthetic Traits*

In contrast to the high plasticity in the area-based rates, no significant difference was found for mass-based photosynthetic and respiration rates between the sites (*p* = 0.69–0.96) (Table 1; Figure 8a,b).

**Figure 8.** Mass-based (**a**) photosynthetic rates and (**b**) dark respiration rates of the sun leaves in the clearing (C) and the shade leaves in the understory (U). Each closed circle indicates one leaf.

#### *3.5. Diurnal Courses of PPFD*

Figure 9 shows the diurnal courses of PPFD incident on the leaves. The estimated critical values were 59–161 μmol m−<sup>2</sup> s−<sup>1</sup> (median: 111.5; these values were calculated for all the six understory leaves for which photosynthetic parameters were measured, and the diurnal course of PPFD was measured for four of them, as shown in Figure 9b,d). On the sunny day in the understory, instantaneous PPFD often exceeded the critical values due to sunflecks (Figure 9b).

**Figure 9.** Diurnal course of light intensity (PPFD) incident on each leaf at each site (the clearing and understory) on a sunny day (3 July 2020) and on an overcast day (24 June 2020). (**a**) Clearing on the sunny day, (**b**) understory on the sunny day, (**c**) clearing on the overcast day, and (**d**) understory on the overcast day. Data for Leaf 1 on the panel (**a**) were not obtained owing to a measurement failure. The red horizontal lines on the understory panels indicate the critical values of PPFD; if PPFD at one moment exceeded that value, the instantaneous net photosynthetic rate would be higher for a typical clearing leaf than for the understory leaf in question.
