The alfalfa seed (Alfalfa cultivar ‘Zhongmu No. 3’) was provided by a specialized seed producer (Xintai Zhouquan Agricultural Technology Co., LTD, China). The similar size of healthy and plump seeds was screened and then sorted into three groups (brown, yellow and green) according to the seed coat color by visual inspection. The seed proportion was calculated by counting the number of seeds with different colors. Yellow seeds comprised the major part (59%) of the seed lot used for this study, while the relative amount of brown seeds was 33%. Green seeds comprised only 8% of the seed lot ( Table 1 ). GA3, D-IAA, D-ABA, D-GA4 were purchased from OlChemIm Ltd. (Olomouc, Czech Republic). IAA was purchased from LGC Standards Ltd. (Tetdington, UK). ABA was purchased from Sigma-Aldrich (Merck, Germany).

The alfalfa seeds were sterilized with 75% alcohol for 30 min, rinsed with distilled water three times to remove the remaining alcohol, and then soaked in distilled water. Thirty soaked seeds of different colors were incubated on double-layered filter paper that was placed in each 12-cm-diameter Petri dish with 8 mL solutions, i.e. CK (0.00 mol/L NaCl solution), 0.05 mol/L NaCl solution, 0.10 mol/L NaCl solution, 0.15 mol/L NaCl solution, and 0.20 mol/L NaCl solution. All experiments were performed with six replicates. Petri dishes were placed in the controlled climatic chamber with a photoperiod of 12 h/12 h (day/night). The temperature was set to 25°C, the light was set to 5000lx, and relative humidity was set to 65% in day time, while they were set to 20°C, 0lx and 65% at night time. Seeds were provided with additional solutions separately to prevent drying. Germinated seeds (as judged by the appearance of a visible 1 mm radicle) were counted every day until the number stopped increasing, and finally, the germination test lasted for 7 days. On the seventh day, the sprout and root length of seedlings in each petri dish were measured. For those petri dishes with more than 20 seeds germinated, the sprout height and root length of 20 randomly selected seedlings were measured, followed by seedling fresh and dry weight measurement. For those petri dishes with less than 20 seeds germinated, all the seedlings in the petri dish were measured. The germinability and final germination rate were estimated to describe the influence of seed color and salt stress on the germination process.

The germinability (Ga) refers to the total germination on which day the daily germinated seed reaches its peak. Based on the pre-experiment, the germination peak of alfalfa seeds was on the 3^rd day in this study, so it was calculated as:

e final germination (Gf) rate was calculated as:

Before analysis, all variables were tested for normal distribution by Kolmogorov-Smirnov’s test, the homogeneity of variances was examined by Levene’s test, and the final germination and the sprout length of yellow seeds were log-transformed to meet the require of homogeneity. The statistical significance of seed color on seed traits, seed germination process and seedling growth were evaluated using one-way analysis of variance (ANOVA) and least significance difference (LSD) test. For data that did not conform to homogeneity of variances after transformation, Mann-Whitney U test was used to evaluate the effects of seed color. The analyses were performed with IBM SPSS 22.0 (IBM Corp., Armonk, NY, USA), and significance was examined at a level of P<0.05. All data are presented as mean ± standard error (SE).

Seed color affected the germination parameters significantly (P< 0.05). The germinability and final germination percentage of yellow and green seeds were significantly higher than that of brown seeds ( Figure 1 ), but they showed no significant differences between green and yellow seeds (P > 0.05).

Salt stress had significant influence on the germination parameters ( Figures 1 and 2 ). Generally, the germinability and final germination percentage decreased with the increase of salt stress. Take green seeds as an example, seed germinability and final germination percentage under CK (0.00 mol/L), 0.05 mol/L and 0.10 mol/L salt stress were not significantly different, but they were significantly higher than that under 0.15 mol/L and 0.20 mol/L salt stress (P< 0.05) ( Figure 1 ).

Seed color affected seedling growth significantly in high salt stress ( Figures 2 and 3 ). In the treatment of 0.15 mol/L and 0.20 mol/L NaCl, the sprout height, root length, seedling fresh and dry weight of green and yellow seeds were significantly higher than that of brown seeds ( Figures 2 and 3 ). Particularly, most brown seeds had no sprout under high salt stress.

Salt stress had great influence on seedling growth ( Figures 2 and 3 ). The sprout height, root length, seedling weight were highest under 0.05mol/L salt stress, while these growth indexes were lowest under 0.20 mol/L salt stress ( Figures 2 and 3 ).

The variation of EC over time is consistent among the three-color seeds ( Figure 4 ). The leachate values for EC increased fast within 4h, and gradually became stable after soaking 12 h. The leachate values for EC varied with seed color significantly. The brown seeds had the highest EC, and the EC of green seeds were larger than yellow seeds (P< 0.05).

Changes in seed weight and volume reflected the water uptake capacity. The seed weight and volume changed rapidly within 4h after soaking, and then became stable after the 8 h ( Figure 5 ). The seed weight and volume of three colors were significantly different at the 2 h, shown as brown seeds > yellow seeds > green seeds (P< 0.05), and the difference among the three colors became not significant over the soaking time. The weight and volume of brown seeds increased fast than that of green and yellow seeds, especially in the first 4h, the increase rate of three colors presented as brown seeds (109.2% and 181.2%) > yellow seeds (81.2% and 114.0%) > green seeds (45.2% and 81.7%).

The thickness of the palisade layer in the ventral of seed was shown that brown seeds (29.95 ± 1.33μm) > green seeds (28.96 ± 0.53μm)> yellow seeds (28.92 ± 1.38μm), but the differences among the three colors were not significant (P > 0.05) ( Figure 6 ). As for the palisade layer in the back of seed, there was no significant difference either (brown 40.46 ± 2.46μm > yellow 35.39 ± 2.87μm> green 34.14 ± 1.86μm) ( Table 2 ).

The hormone content remarkably varied with seed color and hormone type ( Figure 7 ). The content of IAA and GA3 in brown seeds were significantly higher than those in yellow and green seeds (P< 0.05), but there was no significant difference between yellow and green seeds (P > 0.05). The content of ABA showed a significant difference among the three colors (brown: 21.02ng/g > green: 6.00ng/g > yellow: 2.92ng/g) (P< 0.05). The value of (IAA+GA3)/ABA in yellow seeds was larger than that in green and brown seeds, but it showed no significant difference among the three colors (P > 0.05).

Due to the long flowering episode, the maturity of seeds in one lot is quite different at the same time and the seed coat acquires its own color along maturity (Lv et al., 2021). In our study, the yellow alfalfa seeds comprised the major part of the seed lot, and yellow seeds had higher vigor and germination parameters (germinability and final germination percentage), so it is recommended that yellow seeds could be the first choice in seed screening and planting for marginal land. More importantly, we found that the germination parameters and seedling growth indicators (sprout and root length, seedling weight) of brown seeds were not significantly different with control (no salt stress) in low salt stress (0.05 mol/L and 0.10mol/L NaCl) ( Figure 1 ), but the germination ability of brown seeds fell sharply under high salt stress (0.20 mol/L NaCl), i.e. the germination rate decreased from 63.9% to 7.3%. However, the germination parameters of green and yellow seeds decreased with the aggravation of salt stress, and the seedling growth indicators under low salt stress were greater than control and high salt stress. Compared with green and yellow seeds, brown seeds were more sensitive to salt stress. And not as expected, the seed germination ability and seedling growth of brown seeds were lower than green and yellow seeds under different salt stress. Therefore, contrary to the first hypothesis, salt resistance of alfalfa seeds did not increase with the deepen of seed color. In fact, the salt resistance of brown seeds was significantly lower than that of green and yellow seeds.

Seed germination and seedling establishment was affected by seed vigor, while seed color is a morphological trait associated with seed vigor (Murray et al., 2004; Moles et al., 2005; Yang et al., 2008). In this study, the results of electrical conductivity showed that the vigor of yellow alfalfa seeds was highest, and the vigor of brown alfalfa seeds was lowest. These findings were consistent with the results of red clover which reported that yellow seeds had higher vigor than brown and purple seeds (Atis et al., 2011).

Seed coat permeability limits the water uptake, which is the primary process of seed germination (Bewley et al., 2013). Previous studies showed that seed color leads to differences in water uptake (Ertekin and Kirdar, 2010). In our study, brown alfalfa seeds had a higher water uptake rate than green and yellow seeds at the early stage ( Figure 5 ). Previous research reported that the cuticle and palisade layers in the seed coat caused the impermeability, so a thicker cuticle and palisade layers may reduce the rate of water uptake. However, the results of our study showed that the seed coat thickness of alfalfa was not significantly different among colors. It indicated that seed coat thickness may play little role in restricting the water from entering the seed. In addition, we found that seeds absorb water quickly did not germinate quickly and had a lower germinability and germination percentage ( Figure 1 ). Our results were different from those of Zhang et al. (2008), they found that yellow rape seeds had rapid water uptake than black seeds. But our results were similar with the research of Atis et al. (2011), they observed that brown seeds of red clover had higher water uptake rate and lower germination percentage. The higher water uptake capability and lower germination of brown seeds may be caused by the damage of the seed coat structure. It has been reported that seed coat structure gradually breaks with the aging of seeds (Srii et al., 2022). Thus, the brown seeds lose membrane integrity, and they were prone to imbibitional injury, resulting in excessive water absorption and less oxygen (Mcdonald et al., 1988; Asiedu and Powell, 1998; Yang et al., 2008; Doria et al., 2019). The results that brown seed showed a faster change rate in electrical conductivity compared with green and yellow seed also exemplified that reason.

Plant hormones regulate seed germination and dormancy (Graeber et al., 2012, Jordan et al., 2017). In this study, the content of ABA, GA3, and IAA remarkably varied with seed color. The hormone content in brown seeds was significantly higher than those in yellow and green seeds, and noteworthy, the content of ABA in brown seeds was significantly higher than IAA and GA3. In the early stage of seed development, a higher concentration of growth promoting hormones (IAA and GA3) is required to promote the formation of the embryo body (Miransari and Smith, 2014); and then with the maturation of the seed, the content of ABA increased (Andrade et al., 2008; Matilla and Matilla-Vazquez, 2008). The difference in hormone content resulted in the lowest germination index of brown seeds ( Figure 1 ). In addition to this, seed germination is not only related to the absolute content of hormones but also depends on the balance between various hormones (Baskin et al., 2000; Wang et al., 2018). The ratio between promoter and inhibitor [(IAA+GA3)/ABA] in yellow seeds was larger than that in green and brown seeds. Therefore, although the hormone content of yellow seeds including IAA and GA3 is lower than that of brown seeds, the germination parameter and seedling growth of yellow seeds are higher than brown seeds under salt stress. The findings of seed germination in our study were following those of Atis et al. (2011). They also reported that the brown-colored seeds had the lowest germination percentage (58%), whereas the yellow-colored seeds had the highest germination percentage (99%). It confirmed that both the content and the balance between IAA+GA3 and ABA played important role in the germination of seeds with different colors. Thus, the results of seed water uptake, seed coat thickness and hormone content supported the second hypothesis.

Based on the results, the green and yellow seeds germinated at a high rate (more than 95%) under normal conditions, while the brown seeds had a significantly lower germination rate (less than 50%), this difference may also associated with seed dormancy. According to Baskin and Baskin (2004), the green and yellow alfalfa seeds that used in the study was non-dormant, while brown alfalfa seed was more influenced by physiological dormancy (Type 1, non-deep PD). The balance of ABA and GA regulates the onset, maintenance and termination of physiological dormancy (Baskin et al., 2000). The change degree of ABA and GA over time may be different among three colors, which might influence the germination process and stress resistance. The water-impermeable palisade layers in seed coat hinder the water uptake, and then reduce the germination percentage. On the other hand, the broken seed coat may result in a very quick water absorption, which in turn lead to imbibition injury. Therefore, seed coat integrity and structure might have potential function in stress resistance. Except that the seed coat thickness was measured in a relatively small sample size, our study only examined the initial hormone content and seed coat thickness of one alfalfa cultivar. Further studies are needed to examine the hormone change and seed coat structure during the germination in alfalfa seeds of different dormancy type.

Despite the limitation, our study demonstrated the importance of the seed color with respect to the screening of seeds with high salt resistance and provided new insights for seed adaptability in terms of seed heteromorphism.