Seeds of L. bicolor were collected in October 2019 from the inland saline soil of the Yellow River Delta (N37°20′; E118°36′) in Dongying, Shandong, China. Uniform seeds were selected for sowing after storage at 4°C for 6 months.

The seeds of L. bicolor were initially surface-sterilized in 75% ethanol for 5 min, followed by soaking in 6% sodium hypochlorite solution with shaking for 15–20 min. Seeds were then washed using sterile water three to five times. The seeds were sown on growth medium containing various phytohormones and grown at 26/22°C (day/night) with a photoperiod of 16/8 h (day/night) and a light intensity of 600 μmol/m²/s.

All seeds were sown on Murashige and Skoog (MS) basal growth medium adjusted to pH 5.8–6.0 (Murashige and Skoog, 1962). The stock solutions of phytohormones were 50 mg/mL SA, 0.5 mg/mL BL, 10 μM MeJA, 0.1 mg/mL GA₃, and 0.1 mg/mL ABA. MeJA was dissolved in absolute ethanol, while all other phytohormones were first dissolved in a small volume of absolute ethanol until full dissolution before the volume was adjusted with water. All stock solutions were filtered using a 0.2-μm filter.

Salicylic acid was added to MS medium to final concentrations of 25, 50, 75, 100, or 125 mg/L (Rajjou et al., 2006; Rivas-San Vicente and Plasencia, 2011). BL was added to MS medium at concentrations of 1.25, 2.5, 5, or 7.5 mg/L (Vogler et al., 2014). MeJA was added to a final concentration of 0.01 or 0.1 μM (Bazabakana et al., 1999; Chen et al., 2012). GA₃ was added to a final concentration of 0.02, 0.04, 0.06, 0.08, or 0.1 mg/L (Li G. et al., 2014). ABA was added to a final concentration of 0.02, 0.04, 0.06, or 0.08 mg/L (Pan et al., 2020). Three replicates were sown for each concentration, and each replicate consisted of 20 seeds.

After growth for 15 days, the first true leaves were collected and fixed in a mixture of ethanol and acetic acid (3:1, v/v), then cleared in 70% ethanol, before being mounted in Hoyer’s solution (Meinke, and David, 1994). The leaves were observed by differential interference contrast (DIC) microscopy (ECLIPSE 80i, Nikon, Tokyo, Japan) with 330–380 nm ultraviolet (UV) excitation Images from five different fields were taken for each leaf, with 15 leaves per treatment. The average numbers of salt glands, pavement cells and stomata were calculated across the five fields using images acquired with a CCD camera (Nikon, Japan). Leaf area and pavement cell area were measured using ImageJ.

Salt gland density was calculated using the number of salt glands in a field, divided by the field area. The total number of salt glands was calculated as leaf area × salt gland density. Pavement cell density was calculated as the number of pavement cells in a field, divided by the field area, while the total number of epidermal cells was defined as leaf area × epidermal cell density. Stomatal density was calculated as the number of stomata in a field, divided by the field area, with the total number of stomata being leaf area × stomatal density.

In parallel, leaves were observed on an upright microscope (DM6B, Leica, Germany) to visualize pavement cells with a 19-mm field of view sCMOS camera with the LAS X Navigator Software.

SPSS (IBM, SPSS Statistics 25) was used for statistical analysis. Duncan’s multiple comparisons were performed to determine significant differences between samples, with a significance cutoff of P < 0.05. Analysis of variance (ANOVA) was used to determine statistical significance. All data collected from the five phytohormone treatments were normalized in SPSS to remove the influence of dimensions. These normalized values were saved as variables and used for correlation analysis according to the phytohormone. Statistical significance was determined by two-tailed Student’s t-test for P < 0.05 and P < 0.01.

We first germinated L. bicolor seeds on MS medium containing a range of SA concentrations (0–125 mg/L) to investigate the development of salt glands, stomata, and pavement cells from the fully expanded first true leaf (Figure 1A). We measured the area of the epidermis occupied by pavement cells with an upright microscope (Figure 1B) and scored the number of salt glands and stomata using autofluorescence resulting from UV excitation between 330 and 380 nm (Figure 1C). Given that salt gland density varied under different treatments (Supplementary Figure 1) due to the different leaf area, the total salt gland number on the first true leaf are further used to be compared with stomata and pavement cells.

The area of the first true leaf increased slightly when treated with 25 mg/L SA relative to control leaves but then decreased with higher SA concentrations (>25 mg/L) (Figure 1D). We observed the same trend for the leaf area covered by pavement cells, which was highest at 25 mg/L SA. Likewise, the number of salt glands, stomata, and pavement cells followed the same pattern, with the greatest cell numbers obtained with 25 mg/L SA, and fewer cells at higher SA concentrations (>25 mg/L) (Figure 1D).

We next germinated L. bicolor seeds and followed seedling growth on MS medium containing 0–7.5 mg/L BL (Figure 2A). As with SA, we observed pavement cells (Figure 2B), salt glands, and stomata (Figure 2C). Again, leaf area and pavement cell area reached their highest values at the lowest phytohormone concentration (here 1.25 mg/L BL) (Figure 2D), both exhibiting a gradual decrease at higher concentrations (>1.25 mg/L) of BL. The total numbers of salt glands, stomata, and pavement cells followed similar trends with BL concentration (Figure 2D), peaking at 1.25 mg/L BL, before dropping with higher BL concentrations (>1.25 mg/L).

Minute amounts of MeJA can substantially affect leaf development, as evidenced by the inhibition of seedling growth we observed with 0.1 μM MeJA (Figure 3A). Accordingly, we determined the developmental status of salt glands, stomata, and pavement cells at low MeJA concentrations of 0.01 and 0.1 μM (Figures 3B,C). Leaf area was markedly smaller upon treatment with 0.1 μM MeJA but not with 0.01 μM MeJA, while pavement cell area already diminished at MeJA concentrations as low as 0.01 μM compared to control seedlings (Figure 3D). Similarly, increasing MeJA concentrations were accompanied by a reduction in the total numbers of salt glands, stomata, and pavement cells relative to control conditions.

We exposed seeds to five different GA₃ concentrations ranging from 0.02 to 0.1 mg/L (along with a control not treated with GA₃) and characterized leaf development (Figure 4A), pavement cell expansion (Figure 4B), and the numbers of salt glands and stomata (Figure 4C). Even the lowest GA₃ concentration of 0.02 mg/L limited leaf expansion; raising GA₃ concentrations to 0.1 mg/L did not substantially further reduce leaf area. Pavement cell area showed a similar trend (Figure 4D). Likewise, the numbers of salt glands, stomata, and pavement cells dropped with 0.02 mg/L GA₃, but then remained constant with higher GA₃ concentrations (>0.02 mg/L) (Figure 4D).

Finally, we tested the effects of ABA treatment on leaf growth (Figure 5A) and epidermis development (Figures 5B,C). Low and intermediate ABA concentrations (0.02–0.06 mg/L) reduced leaf and pavement cell areas to roughly the same extent (Figure 5D). However, the highest ABA concentration applied (0.08 mg/L) further reduced both leaf and pavement cell areas. The total numbers of salt glands, stomata, and pavement cells displayed the same trends as leaf area (Figure 5D).

We then explored the correlations between the total numbers of salt glands, stomata, and pavement cells; leaf area; and pavement cell area for each phytohormone using SPSS. For SA treatments, the total salt gland number showed a significant (P = 0.01) and positive correlation with leaf area, pavement cell area, total stomatal number, and pavement cell number (Supplementary Table 1). We also obtained strong, positive correlations between stomatal number and leaf area, as well as stomatal number and pavement cell number, indicating that the development of salt glands, stomata, and pavement cells might be coupled or coordinated during leaf expansion.

We then applied the same analysis to the other four phytohormones (Supplementary Tables 2–5). We observed similar results, supporting the notion that the number of salt glands is highly correlated with all other parameters tested. We therefore conclude that larger leaves will present more salt glands, stomata, and pavement cells.

Finally, we integrated the data from all five phytohormones after normalization to remove any scaling effects and repeated the correlation analysis between the total numbers of salt glands, stomata, and pavement cells; leaf area; and pavement cell area (Table 1). Again, we obtained highly significant and positive correlations between salt glands and other parameters.

As different phytohormones have different effects on leaf development, we combined all normalized data to determine the overall correlation between leaf development parameters. We detected strong correlations between salt gland number and pavement cell area, and between salt gland number and leaf area when considering each phytohormone separately (Supplementary Tables 1–5) or in combination (Table 1). In previous studies of L. bicolor, salt gland number always showed a positive correlation with leaf area (Yuan et al., 2018); for example, Ca²⁺ treatment enhanced salt gland development and leaf expansion (Ding et al., 2010). Here, we observed strong and positive correlations between salt gland number and leaf area, indicating that salt gland development is coupled with leaf development.

Salt glands are not the only epidermal structure of L. bicolor. Stomata also responded to the different phytohormone treatments. The development of salt glands and stomata is thought to take place at distinct leaf developmental stages (Yuan et al., 2015). We obtained very strong, positive correlations between the numbers of salt glands and stomata, indicative of an indirect connection between salt gland and stomatal development. Moreover, we noted a positive correlation between stomatal number and pavement cell area, which was consistent with a previous study reporting that stomatal development is correlated with leaf area (Gay and Hurd, 2010).

Besides of the relationship between salt gland and pavement cell or stomata, there is an interesting phenomenon that no trichome distributed on the surface of L. bicolor. Given that trichomes also have the similar distribution pattern to salt gland, and trichomes from some species are reported to have secretory function (Wagner et al., 2004; Olsson et al., 2009), it is believed that there may be some homologous relationship between salt gland and trichome (Yuan et al., 2015). More will be verified by transformation the homologous genes involved in trichome development and salt gland development, such as a WD40-repeat protein (Yuan et al., 2019) and MYB transcription factor LbTRY (Leng et al., 2021).

Based on the former proposed salt gland development pattern (Yuan et al., 2015; Xu et al., 2020), we further improve the leaf development model (Figure 6) including MESCs (multipotent epidermal stem cells) stage, salt gland differentiation stage, stomata stage and pavement cell formation stage. The plant regulators proposed in the current report may directly act on the regulation of MESCs with promotion of SA and BL, and inhibition of GA₃, MeJA and ABA to further differentiate into different cell types. In conclusion, the leaf area of expanded leaves is a good direct indicator of the number of salt glands, stomata, and pavement cells. Salt glands on L. bicolor leaves can excrete excess Na⁺ out of the plant to avoid salt overaccumulation, but their development is not disconnected from that of the rest of the leaf. Our present results revealed connections between the development of salt glands, stomata, and pavement cells, which will be benefit to the further study of salt glands development.