Higher adaxial stomatal density is associated with lower grain yield in spring wheat

Stomata are microscopic pores, each formed by two guard cells, on the leaf surface that enable gas exchange between the leaf and the atmosphere. Stomatal traits, for example, stomatal density (SD, number of stomata per unit leaf area), stomatal distribution between leaf surfaces (stomatal ratio (SR), ratio of adaxial SD to abaxial SD), guard cell length (GCL), stomatal aperture width, stomatal conductance (gs), and maximum anatomical stomatal conductance (gsmax), vary considerably among different species, genotypes, and environmental conditions, affecting leaf and plant gas exchange (Salisbury, 1928; Maruyama & Tajima, 1990; Weyers & Lawson, 1997; James & Bell, 2000; Hetherington & Woodward, 2003; Muir, 2015; Stevens et al., 2021; Nunes et al., 2022; Caine et al., 2023; Liu et al., 2023; Wall et al., 2023). High values of gs denote a high rate of water vapour diffusion from leaves and are associated with efficient photosynthesis, active water movement through the plant, and canopy cooling, leading to not only higher yields but also higher water loss as a drawback (Wong et al., 1979; Roche, 2015). SD affects photosynthesis, growth, stress tolerance, and production in both dicots and cereal crops (Hepworth et al., 2015; Hughes et al., 2017; Caine et al., 2019; Dunn et al., 2019; Lawson & Matthews, 2020; Jalakas et al., 2024; Tulva et al., 2024). Depending on plant species and environment, stomata can be present on only one leaf surface or on both. Amphistomatous species with stomata on both leaf surfaces usually have a higher SD on the abaxial leaf surface (Salisbury, 1928; Pemadasa, 1979; Mott et al., 1982; Muir, 2015; Caseys et al., 2024), whereas crops of the Gramineae family (like wheat) may have equal or even higher SD on the adaxial surface (Pathare et al., 2020; McAusland et al., 2021; Wall et al., 2022, 2023). The presence of stomata on both leaf surfaces results in enhanced photosynthetic rates via boosting stomatal and mesophyll conductances for CO2 diffusion, with potential disadvantages, such as pathogen sensitivity or higher dehydration risk (Parkhurst, 1978; Drake et al., 2019; Muir, 2020; Pathare et al., 2020; Xiong & Flexas, 2020). Higher dehydration risk associated with adaxial stomata may be due to the longer distance to the closest veins and larger leaf-to-air vapour pressure differences representing higher evaporative demand (Drake et al., 2019). Here, we investigated stomatal traits of adaxial and abaxial leaf surfaces (SD, SR, GCL, gs, and gsmax) in seven spring wheat genotypes under field conditions to assess their genetic and environmental variation, examine their relationships with grain yield, and provide insights for future breeding strategies. Agronomic traits of the studied genotypes have been reported elsewhere (Ingver et al., 2024) along with the effects of two nitrogen (N) regimes used: N150, the typical fertilization rate for spring wheat in the Baltics, and N75, representing a less intensive, environmentally friendly management approach (Jansone et al., 2024). In 2022 and 2023, fully expanded flag leaves of spring wheat were measured with a porometer for abaxial and adaxial gs values and with an N-tester for leaf Chl contents. The leaves were then collected for further stomatal anatomical analyses. The 2022 vegetative period experienced near-average weather conditions (Supporting Information Fig. S1), with June being slightly warmer, sunnier, and drier than the long-term regional average. By contrast, 2023 saw a pronounced dry and sunny period in May–June (Fig. S1), followed by exceptionally high rainfall in July. Above-average temperatures and soil water deficiency experienced during the period of rapid vegetative growth and development of generative organs (the second half of May and June) can significantly reduce the yield of all spring cereals. Therefore, weather differences in 2022–2023 likely contributed to the variation in grain yields, which ranged from 5200 to 6900 kg ha−1 in 2022 but were lower in 2023 (2800–4600 kg ha−1). Genotype and year had significant main effects on SD and GCL values of adaxial and abaxial leaf surfaces, but only genotype affected adaxial to abaxial SR (Table 1), suggesting that SR is largely genetically determined, and environmentally induced changes in SD were proportional in different leaf surfaces. N fertilization had no significant effect on stomatal traits except abaxial GCL, but it increased flag leaf area and had a small positive effect on grain yield. In 2023, most of the genotypes showed significantly higher adaxial and abaxial SDs than in 2022 (Fig. 1a,b). Higher SDs in 2023 were associated with significantly smaller leaf areas due to early season drought (Fig. 1c). Leaf area has been found to correlate with SD, either negatively or positively (Salisbury, 1928; Brodribb et al., 2013; Peel et al., 2017; Wall et al., 2023). Stomata are formed early in leaf development before full leaf expansion is reached (Gay & Hurd, 1975; Hepworth et al., 2018; Nunes et al., 2020). Grass stomatal development follows a strict pattern and adjusting SD to the environment may happen via changes in stomatal spacing within stomatal files or between them, or by changes in the number of stomatal files (Stebbins & Shah, 1960; McKown & Bergmann, 2020). Different abiotic stresses suppress leaf expansion growth, which relies on cell turgor and is strongly affected by leaf water status, leading to smaller leaves under stress conditions (Munns et al., 2000; Munns & Tester, 2008; Driesen et al., 2020). In 2023, drought started to develop already in May and intensified in June; hence, both stomatal development and expansion of flag leaves were affected by it. On average, flag leaf area (calculated as average of all genotypes and N-treatments) was c. 60% smaller in 2023 than in 2022 (Fig. 1c), likely due to drought-induced suppression of cell expansion. Variation in SD values (Fig. 1a,b) indicates that SD provides a valuable source of genetic diversity that has not yet been harnessed in breeding. As previously found in wheat (Wall et al., 2023), more stomata were detected on the adaxial than the abaxial leaf surface as the values of SR (ratio of adaxial SD to abaxial SD) were always above 1 (Fig. 1d). In 2022, guard cells were longer than in 2023, and adaxial guard cells were always longer than corresponding abaxial guard cells (Fig. S2a,b). The previously detected significant negative relationship between SD and GCL (Franks & Beerling, 2009; Haworth et al., 2023; Wall et al., 2023) was found only when years were pooled. Flag leaf total stomatal conductance of all genotypes was significantly lower in 2023 than in 2022 due to drought (Fig. 1e). In 2022, gs was similarly high in all genotypes. Higher values for adaxial than abaxial gs were measured in both years (Fig. S3). A significant correlation between abaxial and adaxial gs values, pointing at a coordinated stomatal regulation between leaf surfaces, was detected only in 2023 (Fig. S3). Values of anatomical gsmax were higher for adaxial than for abaxial leaf surface (P = 0.00) and higher in 2023 than in 2022 (Fig. S4; Table 1). N-tester values of flag leaves were significantly affected by genotype, year, and N fertilization (Table 1). While the main effect of N fertilization on N-tester values was significant (Table 1, Fig. S5), there were no significant differences between differently fertilized pairs within each year. In the pooled dataset (2022–2023), N-tester values of flag leaves that developed and expanded during the drought year of 2023 were higher (Fig. S5). These leaves had significantly smaller areas (Fig. 1c) and likely other drought-related characteristics, for example, higher leaf mass per area, which means thicker leaves and less transpirational area (Poorter et al., 2009). Thicker leaves methodologically give higher N-tester values (see also Fig. S6 for Arabidopsis) that in this case do not indicate vigorous plant N nutrition and cannot be used for predicting high yields. Rather the opposite, N-tester values were negatively associated with grain yield within the pooled dataset of 2022–2023 but were not related to yield in individual years, irrespective of whether N levels were pooled or not (Table S1). This, together with the fact that N fertilization only slightly affected grain yield, shows that in our experiment, N was not a strongly limiting factor. Next, we analysed the relationships between stomatal traits and wheat grain yield by fitting linear regression models (Table S1). Interestingly, a strong significant negative relationship between adaxial SD and grain yield was detected for N75 plots in 2022 and 2023 and for N150 plots in 2023 (Fig. 1f). At the same time, abaxial SD was not significantly related to yield (Table S1). Neither total gs nor gs of leaf surfaces separately was related to grain yield in individual years (Table S1). This suggests that the significant negative relationship between adaxial SD and grain yield was not due to higher water loss of genotypes with higher adaxial SD, as gs combines SD and stomatal aperture width to determine the diffusion of water vapour out of the leaf. This point is supported by a similarly strong decrease in gs values of all wheat genotypes in response to water stress of 2023 compared with 2022, showing that all genotypes, irrespective of their differences in SDs, were able to restrict water loss through decreasing gs under drought (Fig. 1e). Alternatively, single point measurements of gs taken around midday did not reveal the differences in diurnal water loss between genotypes. For example, genotypic variation in nocturnal stomatal conductance has been detected for wheat (McAusland et al., 2021), and higher adaxial SD could result in larger night-time water loss especially in Northern countries with dusky rather than dark summer nights. Since N fertilization had no significant effect on the values of SD and gsmax (Table 1), we pooled data from different N levels and found a significant negative relationship between adaxial SD and grain yield and adaxial gsmax and grain yield in both years (Fig. S7), whereas the respective abaxial traits were not correlated with yield (Table S1; Fig. S7). Thus, cumulative water loss through more numerous and potentially more open adaxial stomata may lead to negative effects on grain yield. Adaxial SD was the highest in the genotype DS-720-3-DH in both years, and since SD is an important determinant of gsmax, adaxial gsmax of DS-720-3-DH was also among the highest (Figs 1a, S4). The genotypes with the lowest adaxial SDs were Hiie and DS-638-5-DH (Fig. 1a). In 2021–2022, the phenotypic diversity of 300 spring wheat genotypes from the Nordic-Baltic region was studied, and DS-720-3-DH ranked among superior genotypes due to its high grain yield, grain protein content, test weight, and thousand kernel weight (Ingver et al., 2024; Jansone et al., 2024). Our study indicates that because of its high adaxial SD, DS-720-3-DH is not the best parental material in breeding for stressful conditions, as also supported by its very low yield in the drought year of 2023. All stomatal traits were significantly related to grain yield within the pooled dataset, except SR (which was negatively related to grain yield in N75 plots in 2023; Table S1). However, for pooled data, these significant relationships are rather due to environmental differences between the years and not due to differences between genotypes. In conclusion, the considerable variation in adaxial SD between wheat genotypes in individual years and between different years was highly indicative of yield: the fewer stomata on the adaxial surface, the larger the grain yield. This result was found for 2023 characterized by serious early season drought and for N75 plots in 2022. At this point, we can only hypothesize about the reasons behind the negative relationship between adaxial SD and grain yield: as the month of June in both years was warmer, sunnier, and drier than usual, larger dehydration associated with higher gas-exchange potential via more adaxial stomata seems the most probable cause (Fig. S7). Different pathogen sensitivity due to variation in adaxial SD is less probable in these conditions, but cannot be excluded. For N150 plots in 2022, there was no significant relationship between adaxial SD and grain yield due to larger biological variation. Correlations between SD, net assimilation rate, and yield have been detected in previous studies (Gaskell & Pearce, 1983; Kumar et al., 2021; McAusland et al., 2021). In rice, drought-resistant lines showed consistently lower adaxial stomatal densities compared with the drought-susceptible ones, and adaxial SD negatively correlated with yield under high vapour pressure deficit (Kumar et al., 2021). In the C4 plant maize, hybrids with higher carbon exchange rates had larger adaxial SD and lower stomatal resistance (Gaskell & Pearce, 1983). In wheat, higher abaxial, but not adaxial SD, led to greater numbers of heavier seed (McAusland et al., 2021), whereas in other studies, neither abaxial SD of wheat flag leaves (Liao et al., 2005) nor adaxial SD of wheat flag leaves (Shahinnia et al., 2016) correlated with grain yield. Stagnation in European wheat yields since the early 1990s has led to discussions about whether and how much it is possible to boost crop yields with photosynthetic improvements (Moore & Lobell, 2015; Faralli & Lawson, 2020; Le Gouis et al., 2020; Matthews & Burgess, 2024) and breeding plants with higher production in well-watered vs drought-prone environments by modifying SD (Hepworth et al., 2018). Stomatal distribution has also been recognized as a functional trait for improved photosynthesis (Lawson & Blatt, 2014; Xiong & Flexas, 2020). Our study simultaneously measured and distinguished between leaf surfaces to highlight their distinct effects on yield. Results showed that higher-than-optimal adaxial SD values could influence yield outcomes of wheat, particularly under water stress conditions, which are likely to occur more often in the future. Its reasons and whether the same is true for crops with lower SR values await further studies. In rainier years, on the contrary, more adaxial stomata may be beneficial for net assimilation rate and yield. Seven spring wheat (Triticum aestivum L.) genotypes representing perspective breeding lines (876, 013-074, DS-638-5-DH, and DS-720-3-DH) and commercial cultivars (Hiie, Robijs, and Runar) developed in the spring wheat breeding programmes of the Baltic States and Norway were studied. The field experiments were carried out at the Centre of Estonian Rural Research and Knowledge (METK, Estonia), Estonia (58°45′N, 26°24′E), in 2022–2023. The wheat genotypes were sown (plot size of 10 m2 and row spacings of 12.5 cm) under two N fertilization levels applied before sowing: 75 kg ha−1 N (referred to as N75) and 150 kg ha−1 (referred to as N150). Two plots per genotype and N level were randomly sown on 03 May 2022 and 25 April 2023. At maturity, the plots were harvested mechanically (dates 18–25 August 2022 and 14–21 August 2023), and the grain yield (kg ha−1) was expressed on a 14% grain moisture basis. The timing of flag leaf measurements was chosen to study fully expanded leaves at their maximum physiological activity (30 June 2022 and 29 June 2023). In both years, the genotypes were around the anthesis stage at the time of assessment, with flag leaves fully unfolded. Adaxial and abaxial leaf surfaces of six flag leaves per plot were measured with a LI-600 (LI-COR) hand-held porometer to get stomatal conductance, gs. Measurements of gs were performed between 10:00 h and 13:00 h, starting with the first replication plots and continuing with the second replication plots. Routinely, the abaxial surface of the middle section of the chosen flag leaf was measured first, followed by the adaxial surface of that leaf, avoiding the spot of the abaxial measurement. The weather did not change significantly during this relatively short measurement period. However, porometer gs data represent in situ values of leaves in their specific growth conditions, resulting in a large variability of gs depending on leaf angles, their positions relative to the sun, shading from other leaves, etc. In further data analysis, total gs was calculated as the sum of adaxial and abaxial stomatal conductances. In addition to porometer measurements, N-tester values of measured flag leaves were taken using a demo mode of the N-tester (Yara, Yara UK Ltd, York, UK). The N-tester is a hand-held spectrometer used in precision agriculture to indicate potential N deficiency of crops; it measures light absorption of leaves and leaf Chl content, which, in turn, is closely correlated with N concentration of leaves (https://www.yara.my/contentassets/6d5ba39b1a364a33be1e4e6b6b2a2be1/n-tester-instruction-manual.pdf/; Lopez-Bellido et al., 2004). The measured leaves were then harvested and taken to the laboratory for leaf area and SD measurements. Leaves were laid flat on a table and photographed, and their projected areas were determined with the ImageJ software (Schneider et al., 2012). For SD measurements, the central area of the flag leaf was sampled, avoiding the midrib region, and impressions of both the adaxial and abaxial leaf surfaces were made using dental silicone. Specifically, Speedex light body (Coltene/Whaledent AG, Alstätte, Switzerland) was used for the adaxial surface and oranwash L (Zhermack, www.zhermack.com) for the abaxial surface. Materials of different colour were used for adaxial and abaxial surface images to help in keeping track of them and for emotional value added by colours in this routine work. Secondary imprints made with nail varnish were collected from the silicone and transferred onto microscope slides with the help of transparent tape. An area of 0.26 mm2 from each imprint was photographed under a microscope (Kern OBF 133; Kern & Sohn GmbH; www.kern-sohn.com) at 200× magnification. SD (stomata per mm2), GCL, stomatal pore length (PL), and apical guard cell width (WA) were determined from these images using the ImageJ software (Schneider et al., 2012). Stomatal ratio (SR) was calculated as the ratio of adaxial SD to abaxial SD. Analysis of variance (GLM procedure) was used to evaluate the main effects of genotype, N fertilization, and year on leaf area, stomatal traits, and grain yield (Statistica, v.7.0; StatSoft Inc., Tulsa, OK, USA). Comparisons between measured variables were performed using three-way analysis of variance (ANOVA) followed by Tukey post hoc test to compare the individual means with the R scripting language, v.4.4.2 (R Team, 2021). Simple linear regression analysis was performed to detect significant relationships between studied traits and grain yield. All effects were considered significant at P < 0.05. This study was funded by the Estonian Research Council, PSG404 and PRG1620, and project ‘NOBALwheat – breeding toolbox for sustainable food system of the NOrdic BALtic region’, Research Council of Lithuania (LMTLT), contract no.: S-BMT21–3 (LT08-2-LMT-K-01-032). We are grateful to Mikk Välbe, Isabella Siil, Daana Morozova, Marilin Poogen, Katarína Geletková, and Dmitrii Kliaus for their help with sample collection, stomatal imaging, and counting. The research was conducted using the TAIM Plant Biology Infrastructure funded by the Estonian Research Council (TT5). None declared. HH initiated the studies concerning stomatal imaging of different leaf surfaces. HH and EM conceived and the and were in the of plant HH and EM collected and analysed the and EM the with from all The data that the of this study are in the Information of this S1). data the of the Fig. data for Fig. cell of adaxial and abaxial leaf surfaces. Fig. between adaxial and abaxial stomatal conductances during 2022–2023. Fig. stomatal conductances of measured flag leaves calculated from anatomical Fig. N-tester values of measured flag leaves, = Fig. S6 between leaf mass per area and N-tester in Fig. between SD and grain yield and gsmax and grain yield. Table linear relationships between and different leaf and stomatal is not for the or of Information by the than be to the The is not for the or of by the than be to the corresponding for the The with to in and in