Alfalfa seeds were collected from Nanshan Forest Farm, Jiyuan, Henan, China. Planted alfalfa populations were used to collect seeds from the understory layer of a walnut (Juglans regia L.)-alfalfa agroforestry system (34° 56′−35° 04′ N, 112° 22′−112° 32′ E) in southern Taihang Mountain. Seeds were evenly distributed to planting cavities (212 ml; 7 cm × 4 cm × 13 cm, top Ø × bottom Ø × height) in trays filled with growing substrates (peat and perlite, 3:1, v/v). The tray surface was covered by a moist towel, and the moisture (>95%) was maintained every day by spraying with distilled water. When plantlets were germinated, the number of specimens was thinned to leave about 8–10 individuals per cavity to eliminate unnecessary competition. Thinned seedlings were cultured with a nutritional solution adapted from Kim et al. (1991). Solutions contained 1 mM potassium nitrate (KNO₃), 0.4 mM monopotassium phosphate (KH₂PO₄), 1 mM potassium sulfate (K₂SO₄), 3 mM calcium chloride (CaCl₂), 0.5 mM magnesium sulfate (MgSO₄), 0.15 mM dipotassium phosphate (K₂HPO₄), 0.2 mM iron–sodium ethylene diamine tetraacetic acid (Fe–Na EDTA), 14 μM boric acid (H₃BO₃), 5 μM manganese sulfate (MnSO₄), 3 μM zinc sulfate (ZnSO₄), 0.7 mM ammonium molybdate ((NH₄)₆Mo₇O₂), and 0.1 Mm cobalt chloride (CoCl₂).

Half of the alfalfa seedlings were cultured with water withdrawal (drought), and the other half were well-watered (control). The water deficit was induced by a 7-day withdrawal of water input. That is, drought-treated seedlings were watered every 14 days, while the controlled seedlings were watered every day. Thus, for alfalfa, drought stress can be induced after 7, 14, or even 21 days of water deficit, but only the 14-day period of drought-induced mostly frequent negative responses (Erice et al., 2011). Because the irrigation seedlings were watered to the pot capacity, the drought treatment and well-watered control resulted in different total volumes of water input. This accords with the total quantity of water used in the water deficit treatments of Erice et al. (2011). The seedlings that received contrasting rates of water were fed with the same dose of total nutrient input. Drought-treated seedlings were fed with nutrients on the same day water was supplied; the dose equaled that which the controlled seedlings received in a week’s time. During the experiment, the temperature was maintained in a range of 17°C and 34°C (night/day), and the relative humidity was maintained at 52%.

Alfalfa seedlings were raised under artificial LED lighting. Throughout the experiment, seedlings were exposed to a 12-h photoperiod from 08:00 am to 20:00 pm. This amount of time was shorter than that (~18 h/day) used for woody plants (Wei et al., 2020; Gao et al., 2021) because alfalfa has a faster-growing speed and does not need longer photoperiod exposure to promote growth. As alfalfa seeds were collected from an understory population beneath the canopy of a walnut agroforestry system, their lighting environment was simulated from a wide range of spectra that were tested for generated saplings of walnut (Gao et al., 2021). Three types of spectra were tested. The blue-light spectrum contained proportions of photosynthetic photon flux density (PPFD) of 33.7% red (600–700 nm), 48.5% green (500–600 nm), and 17.8% blue lights (400–500 nm). The red-light spectrum contained PPFD proportions of 71.7% red, 13.7% green, and 14.6% blue lights. The green-light spectrum contained PPFD proportions of 26.2% red, 56.4% green, and 17.4% blue lights. The test of ranged spectra was also employed on understory medicinal herbs (Guo et al., 2020; Zhou et al., 2021a; Zhou et al., 2021b).

Spectra were emitted by illuminations from LED panels (0.5 m × 1.2 m, width × length). It was determined that daytime PPFD under forest canopy ranged from 3.60 to 175.67 μmol m⁻² s⁻¹ (Wei et al., 2020). For each generation of walnut saplings, PPFD in touchable space was around 96 μmol m⁻² s⁻¹ (Gao et al., 2021), which fell in the range of PPFD in transmittance of sunlight. Therefore, PPFD was adjusted to be 97.88 μmol m⁻² s⁻¹ 10 cm beneath the LED panel. The LED panels were hung 50 cm over the tray. Three transformers were responsible for controlling the electrical currents of panel diodes. The 200-W transformer accounted for the electrical current adjustment of red-light diodes, and the 135-W transformer controlled that of green- and blue-light diodes.

The combined treatments of light spectra (df = 2) and water deficit (df = 1) were replicated three times, each of which was assigned as a tray of alfalfa seedlings. When the maximum height of most alfalfa per tray nearly reached ~50 cm (tips touched the panel), seedlings were mowed to remove all above-ground parts. Seedlings were clipped to mimic mowing about 50 days after sowing, and 58 days later the seedlings were clipped again to harvest for sampling. The mowing treatment was incorporated into the experimental arrangement as a repeated manipulation and did not increase the number of fixed-factor replicates. Most aerial organs were mowed, leaving shoots at a height of about 5 cm, as suggested by Shen et al. (2013). Mowed samples were divided into two halves. One-half of the samples were measured for height and then dried in an oven (70°C) for 72 h. Their dry mass (DM) was measured, and chemical analyses followed. The other half was freeze-dried and used for measuring physiological parameters.

Oven-dried samples were ground to pass a 1.0-mm screen. Soluble sugar and starch contents were determined by a colorimetric method (DuBois et al., 1956). A 0.5-g sample was used to calculate colorimetric measurement at 490 nm using an UV-Visible 8453 analyzer (Agilent Inc., San Francisco, CA, USA). Crude fiber and fat contents were determined using the standard methods endorsed by relevant national standards. Crude fiber determination was adapted from the method of SN/T 0800.8-1999 (2000) and crude fat from GB/T 6433-2006/ISO 6492: 1999 (2006). C-isotope discrimination was determined using freeze-dried samples that passed a 1-mm sieve. δ¹³C was determined using a mass spectrometer (Thermo Finnigan, CA, USA) following the equation:

where R _Sample and R _Standard are the ratios of ¹³C/¹²C in plant samples and the standard (Pee Dee Belemnite). Total C content was determined by an element analyzer (EA-3000, Boaying Tech., Shanghai, China).

Results were analyzed in a mixed-model analysis of variance (ANOVA), where light spectra and drought treatment were two fixed factors that were replicated three times, and seedlings were sampled twice pre/postmowing. The random placement of trays was designated as a random factor. SAS software (SAS Inc., Charlotte, NC, USA) was used to analyze the data. Factors of water deficit, mowing treatment, and light spectra were combined as a multiple-factorial interaction design. When significant effects were indicated, results were compared across treatments following the Tukey test (α = 0.05). To reveal the joint driving forces of ecophysiological parameters, multivariate linear regression was used to regress the contributions of crude fat and fiber contents. Pearson correlation was used to detect relationships between pairs of ecophysiological parameters.

Light spectra had an interactive effect with mowing on height and DM ( Table 1 ). Before shoot-mowing, the red-light spectrum induced greater shoot height compared to the blue-light spectrum ( Figure 1A ). Regarding the blue-light spectrum, postmowing seedlings had greater shoot height compared to those which had not yet been mowed ( Figure 1A ). Blue light also induced greater shoot height in the postmowing seedlings compared to green light. Water also had an effect on height and DM. Both height and DM increased in the well-watered plants compared to those under drought conditions ( Table 2 ).

In the seedlings that were not mowed, exposure to the red-light spectrum resulted in greater shoot DM than in those under the blue-light spectrum, as well as in the mowed seedlings ( Figure 1B ). Mowing increased shoot DM weight in the seedlings exposed to green- and red-light spectra.

Although watering conditions and light spectra had no interactive effects on shoot height ( Figure 2A ), their interactions had a significant impact on DM weight ( Figure 2B ). Drought-exposed seedlings were depressed to accumulate shoot DM in the blue- and green-light spectra, but water conditions did not change the shoot DM weight in the red-light spectrum ( Figure 2B ).

Mowing and light spectra had an interactive effect on soluble sugar and starch concentrations ( Table 1 ). Soluble sugar content was increased by 53% after mowing (before mowing, 4.05 ± 1.91 mg g⁻¹ DW; after mowing, 6.19 ± 1.85 mg g⁻¹ DW). The green-light spectrum resulted in higher soluble sugar content (6.77 ± 2.22 mg g⁻¹ DW) compared to the blue- (3.64 ± 1.70 mg g⁻¹ DW) and red-light (4.94 ± 1.52 mg g⁻¹ DW) spectrums. Starch content was the highest in mowed seedlings exposed to green light ( Figure 3B ). Starch content decreased during the drought treatment, while the change of soluble sugars was not significant ( Table 2 ).

Mowing increased crude fat content from 1.66% ± 0.77% to 2.86% ± 1.59% but decreased crude fiber content from 28.97% ± 3.86% to 25.88% ± 4.11%. The varying light spectra did not significantly affect crude fat content ( Table 1 ), which ranged between 1.7% and 2.8% among the three types of spectra ( Figure 4A ). Crude fiber content under the blue-light spectrum decreased by 32% and 30%, compared to that in the green- and red-light spectra, respectively ( Figure 4B ). The drought treatment induced a decrease in crude fat content but an increase in crude fiber content ( Table 2 ).

Mowing and light spectra had an interactive effect on δ¹³C and total carbon content ( Table 1 ). Mowing decreased δ¹³C for seedlings exposed to the red-light spectrum, but no variation of δ¹³C was induced by mowing in the blue- and green-light spectra. Before mowing, the blue-light spectrum induced lower δ¹³C compared to the green- and red-light spectra. However, after mowing, the variation of δ¹³C disappeared among spectra ( Figure 5A ). Before mowing, total C content was higher in seedlings subjected to the green- and red-light spectra, but a postmowing difference in total C content disappeared again ( Figure 5B ). The drought resulted in higher δ¹³C ( Table 2 ).

Mowing and water conditions did not result in a significant difference in δ¹³C ( Table 1 ; Figure 6A ). However, the drought treatment increased total C content regardless of whether the shoots had been mowed ( Figure 6B ). Mowing decreased total C content in drought-treated seedlings. Drought increased δ¹³C, fiber content, and total C content but decreased DM and fat content ( Table 2 ).

The linear regression model indicated three physiological parameters (shoot DM weight, soluble sugar content, and total C content) that may contribute to the accumulation of crude fat ( Table 3 ). However, the only sugar content was estimated to be a driving force for crude fat. DM weight and total C content were screened, and DM was further estimated as a significant parameter. Water-soluble sugar contributed to a positive correlation with crude fat content, which can be shown by a curve with a low slope (~0.18) with the narrowest 95% confidence falling in a range of 4.3−6.1 mg g⁻¹ DW. Shoot DM weight positively contributed to crude fiber content, whose 95% confidence was narrowest in a range of 2.1−3.3 g DW ( Figure 7B ).

Apart from forage-quality variables, physiological parameters had close relationships ( Table 4 ). Shoot DM weight had a positive correlation with soluble sugar content, which further had a positive correlation with starch content. Total C content had a positive correlation with δ¹³C.

We found that mowing can increase shoot DM weight in alfalfa when exposed to different light spectra. Moreover, shoot DM can be promoted by mowing in red- and green-light spectra, but no response was found in the blue light. A field trial also reported an increased DM production in alfalfa following mowing (Al-Gaadi, 2018). The increase in DM production in alfalfa populations resulted from the promotion of the regrowth of shoot parts under field conditions following mowing (Giese et al., 2013; Han et al., 2014). However, under blue light, mowing did not cause any changes in DM but did increase shoot height. DM increment in mowed alfalfa may be due to joint increments in plant organs such as flowers, buds, nodules, and initial shoots (Anower et al., 2017; Fiutak et al., 2019). This can account for the irrelevance of height growth with a null response of DM. Mowing was also found to increase the stem length of Aralia elata, a woody species, under plant factory conditions (Wei et al., 2020). Therefore, our first hypothesis can be accepted under the condition of blue-light radiation.

Our drought treatment limited shoot height across all light spectra, but water conditions did not affect DM weight in any of the light spectra. Aranjuelo et al. (2007) also reported that drought depressed growth and DM accumulation in nodules of alfalfa. Some accessions of alfalfa are extremely drought tolerant and showed more shoot biomass when subjected to drought (Anower et al., 2017). The cultivar, however, was not as tolerant. The results endorse parts of our results that the well-watered condition can promote the shoot growth of alfalfa.

It was proven that DM production in alfalfa can be easily modified by changing lighting spectra (Fiutak et al., 2019). Compared to the green-light spectrum, a blue-light–enriched spectrum more efficiently promotes dry matter accumulation in alfalfa’s aerial organs (Fiutak et al., 2019). Regenerated oak saplings also showed greater shoot biomass under blue light compared to their performance under green light (Gao et al., 2021). Our results did not follow the trend of these findings. The blue light resulted in lower shoot DM accumulation compared to both the green- and red-light spectra. This result was not influenced by mowing. In the interaction with water conditions, the red light only increased shoot DM accumulation. In dill (Anethum graveolens L.) and lettuce cultivars, red light caused greater DM production than blue light (Fraszczak, 2013; Clavijo-Herrera et al., 2018). The red-light spectrum resulted in greater shoot DM under drought conditions. Therefore, the red-light spectrum can be identified to benefit DM production in shoots.

Mowing increased starch content in alfalfa as a general effect, especially for those plants under green light. Annicchiarico et al. (2013) reported positive responses of increased starch accumulation in five alfalfa cultivars to a mown environment. They concluded this controls starch degradation. The accumulation of nonstructural carbohydrates is the main support in the regrowth of perennial root shoots (Annicchiarico et al., 2013). Both mowing and the green-light spectrum promoted shoot starch content in A. elata (Wei et al., 2020), which depended on its sprouts to regrow its shoots. In annual grasses, however, mowing caused uncertain responses with large variations (Peterson et al., 2013) or a complete failure to change (Wilen and Holt, 1996). However, not all perennial plants responded to the green light by increasing their starch content. For example, the tropical perennial plant Alpinia oxyphylla showed lower starch content in shoots under the green light as opposed to the red light (Zhou et al., 2021b). Starch in another temperate perennial plant, Allium victorialis, was not affected by different light spectra (Zhou et al., 2021a). Overall, mowing can benefit starch accumulation in alfalfa shoots as a stable effect, but its interaction with light spectra is species-specific.

Accumulating evidence suggests that nonstructural carbohydrate metabolism is a tradeoff between reserve and consumption. During the growing process, when photosynthetic assimilates are continuously used to produce carbohydrates, both coagulation and hydrolyzation occur in carbohydrate granules, and both sugar and starch exist in alternating high/low concentrations (Liu et al., 2021). During consumption, however, starch is intensively depleted for physiological demand while sugars can be accumulated (Wei and Guo, 2017). According to our research, both mowing and the green-light spectrum can increase the content of sugars. These findings suggest that alfalfa seedlings were subjected to a process that reserves photosynthetic production when exposed to green light and mowing. These two treatments did not generate any combined effects on sugar accumulation. The green light likely induced a control on hydrolyzations of both starch and sugars, but it had no further impact on activated conversions.

Mowing resulted in a decline of δ¹³C, suggesting higher conductance and gas exchange. In contrast, on abandoned cropland, it was found that mowing increased the δ¹³C of Artemisia frigida, suggesting elevated WUE (Diao et al., 2021). The elevation of δ¹³C is an evolutionary strategy to reserve water loss for a wide spectrum of plant species. However, elevated WUE indicated by δ¹³C is not needed during the evolutionary process for alfalfa. In our study, mown shoots showed higher nonstructural carbohydrates but lower C content. This suggests structural C depletion and likely reserved conversion of structural carbohydrates towards the nonstructural forms. This is evidence of resistance to drought by reserving sugar and starch post-mowing because the depletion of nonstructured carbohydrates is an immediate response to provide energy to fuel enzymic activity (Wei and Guo, 2017; Lauriks et al., 2022).

The elevation of δ¹³C in drought-treated alfalfa correlated with controlled conductance and lowered gas exchange (Erice et al., 2011; He et al., 2020). The depletion of starch was an alternative response to an increase in crude fibers, which further accounted for the increase in total C content. The decrease in sugar content corroborated the significant reduction in gas exchange and intercellular CO₂ concentration caused by drought (Gorthi et al., 2019; Du et al., 2020). In regard to this study, unchanged sugar content means the close of the stomata did not cause high intercellular CO₂ to a level that enforced sugar decline. The reduction of starch content in drought is driven to balance the high demands of glucose following limited photosynthesis (Abdelhakim et al., 2021).

Green light induced consistently higher levels of δ¹³C and total C relative to the blue light, which accounts for the above-mentioned increases in DM production and growth. Together with dual increases of soluble sugars and starch contents, it can be concluded that the green light promoted DM accumulation by controlling consumption and accumulating structural and nonstructural carbohydrates. However, these responses to light spectra were interrupted by the mowing. Thus, postmowing differences between δ¹³C and total C content were dismissed. Overall, mowing is the stronger driver compared to light spectra.

Fiber is a type of structural carbohydrate and, following mowing, decreased during the experiment. This result concurs with our hypothesis that fiber was decreased by mowing. Davies et al. (2009) also reported a decrease of crude fibers in mown Artemisia tridentata ssp. wyomingensis, and they further revealed that the decrease was mainly attributed to the decline of acid detergent fiber. Increased crude fat content in mowed alfalfa was also found in forage bermudagrass (Cynodon dactylon [L.] Pers.) (Zhang et al., 2020). Alfalfa is a legume C3 plant, and bermudagrass is a C4 plant dwelling in a warm climate. Both alfalfa and bermudagrass are perennial, and their shoots regrow to stubbles after mowing; their responses of fat accumulation reflect a common physiological consequence of shoot removal. Decreased fiber and increased fat suggest better forage quality in alfalfa. In contrast, drought decreased the forage quality by increasing crude fiber content and decreasing crude fat content. All of these responses were reported in previous studies on grasslands (Grant et al., 2014; Delfani et al., 2022). The lower content of crude fiber in the blue-light spectrum indicates improved forage quality, which, in addition to previous conclusions, demonstrates that blue light limits DM production but improves forage quality.

We found that across treatments, crude fat contents were positively associated with soluble sugar content. Soluble sugars can be dissolved in glycerin because of the formation of hydrogen bonds between glucose and glycerin molecules (van der Sman, 2017). These changes were irrelevant from the exposure to different light spectra because both soluble sugars and crude fat revealed scarce responses to light treatments. DM weight was found to be positively correlated with crude fiber content, which was formed due to dual changes following mowing, drought, light spectra, and their interactions. This means that the increase in DM accumulation is an alternative approach to improving forage quality by increasing fiber content.