The photosynthesis measurements were conducted under the same conditions in experiments 1, 2, and 3. Fully-expanded mature leaves were used for the measurements. Photosynthesis measurements were performed using a photosynthesis system, LI-6400XT (Li-Cor Biosciences, Lincoln, NE, USA). Measurements were performed at a leaf temperature of 25 °C, a flow rate of 500 ml min^–1, and PPFD (photosynthetic photon flux density) of 1500 μmol m^–2 s^–1. Unless otherwise stated, measurements were taken at VPD 1.0–1.3 kPa. For the estimation of the photosynthetic parameters, maximum Rubisco carboxylation rate (V_cmax) and maximum electron transport rate (J_max), the CO₂ concentration was changed to 400, 300, 200, 100, 60, 400, 500, 600, 800, 1000, 1500, and 1900 μmol mol^–1, to obtain A/C_i curves. Curve fitting was performed using R package, “plantecophys” (Duursma, 2015), with the same assumptions about Rubisco specificity factors such as K_c, K_o and Γ^* for G. biloba and R. pulchrum. The photosynthesis rate and stomatal conductance at a CO₂ concentration of 400 μmol mol^–1 were abbreviated as A₄₀₀ and g_s400, respectively.

We re-analyzed the data obtained in 2014 and 2017, which was a subset of our published dataset (Matsumoto et al., 2022; Matsuura et al., 2025). In this subset data, branches of G. biloba and R. pulchrum were collected from the same four study sites in Kyoto city, S1 and S2 in 2014 and S3 and S4 in 2017 (Figure 1A). The sampling method was described in the previous study. Briefly, for each study site, the sunlit branches of 3–4 G. biloba and R. pulchrum trees planted along the roads were collected from July to November. They were brought back to the laboratory and then leaf photosynthesis measurements were performed using the method described in 2.1. In 2014, collection of the branches was conducted 2–4 times per year (3–4 leaves for each sampling time, n = 12–16 for each site). In 2017, collection of the branches was conducted in September (two leaves for each tree, n = 8 for each site). The anatomical data, stomatal density, stomatal pore length, and mesophyll thickness were obtained by re-analysis of light micrographs of the stomata and leaf sections of 5–6 leaves obtained in 2014. Data on leaf nitrogen content were obtained from Kiyomizu et al. (2019). The anatomical analyses were performed for the leaves in which photosynthesis was measured. For the stomatal density measurements, two randomly-selected areas of 200µm × 200 µm from five images were used, counting the stomata number (n = 10). For the measurement of stomatal pore length, four randomly-selected stomata from five images (n = 20) were used. Leaf thickness was measured by selecting four randomly-selected locations from five images of leaf section (n = 20).

The data of the concentration of atmospheric pollutants were obtained from a public database provided by the Japan National Institute for Environmental Studies (http://www.nies.go.jp/igreen/td_down.html). For the data of the atmospheric pollutants, monthly averaged data from April to November were used. For obtaining the monthly averaged data, daily mean values were used for ozone (O₃), nitrogen oxide (NO), nitrogen dioxide (NO₂), and suspended particulate matter (SPM), while for particles that are 2.5 µm or less in diameter (PM_2.5), the mean of the daily maximum values were used. The meteorological data were obtained from AMeDAS provided by the Japan Meteorological Agency (https://www.data.jma.go.jp/stats/etrn/index.php).

Four seedlings of G. biloba and R. pulchrum, 30 cm in height, were purchased from a commercial nursery (Kumamoto Greenery Center). The seedlings were grown in pots with a diameter of 13.0 cm and a height of 11.3 cm, filled with a soil mixture of culture soil, Akadama (pumice lava), and perlite at a ratio of 11:11:3. Fertilizer (Hyponex, Hyponex Japan) diluted 500 times with water was applied once a week, 100 ml per pot. The seedlings were grown in a greenhouse at the campus of Kyoto Institute of Technology, with the temperature set at 25 °C. A data logger (HOBO pro v2 U23-002, Onset, Bourne, MA) was placed 1.6 m above the ground in the glasshouse to record temperature and relative humidity every 30 minutes.

We conducted a preliminary experiment using 50, 100, 150 and 200 mM NaCl. We chose 50 mM NaCl because the photosynthesis of R. pulchrum decreased to nearly zero at concentrations above 100 mM NaCl. Before the salt treatment (control), the plants were sufficiently watered for three weeks, that is, watered with 150 ml of tap water every other day. For the salt treatment of three weeks, 150 ml of 50 mM NaCl solution was supplied per pot every other day. Images of the whole plant were taken using a digital camera (PowerShot SX 210 IS, Canon, Tokyo and TOUGH TG-5, OLYMPUS, Tokyo).

Fully-expanded undamaged leaves were selected per individual and the photosynthesis rate was measured twice for each leaf using the method described in 2.1 (n = 8). To estimate the chlorophyll content of the leaves, we measured the SPAD value, which is an indicator of chlorophyll content, using a chlorophyll meter (SPAD-502Plus, Konica Minolta, Tokyo) for the leaves used for gas exchange measurements (n = 4). The SPAD value was measured at six places on each leaf, excluding the leaf veins, and the average value was calculated.

To measure stomatal density, a silicone rubber (KE-14, Shin-Etsu Silicone, Tokyo) and a catalyst (CLC-229, Shin-Etsu Silicone) were mixed and applied to the lower surface of fully-expanded mature undamaged leaves to create a primary replica. A secondary replica of the leaf was made using nail varnish, and then stomata images of the secondary replicas were taken using a digital camera (DP22, Olympus) with a light microscope (BX51, Olympus). Stomatal density and stomatal pore length were measured in four leaves from different individuals before the salt treatment. One image per leaf was used to measure stomatal density (n = 4), and five stomata per image were used to measure stomatal pore length (n = 20).

To measure the amount of Na in the leaves, the dried leaves were ground, and then 25 ml of distilled water was added per 0.5 g of dried leaves. After centrifugation (14000 rpm) for 15 minutes, 10 ml of the supernatant was filtered through a Stericup-GP (filter pore size 0.22 μm, Merck, German), and then the amount of Na was measured twice (n = 8) using an ICP emission spectrometer (ICPE-9820, Shimadzu, Kyoto) at the Center for Environmental Science, Kyoto Institute of Technology.

To evaluate the changes in leaf hydraulic conductance, leaf water potential was measured using a pressure chamber (Model 600 Pressure Chamber Instrument, PMS Instrument Company, Oregon, USA) in four leaves for each species. Leaf water potential at predawn (Ψ_predawn, 03:00-05:30) and during the day (Ψ_midday, 11:30-12:30) were measured using one fully-expanded mature undamaged leaf each from four individuals of each tree species. The leaf hydraulic conductivity (K_leaf, mmol MPa^–1 m^–2 s^–1) was calculated using the following equation (Sperry and Pockman, 1993).

where E (mmol m^–2 s^–1) is the leaf transpiration rate. We assumed that the E values at predawn were zero. We obtained E values twice at midday using leaf gas exchange measurements with a Li-6400 XT (n = 8 for K_leaf). The leaf temperature was 25 °C, the flow rate was 500 ml min^–1, PPFD was 1500 μmol m^–2 s^–1, VPD was 1.0 kPa, and CO₂ concentration was 400 μmol mol^–1.

In experiment 1, that is, the air pollution investigation, we conducted ANOVA and Holm’s post-hoc test, using monthly mean data of the pollutants, to compare the air pollutant levels between low- and high-pollution sites for G. biloba and R. pulchrum. First, we first calculated the average leaf trait values for G. biloba and R. pulchrum in low-pollution sites 2014 and 2017 separately. Then, we calculated the ratio relative to the low-pollution area for all data from both years. This minimized the impact of differences in climate conditions between the two years. Finally, we pooled the data from both years and performed an ANOVA and a Holm’s post-hoc test on the trait ratios for G. biloba and R. pulchrum.

In experiment 2, that is, 50 mM NaCl treatment, we measured the traits using the same tree for both the control and the NaCl treatment. We calculated the increase in concentration of leaf Na and decrease in K_leaf for each individual tree after three weeks of 50 mM NaCl treatment. These values were compared between G. biloba and R. pulchrum, using Welch’s t-test. The ratios of the leaf traits of 50 mM-NaCl treatment to those of the controls were calculated for each individual tree. The differences in the leaf traits between G. biloba and R. pulchrum were tested using Welch’s t-test. We first performed ANCOVA for the relationships between leaf traits, considering “species” as a covariate. The effect of species was not statistically significant (p > 0.1) for the relationship between A₄₀₀ and g_s400, so we pooled the data for the two species and performed a regression analysis. However, for the relationship between g_s400 and K_leaf, the species effect was statistically significant (p< 0.05). Therefore, so we performed separate regression analyses for the two species.

In experiment 3, that is the long-term excess humidity experiment for G. biloba, the ratios of the leaf traits at excess humidity and recovery were calculated with respect to the traits under the control condition, which is similar to experiment 1. ANOVA and Holm’s post-hoc test were used to analyze trait ratios, as well as time-dependent changes in the decrease of A₄₀₀ and g_s400.

For all statistical analyses, the outlier values were eliminated using the Kolmogorov–Smirnov test. We performed statistical analysis using R version 4.5.0 and R Commander Version 2.9-5.

High atmospheric pollution sites had higher concentrations of NO₂, NO, and SPM than low atmospheric pollution sites, while PM_2.5 and O_x showed no difference between high and low pollution sites (Figure 1A, Table 1). Although many previous studies reported negative impacts of O₃ on the photosynthesis of urban trees (Xu et al., 2015; Gao et al., 2016; Hoshika et al., 2020), the variation in O_x in Kyoto city is very small (Matsumoto et al., 2022), in which changes in O_x levels are independent of the changes in other air pollutants (Kiyomizu et al., 2019). In Kyoto city, concentrations of air pollutants other than ozone are strongly correlated with traffic volume (Kiyomizu et al., 2019; Matsumoto et al., 2022). It is likely that multiple air pollutants emitted from vehicles collectively affect the photosynthesis of urban trees in Kyoto city. Several studies reported toxic effects of atmospheric pollutants NO₂ and NO on photosynthesis. Research on Arabidopsis thaliana showed that exposure to NO₂ leads to oxidative stress and functional impairment, which enhances the expression of genes associated with the production and metabolism of harmful reactive oxygen species (Antenozio et al., 2024). In the Populus species, exposure to NO₂ significantly reduces photosynthesis, which is related to partial stomatal closure and changes in chloroplast ultrastructure (Hu et al., 2015). In the present study, the negative effect of atmospheric pollution on the photosynthetic traits A₄₀₀, g_s400, V_cmax, and J_max was significantly different between Ginkgo biloba and Rhododendron pulchrum (Figure 1B; Supplementary Table S1). High levels of atmospheric pollutants increased these traits for G. biloba but overall decreased them for R. pulchrum. This suggests that photosynthesis in G. biloba had a high tolerance to atmospheric pollution, which supports our previous studies (Kiyomizu et al., 2019; Matsumoto et al., 2022; Matsuura et al., 2025).

The high tolerance of photosynthesis to atmospheric pollution in G. biloba should be related to the anatomical traits of the leaves that mitigate the harmful effects of air pollutants. First, mesophyll thickness of G. biloba was 1.6–2.4-fold compared to R. pulchrum (Figure 1C, Table 2). Furthermore, high atmospheric pollution increased mesophyll thickness of G. biloba (Figure 1B). A large mesophyll thickness would increase the path length and resistance for the diffusion of air pollutants to the upper mesophyll tissues (Kiyomizu et al., 2019). Second, the stomatal density of G. biloba was only 0.27–0.43-fold of R. pulchrum (Figure 1D, Table 2). Low stomatal density and sunken stomata from the leaf lower epidermis in G. biloba (Figure 1D) would reduce the influx of atmospheric pollutants into the leaves (Kiyomizu et al., 2019). It should be noted that G. biloba had larger stomatal pore length, 1.8-fold that of R. pulchrum (Table 2, Figure 1D), which potentially increases the stomatal conductance and thus diffusion of air pollutants into the leaves. There is a strong negative correlation between stomatal density and stomatal size (Franks and Beerling, 2009), so it is not easy to determine which is the more decisive factor for stomatal conductance. In the introgression lines of Solanum lycopersicum, stomatal pore size rather than stomatal density is reported to be key for the genetic variation of stomatal conductivity (Fanourakis et al., 2015). In contrast, low stomatal density relates to the low stomatal conductance in Arabidopsis taliana (Tulva et al., 2024) and in a variety of rice plant (Caine et al., 2023). In the low pollution site in the present study, g_s400 of G. biloba is only 0.6 times that of R. pulchrum (Table 2). This suggests that, in terms of gas conductivity through the stomata, the effect of lower stomatal density is more pronounced than that of larger stomatal pore size in G. biloba.

Unlike with the photosynthetic traits, for both tree species, pollution levels alter neither the leaf chlorophyll content (SPAD value, Supplementary Table S1) nor leaf nitrogen content (Table 2). G. biloba had significantly higher SPAD and leaf nitrogen content than R. pulchrum at both high and low pollution sites (Supplementary Table S1; Table 2). However, G. biloba did not exhibit higher photosynthetic rates than R. pulchrum at either S1 or S2 sites (see Supplementary Table S1), suggesting that high leaf nitrogen content does not correlate with high photosynthetic rates. Research in Chinese cities reported increased chlorophyll levels in urban areas with elevated NO_x, suggesting that NO_x absorbed by leaves contributes nitrogen for chlorophyll biosynthesis (Yan et al., 2024; Zhang et al., 2024). However, this response was less possible in our study, because no increase in leaf nitrogen content in areas with higher pollution was obtained (Table 2).

A 50 mM NaCl treatment, that is, moderate salinity, resulted in partial leaf falls in both G. biloba and R. pulchrum (Figure 2A), which indicates that 50 mM salt treatment imposed harmful effect on the leaves of both species. On the other hand, accumulation of Na in the leaves after salt treatment was much lower in G. biloba than in R. pulchrum (Figure 2B, Table 3). The decrease in K_leaf was also lower in G. biloba than in R. pulchrum (Figure 2B, Table 3). Accordingly, the degree of the harmful effects of salt treatment on leaf photosynthesis was smaller in G. biloba than in R. pulchrum; G. biloba exhibited a smaller decrease in the photosynthesis rate, 25% reduction in A₄₀₀, compared to R. pulchrum, which showed a 52% decrease in A₄₀₀ (Figure 2C).

What is the physiological mechanism that causes the smaller decline in photosynthesis under salt stress in G. biloba? The smaller decline in g_{s_400} under salt treatment in G. biloba (Figure 2C), together with the strong linear correlation between A₄₀₀ and g_s400 (Figure 2D), suggest that stomatal regulation is key for the different response of photosynthesis to salt stress between G. biloba and R. pulchrum. Then, why is g_s400 of G. biloba less responsive to salt stress than R. pulchrum? Although the lower stomatal density and large stomatal pore size of G. biloba may impose an opposite effect on stomatal conductance (Table 3), as previously explained, in terms of gas conductivity through the stomata, the effect of lower stomatal density should be more pronounced than that of larger stomatal pore size in G. biloba. In strawberry cultivars, low stomatal density is correlated with a low sensitivity of stomatal conductance to salt stress (Orsini et al., 2012). Low stomatal density also induces low transpiration, and thus, involves small K_leaf as well as low g_s. Such a relationship is shown as a linear correlation between K_leaf and g_s (Figure 2E).

Another possible trait that may cause a lower reduction in A₄₀₀ under salt treatment in G. biloba is low sensitivity of the photosynthetic biochemistry, that is, lower reductions in V_cmax and J_max (Figure 2C) to salt stress. Salt treatment is reported to reduce V_cmax and/or J_max significantly in G. biloba and in a thermophilic tree, Ziziphus spina-christi, which is connected to a reduction in the leaf photosynthesis rate (Zait et al., 2019; Li et al., 2025). Then, why are V_cmax and J_max of G. biloba less sensitive to salt stress than R. pulchrum? This low sensitivity is most likely attributed to small K_leaf of G. biloba (Table 3), that is, only 50% of R. pulchrum, resulting a slow transpiration stream that may contribute to less transportation of toxic Na from the soil to the leaves in G. biloba. In fact, the leaf Na concentration of G. biloba was only 29% of that of R. pulchrum under salt stress (Table 3). A large interspecific difference in leaf Na accumulation, nearly 30-fold, has been reported for urban trees in Warsaw, Poland (Dmuchowski et al., 2022). In Z. spina-christi, as the Na concentration in the leaves increases, V_cmax decreases accordingly (Zait et al., 2019). Note that for G. biloba, even when Na accumulation is high, salt stress damage remains small. The urban G. biloba in Warsaw, Poland, accumulates a remarkable amount of Na, that is, 1061 mg kg^–1, while no notable damage is observed (Dmuchowski et al., 2019, Dmuchowski et al., 2022).

The low K_leaf of G. biloba should be affected by two factors: 1) low evaporation rate that is attributed to low stomatal density (Table 3) and 2) low hydraulic conductivity in the xylem. The conduit diameter is reported to be important for the response of hydraulic conductivity to soil salinity for angiosperm species, mangroves; species with smaller vessel diameters showed little effect of soil salinity on hydraulic conductivity (Chowdhury et al., 2025). However, the stem tracheid diameter of G. biloba is reported to be 17–31µm (Purusatama and Kim, 2020), which is not larger than the stem vessel diameter of R. pulchrum, 25 µm (Nilsen, 2009). When we observed the cross section of the petiole of G. biloba and R. pulchrum, G. biloba had a smaller proportion of vascular tissue with lower numbers of tracheids compared to R. pulchrum (data not shown). This might be partly related to the low hydraulic conductivity of G. biloba (Horike et al., 2023).

To assess the response of Ginkgo biloba to long-term excess humidity, seedlings of Ginkgo biloba were grown under three sequential humidity conditions. The seedlings were grown for two weeks under 1.5 ± 0.2 (mean ± SD) kPa of VPD (control condition), then for three weeks under 0.5 ± 0.2 kPa of VPD (excess humidity condition), and finally for two weeks under 1.6 ± 0.3 kPa of VPD (recovery condition). Excess-humidity-grown plants are expected to show increases in stomatal conductance and photosynthesis due to acclimation of stomata (Nejad and Van Meeteren, 2005). However, when measured at moderate humidity (VPD of 1.3 kPa or 0.8 kPa), the A₄₀₀ and g_s400 of the excess-humidity-grown G. biloba were no larger and tended to be smaller compared to the control conditions, for both mature and young leaves (Figure 3A). This result shows that the stomata of excess-humidity-grown G. biloba will no longer be more open when measured at a VPD higher than during growth, even when the difference is small (0.3 - 0.8 kPa). These results suggest sensitive closure of stomata to the increase in VPD.

The sensitive closure of stomata in the excess-humidity-grown G. biloba was more apparent in the young leaves when the VPD increased temporarily from 0.8 kPa to 3.0 kPa; a more rapid decrease in A₄₀₀ and g_s400 compared to the control plants was observed (Figure 3B). These results support previous research showing that Vicia faba acclimatized to high humidity conditions had increased stomatal sensitivity to environmental changes, such as changes in the CO₂ concentration (Talbott et al., 2003). In contrast, a perennial herbaceous plant, Tradescantia virginiana, grown under long-term high humidity, showed an increase in g_s and slow closure of the stomata under low humidity (Nejad and Van Meeteren, 2005). This suggests that the stomatal response to prolonged high humidity is highly species-specific. Such a species-specific response of g_s may be partly affected by morphological traits of the stomata. The excess-humidity-grown plants showed a significant decrease in stomatal density and stomatal pore length (Figures 3A, C), both of which would decrease stomatal conductance (Franks and Beerling, 2009).

Significant reductions in V_cmax and J_max in excess-humidity-grown G. biloba were observed for mature leaves (Figure 3A), which suggests that long-term excess humidity impaired the biochemical functions of G. biloba even after leaf maturation. For the G. biloba grown under recovery conditions, the significant decline in gas exchange traits, A₄₀₀, g_s400, V_cmax, and J_max, were even more pronounced compared to the excess-humidity-grown plants (Figure 3A). These results suggest that long-term exposure to excess humidity causes a prolonged negative impact on stomatal function and photosynthetic biochemistry. Even a VPD of 1.6 kPa, which is normally optimal, causes significant stress. On the other hand, the sensitive responses of A₄₀₀ and g_s400 to increased VPD observed in plants grown under excess humidity were not so clear under recovery conditions (Figure 3B).

What are the effects of mesophyll anatomical changes on the photosynthetic response of G. biloba to excess humidity? If the leaf mesophyll tissue of young leaves develops during a treatment of excess-humidity, the surface area of the chloroplasts facing the intercellular spaces (S_c) may increase, which could potentially compensate for the negative impact of biochemical impairment. However, no increase in S_c was obtained for the excess-humidity-grown plants, nor were there any increases in mesophyll thickness, palisade and spongy cell size (Figures 3A, D) or mesophyll cell layers (Supplementary Table S2). In contrast, the mesophyll thickness and the sizes of palisade and spongy cells were increased after the recovery treatment (Figures 3A, D). These suggests that long-term high humidity inhibited the development of mesophyll in G. biloba. The decrease in mesophyll thickness and S_c due to the inhibition of mesophyll development has been observed in tomato cultivars, but these results were obtained under low humidity, not high humidity (Du et al., 2019). On the other hand, according to Torre et al. (2003), the leaves of roses grown under high humidity become thin, and the numbers of palisade cells and stomatal cells decrease. These suggest that the morphological changes in leaves in response to high humidity are again highly species-specific.