The sugarcane cultivars NiF8 (drought-susceptible) and Ni9 (drought-tolerant); Erianthus accessions JIRCAS1 (unknown origin) and JW630 (collected in Shizuoka, Japan); and their intergeneric hybrids F₁ J08-12 (NiF8 × JIRCAS1) and J16-77 (NiF8 × JW630) were included in the study. The hybrids J08–12 and J16–77 were confirmed true intergeneric hybrids using PCR-based simple sequence repeat (SSR) markers (D'Hont et al., 1995) and nuclear DNA content by flow cytometry assays (Pachakkil et al., 2019) (Supplementary Figure S1). Two Erianthus accessions are known to belong to genetically distinct groups (Tsuruta et al., 2012, 2017) and both show robust root system in the field (Terajima et al., 2023). The plants were grown in a temperature- and humidity-controlled glasshouse at the Tropical Agriculture Research Front, Japan International Research Center for Agricultural Sciences (24°22'43" N, 124°11'4" E). The day temperature was maintained at 31°C from 7:00 am to 7:00 pm, while the night temperature was set at 27°C; the relative humidity was maintained at 60% (Supplementary Figure S2). The daily cumulative solar radiation in the greenhouse during the growing season averaged 12.4 ± 5.1 mol m^-2 day^-1.

The pot experiment was performed with a two-factorial design examining sugarcane genotypes and soil moisture conditions (6 genotypes × 2 water regimes). Seedlings of single-bud setts were raised in containers filled with potting mix (Minori, JA Okinawa, Okinawa, Japan) from May 14 (for Erianthus) and June 2 (for sugarcane and intergeneric hybrids) 2021. Due to the slower germination and initial growth of Erianthus, their seedlings were germinated approximately two weeks earlier to synchronize with the growth stage of the other plants. On July 19, 2021, 12 plants of each genotype were transplanted into 1/2000a Wagner pots filled with 10 kg of FW potting mix. Fertilization was performed at transplantation using a solid slow-release fertilizer with a nutrient ratio of N:P:K = 2.4:0.7:1.0 g pot^-1. To reduce evaporation, cobble gravel was spread at a depth of 2 cm on the soil surface, as described in Jackson et al. (2016). Irrigation was initially provided three times daily using an automatic drip system until irrigation control began. On August 18, 2021, drainage was stopped using rubber plugs, and manual irrigation control was implemented. Irrigation was controlled according to Dinh et al. (2017) by reading the volumetric water content (VWC) at 8:00 am using a soil moisture sensor (EC-5, Meter) placed at a soil depth of 13 cm (center of the pot) and estimating the water consumption per pot from the previously obtained soil bulk density. Beginning August 25, 2021, the soil pF value was estimated from the VWC (Supplementary Figure S4) using the moisture characteristic curve of the test soil obtained earlier (Supplementary Figure S3). Two treatments were applied: a wet treatment where the soil was irrigated to a well-watered condition (0.445 m³ m^-3; pF 1.4), and a dry treatment, where irrigation was gradually reduced by approximately 1% until reaching the permanent wilt point (0.131 m³ m^-3; pF 4.2). The pots were randomly placed with four replicates per treatment.

The gas exchange parameters—photosynthetic rate (A), stomatal conductance (g_s), transpiration rate (E), and intercellular CO₂ concentration (C_i)—of the uppermost fully expanded leaves were measured using a portable gas exchange measurement device (LI-6400, LI-COR BioSciences, Lincoln, Nebraska, USA) on August 24 (prior to the start of the irrigation treatment), and on September 6, 16, 23, and October 1 during the treatment period in 2021. A 6-cm² (2 cm × 3 cm) LED chamber (LI-6400B, LI-COR) was used, with two light intensity levels: unsaturated (500 µmol m^-2 s^-1) and saturated (2000 µmol m^-2 s^-1) photosynthetic photon flux density (PPFD). Light curves previously measured for NiF8 and JW630 confirmed no difference between the two species regarding light saturation and unsaturation (Takaragawa and Matsuda, 2023). The flow rate and reference CO₂ concentration were set to 400 µmol s^-1 and 400 µmol mol^-1, respectively. Leaf temperature was maintained at 30.9 ± 1.0°C via a block temperature set at 30°C. Leaf vapor pressure deficit (VPD) was manually controlled at 1.9 ± 0.2 kPa using a desiccant bulb filled with Drierite® (W. A. Hammond Drierite Co., Xenia, OH, USA). iWUE was calculated from the obtained A and g_s, using the equation:

The choice of g_s to calculate gas exchange water use efficiency is based on its role as a transpiration index that accounts for VPD. This approach is easier and equitable, facilitating comparison across studies. In contrast to using transpiration rate or photosynthetic water use efficiency (A/E), g_s provides a more consistent and fairer metric for comparison with other literatures (Jackson et al., 2016; Li et al., 2017; Nakabaru et al., 2020).

A thin layer of nail polish was applied to both sides of the leaf used for gas exchange measurements, and stomatal samples were collected using double-sided tape (Wu and Zhao, 2017; Takaragawa and Wakayama, 2024). Cross-sections of the tested leaves were prepared manually and fixed onto glass slides to measure the interveinal distance—defined as the distance between vascular bundles. Observations were made using an optical microscope system (Eclipse E800, Nikon, Tokyo, Japan) equipped with image analysis software (NIS-elements, Nikon).

At the beginning of the treatment, four plants per genotype were harvested, with all remaining plants harvested 49 days after treatment (on October 13, 2021). The culm length of the main stem, total leaf area, and dry matter weight of each organ were recorded. Leaf area was measured using a leaf area meter (LI-3100, LI-COR). The rate of main-stem elongation during the treatment period was calculated based on the culm length before and after treatment. Underground parts were washed to remove soil and separated into roots and underground stems (stubbles). The underground stem weight was included in the aboveground weight. The leaf area ratio (LAR) during the treatment period was calculated using the leaf area (L₁, L₂) and total dry matter weight (W₁, W₂) measurements taken before and after the treatment, according to the following equation:

The water use efficiency of biomass production, defined as gravimetric WUE (gWUE), was calculated by dividing the increment in dry matter (ΔDW) by the water consumed (ΔWU) during the treatment period, using the following equation (Dinh et al., 2017):

Total nitrogen content of each plant part was analyzed using an NC analyzer (NC22F; Sumika Chemical Analysis Service, Ltd., Osaka, Japan) to calculate the nitrogen uptake (ΔNU) during the treatment period, and the nitrogen use efficiency (NUE) was calculated using the following equation:

Data analysis was conducted using the Bell Curve for Excel statistical analysis software (Social Survey Research Information Co., Ltd., Tokyo, Japan). A two-way factorial analysis of variance (ANOVA) was performed to assess the effects of genotype (six genotypes), water regime (two water regimes), and their interactions on leaf anatomical and dry matter parameters. A four-way factorial ANOVA was also conducted to evaluate the effects of genotype, water regime, PPFD for measurement (two levels), measurement date (five dates), and their interactions with gas exchange parameters. Results of ANOVA were shown with percentage of each factorial variance to total variance. Differences among mean values of the examined parameters for each genotype were determined using Tukey’s test, with statistical significance assumed at P < 0.05 (n = 4). Measured A and g_s values were plotted for each genotype under each PPFD condition, and a correlation analysis was conducted to derive the A vs. g_s slope. Differences in the A vs. g_s slope values between NiF8 and each genotype were assessed using a t-test, with statistical significance assumed at P < 0.05, 0.01, and 0.001.

The soil water conditions during the water treatment are shown in Figure 1. Based on changes in VWC under dry conditions, leaf gas exchange measurements were performed on August 24, September 6, September 16, September 23, and October 1, with mean VWC values of 0.43, 0.32, 0.25, 0.16, and 0.14 m³ m^-3, respectively (Table 1).

The relationship between A and g_s was plotted for all measurements for both dry and wet treatments across each genotype (Figure 2). The correlation between A and g_s was statistically significant under both unsaturated and saturated light conditions, with a steeper slope observed under saturated light than under unsaturated light. Among the genotypes, NiF8 exhibited a consistent tendency for higher g_s (>0.3 mol m^-2 s^-1), regardless of light conditions. The slope under saturated light was significantly higher for Ni9, J08-12, and J16–77 than for NiF8, while JIRCAS1 and JW630 showed higher but not statistically significant trends. Under unsaturated light conditions, the slope was significantly higher for J08–12 and J16–77 compared with NiF8, whereas Ni9 and JW630 exhibited higher but non-significant trends.

The relation of A, g_s, and iWUE to soil moisture was plotted to observe genotype-specific differences relative to NiF8 (Supplementary Figures S5, S6, Figure 3). Under saturated light conditions, NiF8 exhibited a higher A and more pronounced inter-genotype differences (Supplementary Figure S5). NiF8 also showed higher g_s, with higher inter-genotype variation observed under unsaturated light than under saturated light conditions (Supplementary Figure S6). Both A and g_s exhibited minimal inter-genotype differences under extremely dry conditions (VWC< 0.2 m³ m^-3) (Supplementary Figures S5, S6). The differences in iWUE between genotypes were smaller under saturated light than under unsaturated light as well as under conditions of extreme dryness (VWC< 0.2 m³ m^-3) compared to wetter conditions (Figure 3, Table 1). The iWUE of Ni9 consistently remained higher than that of NiF8 under both unsaturated and saturated light conditions, regardless of soil moisture levels (Figures 3A, B). Although the difference in iWUE between Erianthus JIRCAS1 and NiF8 was minimal under saturated light, the iWUE of JIRCAS1 tended to remain higher than that of NiF8 under unsaturated light conditions (Figures 3C, D). The iWUE of Erianthus JW630 was higher than that of NiF8 under both unsaturated and saturated light conditions (Figures 3E, F). The iWUE of the intergeneric hybrids J08–12 and J16–77 was comparable to or higher than that of their Erianthus parents JIRCAS1 and JW630, respectively (Figures 3C–F).

ANOVA results based on the mean of each genotype across all measurement conditions and dates indicated that gas exchange parameters were significantly influenced by soil moisture conditions, light conditions, and genotype (Supplementary Table S1). Among these parameters, A, iWUE, C_i, E, and A/E (photosynthetic water use efficiency) were most strongly affected by PPFD, whereas g_s was primarily influenced by genotype. Significant differences in iWUE between genotypes were observed on each measurement date, with JW630 and J16–77 typically exhibiting significantly higher iWUE than NiF8, except under saturated light conditions during the dry treatment (Table 1).

The stomatal distribution of the test genotypes, including those of Erianthus and intergeneric hybrids, was amphistomatous, with a higher density of stomata on the abaxial surface than on the adaxial surface, consistent with other Poaceae species (Supplementary Figure S7, Table 2). ANOVA results revealed that genotype had a significant effect on all anatomical traits, whereas the effects of water regime and genotype–water interaction were relatively small (Table 2). Regardless of soil moisture conditions, NiF8 exhibited significantly higher stomatal density on the abaxial surface than JW630 and J16-77. The stomatal density on the adaxial surface was generally lower in NiF8 than in other genotypes, regardless of soil moisture conditions, with significant differences observed only in J08–12 under wet conditions and in JIRCAS1 and J16–77 under dry conditions. Overall, JW630 showed a lower stomatal density than the other genotypes. The ratio of adaxial to abaxial stomatal density under wet conditions was highest for JW630 and significantly higher for all other genotypes compared to NiF8. No significant differences in stomatal density ratios were observed between sugarcane cultivars or Erianthus accessions under dry conditions; however, Erianthus and intergeneric hybrids showed higher values compared to the two sugarcane cultivars. The interveinal distance did not differ among the sugarcane cultivars, whereas significant differences were found among Erianthus accessions (Table 2, Supplementary Figure S8). Specifically, Erianthus JW630 and the intergeneric hybrid J16–77 showed significantly longer interveinal distances than NiF8, whereas the intergeneric hybrid J08–12 showed a tendency for longer interveinal distance, although not significantly.

ANOVA results indicated that both genotype and water regime significantly affected all parameters related to dry matter production (Table 3), except for NUE (Supplementary Table S2). Furthermore, the interaction between genotype and water regime was significant for all parameters, except for LAR and NUE. Genotypic differences in gWUE displayed varying trends across treatments. Under wet conditions, the genotypic differences in the gWUE of shoot dry mass were not significant, though the gWUE tended to be higher for the two sugarcane cultivars and the intergeneric hybrid J08–12 than for the two Erianthus accessions and the intergeneric hybrid J16-77. However, under dry conditions, genotypic differences in the gWUE of shoot dry mass were significant, with the two Erianthus accessions showing the lowest values, followed by the two sugarcane cultivars and the two intergeneric hybrids. Comparing the values between two treatments, the gWUE ratios of dry to wet conditions for average shoot dry mass tended to be higher for sugarcane and the intergeneric hybrids than for Erianthus, with similar trends observed for the gWUE of total dry mass. LAR was minimally affected by soil moisture conditions, with clear genotypic differences. The two Erianthus accessions exhibited significantly higher LAR values than the other genotypes. The shoot mass/root mass (S/R) ratios showed similar trends under both dry and wet conditions, being lower for Erianthus and higher for both sugarcane and intergeneric hybrids. The S/R ratio was the lowest for Erianthus JW630, and intermediate to higher for intergeneric hybrids compared to those for the parental genotypes. Under wet conditions, the stem elongation rate was significantly higher for the other genotypes than for Erianthus JW630, whereas under dry conditions, it was significantly higher or tended to be higher for the other genotypes than for the two Erianthus accessions. NUE variation among replicates was large, and genotypic differences were unclear (Supplementary Table S2).

Dry matter partitioning for each organ is shown in Figure 4. In sugarcane, a higher proportion of dry matter was allocated to the stem, with reduced allocation to the leaves due to drought, leading to an increase in dead tissue. In contrast, the Erianthus accessions exhibited higher partitioning to leaves and roots. Although intergeneric hybrids tended to increase root partitioning under drought, their overall dry matter allocation was similar to that of sugarcane, with more dry matter directed to the stems.

The relationship between A and g_s (Figure 2) indicated that Erianthus exhibited a more sensitive stomatal response, with lower g_s and lesser transpiration than the drought-susceptible cultivar NiF8, regardless of light conditions. This trend was particularly evident under unsaturated light conditions. In contrast, compared to that of the drought-tolerant cultivar Ni9, the A vs g_s slope for Erianthus was not high, indicating that stomatal responsiveness in Erianthus was not necessarily higher than that in sugarcane (Figure 2). A key feature of gas exchange in Erianthus, as compared to sugarcane, is the high stomatal responsiveness while maintaining g_s at consistent low levels, which indicates the presence of an underlying anatomical mechanism (Lawson and Blatt, 2014; Bertolino et al., 2019). Typically, longer interveinal distances and fewer stomata result in lower g_s (Kawamitsu et al., 2002; Xu and Zhou, 2008; Pitaloka et al., 2022). When factors affecting stomatal responses, such as water status (soil moisture, VPD, etc.) and solar radiation, are variable, the stomatal reactivity—the ability to adjust stomatal opening and closing in response to these factors—plays a critical role in maintaining high iWUE (Kawamitsu et al., 1993; Tominaga et al., 2014; Matthews et al., 2017; Dinh et al., 2019; Lawson and Vialet-Chabrand, 2019; Ozeki et al., 2022). A better stomatal response has been reported in leaves that are more amphistomatous, with a higher distribution of stomata on the adaxial surface relative to the abaxial surface (Haworth et al., 2018; Drake et al., 2019; Xiong and Flexas, 2020). Amphistomatous leaves contribute to maintaining optimal leaf water status in response to transpiration demand. When stomata are open, the temperature gradient between the atmosphere, stomatal cavity, and leaf chloroplast is reduced (Drake et al., 2019), suggesting that stomatal responses to VPD—the driving force for transpiration—can be effectively regulated (Kawamitsu et al., 1993, 2002). Furthermore, in C₄ grasses, more stomata on the adaxial surface can increase the surface area of mesophyll cells in contact with intracellular air space, enhancing iWUE and mesophyll conductance (Pathare et al., 2020). In the present study, the Erianthus species, particularly JW630, exhibited fewer stomata on the abaxial surface (resulting in a longer interveinal distance) and a more amphistomatous stomatal distribution (Table 2), which may explain its heightened stomatal responsiveness.

Furthermore, Erianthus JIRCAS1 exhibited a trend toward higher iWUE than the susceptible cultivar NiF8 under both wet and dry conditions, although its iWUE was not consistently higher than that of the drought-tolerant cultivar Ni9 (Figure 3, Table 1). In contrast, Erianthus JW630 consistently showed significantly higher iWUE than both NiF8 and Ni9. The high iWUE of Erianthus JW630 was likely attributed to leaf anatomy, including low stomatal density, (Table 2), leading to low g_s, and may have a different physiological mechanism compared to that of drought-tolerant Ni9. The ability of Erianthus to maintain high A despite a low g_s (that is, a high iWUE) may be linked to ultrastructural features such as mesophyll cell wall thickness and surface area in contact with the stomatal cavity, both of which are involved in bundle sheath leakiness (von Caemmerer and Furbank, 2003). Further investigation of gas exchange characteristics, such as A-C_i curves, and anatomical features of this species will provide deeper insights into these mechanisms.

Erianthus exhibits genetically distinct lineages (Tsuruta et al., 2012, 2017, 2022) which influence variations in its morphological, ecological (Tagane et al., 2012), and agronomic (Terajima et al., 2022) traits. In the current study, variation in iWUE was observed between two Erianthus accessions (Table 1) which are classified into different genetic groups (Tsuruta et al., 2012, 2017). The accession JW630 was collected from Shizuoka Prefecture, a temperate region in Japan, while the origin of JIRCAS1 remains unknown. Previous studies have primarily focused on tropical accessions, such as the IJ series, which also exhibit high iWUE and related variations (Jackson et al., 2016; Li et al., 2017). These findings highlight the need for further investigation into the variation in leaf characteristics across different genotypic groups in Erianthus. Additionally, selecting Erianthus genotypes for improving drought tolerance in sugarcane will require considering both root system and aboveground traits.

Jackson et al. (2016) reported a strong correlation between iWUE and gWUE in sugarcane germplasm, with several Erianthus accessions showing higher values for both parameters compared to sugarcane. However, in the current study, high iWUE did not necessarily lead to high gWUE in Erianthus. This discrepancy may be attributed to differences in growing conditions: Jackson et al. (2016) conducted their trial in larger pots under outdoor conditions, while our study, was performed in smaller pots in a glasshouse. These differences may have limited branching in sugarcane varieties and caused greater root restriction in Erianthus. It is recommended that gWUE evaluation and screening in pot trials should take into account pot size and that evaluation of leaf traits at very early growth stages, rather than gWUE, would be more appropriate to validly evaluate genotypic differences under pot experiments. Despite this discrepancy with previous studies, low iWUE in sugarcane may result in low gWUE under field conditions, particularly under field canopy conditions, due to the larger leaf area and high LAR (or high LAI) of tillers. Additionally, because many leaves under a shaded canopy perform photosynthesis under low-light conditions (Almeida et al., 2022), where iWUE, which exhibits higher genotypic variation under low-light conditions, may have a more pronounced impact on gWUE. Erianthus, recognized for its drought-tolerance, may achieve high gWUE even at high LAR (caused by presence of many tillers), owing to its robust root system in the field (Terajima et al., 2023), in addition to its high iWUE (Figure 3, Table 1). The intergeneric F₁ hybrids, which exhibited high gWUE under pot conditions—unlike the parental Erianthus accessions—showed high iWUE (Table 1) and demonstrated sugarcane-like dry matter partitioning characteristics (low LAR, high S/R ratio, and higher stem partitioning) (Table 3, Figure 4). Consequently, these hybrids may potentially maintain high gWUE even under field canopy conditions with high LAR. Although the intergeneric hybrids exhibited limited dry matter partitioning to roots in this study, which focused on relatively early growth under pot conditions, field studies have shown that the hybrid J08–12 forms roots with intermediate potential between parental species, exhibiting higher root mass and depth than sugarcane (Takaragawa et al., 2022; Terajima et al., 2023). These findings in the present study highlighted the "best of both worlds" scenario demonstrated by the hybrids, where they inherited high iWUE from Erianthus with favorable biomass partitioning characteristics from sugarcane. This fact could represent the ideal outcome for breeding drought-tolerant varieties via intergeneric hybridization with Erianthus. Further field trials will assess the relationships among iWUE, canopy coverage, root system formation, and gWUE using a hybrid population derived from several sets of parental genotypes.

The PPFD for gas exchange measurements had the greatest influence on iWUE (Supplementary Table S1). Genotypic differences in iWUE were particularly pronounced under unsaturated light (500 µmol m^-2 s^-1) conditions than under saturated light (2000 µmol m^-2 s^-1) conditions (Figure 3, Table 1). Additionally, the slope of the A vs g_s curve was smaller, and the stomatal response was notably lower under unsaturated light conditions than under saturated light conditions (Figure 2). Therefore, iWUE screening under low-light conditions may prove effective and provide practical implications for developing high-throughput phenotyping protocols for drought tolerance screening. In contrast, such differences in stomatal responses due to varying light conditions suggest that obtaining stable results when measuring the response at multiple sites under field conditions may be challenging, especially in regions such as Okinawa (which comprise small islands and represent our study site), where the weather frequently shifts between cloudy and sunny within short time frames. Genotypic differences in iWUE vary depending on the measurement date; therefore, measuring within a moderate g_s range (0.2–0.3 mol m^-2 s^-1, Jackson et al., 2016; 0.1–0.4 mol m^-2 s^-1, Natarajan et al., 2021) or averaging multiple measurements, is recommended (Jackson et al., 2016; Li et al., 2017). Considering this climate instability, investigating genotypic differences in response to fluctuating light conditions (Eyland et al., 2021; Tanaka et al., 2019) and exploring non-destructive methods for measuring daily variations in gas exchange, such as sap flow for transpiration (Contreras and Ozawa, 2005), are essential.

Gas exchange measurements are strongly influenced by environmental variations during data collection, which can compromise the stability and efficiency of the measurements. In recent years, the throughput of photosynthesis measurements has been enhanced by reducing measurement time through the use of closed-type equipment (Honda et al., 2021; Tanaka et al., 2021; Takaragawa and Matsuda, 2023). There have been no previous reports on high-throughput estimating and screening iWUE using UAVs and hyperspectral images, while component parameters for iWUE can be estimated by aerial image analysis: transpiration indices from leaf or canopy temperatures obtained from thermal images (Basnayake et al., 2016; Iseki and Olaleye, 2020; Natarajan et al., 2019) and photosynthetic activity using hyperspectral images (Kohzuma et al., 2021). However, despite recent advancements and attempts (Takaragawa et al., 2025), improvements in the measurement throughput of iWUE, which requires simultaneous measurement of photosynthesis and g_s, remain incomplete.

Gas exchange is governed by complex biochemical processes influenced by metabolites, enzymes, and morphology (Sage et al., 2013). Among these, leaf morphological and anatomical traits, particularly stomatal characteristics, play a critical role in supporting gas exchange and mechanical function (Franks and Farquhar, 2007; Haworth et al., 2021). Although anatomical traits, such as stomatal density, are not sufficiently robust or universal enough to be used for species classification (Davis, 1987), they exhibit a smaller environmental variation compared to gas exchange characteristics and can show stable genotypic variation (Moreno-Sotomayor et al., 2002). The current study also demonstrated that environmental variation in leaf anatomical traits was relatively small as indicated by ANOVA results (Table 2, Supplementary Table S1). The leaf anatomical characteristics of the intergeneric F₁ hybrids were intermediate between the parental genotypes, with intergeneric hybridization with Erianthus resulting in a progeny having longer interveinal distances, fewer stomata on the abaxial surface, and a higher stomatal distribution ratio (Table 2). These findings suggest that the improvement in iWUE through intergeneric hybridization was facilitated by changes in leaf anatomy. When examining hybrid populations, the throughput of morphological and anatomical observations may need to be enhanced through rapid image acquisition or other methods (Strock et al., 2022).

Although intergeneric hybridization offers potential for improving leaf traits of sugarcane, F₁ hybrids typically exhibit lower sugar content with higher fiber content than sugarcane parents (Pachakkil et al., 2019), which discourages their direct utilization in breeding programs for sugar industry. Therefore, backcrossing using sugarcane variety must be performed to improve sugar content of hybrids, requiring a further investigation of leaf traits in the backcross populations.