The experiment lasted about 5 months from April to September 2018 at the Fanggan Research Station of Shandong University, Shandong Province, China (36°26′N, 117°27′E). The seeds of R. pseudoacacia and A. fruticosa collected from Shandong Province in the early winter of 2017 were obtained from Qiluyuanyi Seed Company (Linyi, China). The area has a warm temperate monsoon climate, with an annual precipitation of 700 ± 100 mm and an average temperature of 13 ± 1°C. During the experimental period, the microclimate in the greenhouse was monitored with HOBO data loggers (U12-012, Onset, Bourne, MA, United States). Mean air temperature was 29.6°C (18.7–36.7°C) during daytime and 20.8°C (10.2–27.5°C) during nighttime, and mean relative humidity was 59.3% (28.2–97.8%) during daytime and 93.6% (56.3–100%) during nighttime. Seedlings were maintained at 60–70% of field water capacity throughout the experiment utilizing daily irrigation. The seeds were germinated in deionized water on plates and transferred into plastic pots (26 cm in depth and 24 cm in diameter) containing 7 kg of growth substrate (one seedling per pot). The plant growth substrate was a mixture of air-dried sandy loam and humus soils in proportions of 2:3 by volume.

For each species, 3-month-old seedlings of similar size were selected and randomly assigned to the following treatments. For each species, a factorial experiment of two factors (light and defoliation) was designed. The seedlings were randomly assigned to two light conditions: (1) high light treatment (L+), grown in the greenhouse covered by plastic films (88% of natural sunlight); (2) low light treatment (L−), conducted in greenhouse covered by woven black nylon nets (8% of natural sunlight). Under each light treatment, plants were divided into two groups. One group was submitted to top-down 50% defoliation, the upper half leaves were defoliated (DE); and the other was served as a control group without leaf removal (CK). The treatments were conducted from July 12 to September 20, lasting 70 days. There were 16 replicates in each treatment for each species. Four individuals of each species and treatment were randomly selected to measure in each sampling. During the experiment, seedlings were harvested at days 1 (July 13), 10 (July 22), 30 (August 11), and 70 (September 20).

Seedling height and basal diameter (BD, at approximately 1 cm above the ground) were recorded at each harvest. Four seedlings were harvested from each treatment at around noon (12:00–3:00 pm) and separated into roots (include root nodules), stems, and leaves. Then, the samples were oven-dried (30 min at 105°C, followed by 72 h at 75°C) and weighed. Total biomass (TB), leaf mass ratio (LMR), stem mass ratio (SMR), root mass ratio (RMR), and root-shoot ratio (R/S) were calculated as follows:

where RB is root biomass, SB is stem biomass, and LB is leaf biomass.

Non-structural carbohydrates (NSCs) of four replicates per treatment were measured after each harvest. After biomass determination, dried samples were grounded with a ball mill (JXFSTPRP-24, Jingxin, Shanghai, China) to analyze NSCs (defined as the sum of starch and soluble sugars) concentration in the leaves, stems, and roots (Cao et al., 2018). Soluble sugars (SS) were extracted twice with 80% ethanol, and starch (ST) content was measured after subjecting the solid residue of each sample to a washing step and hydrolysis. The absorbance of the extracts was measured at 620 nm (UV-9000S, Metash, Shanghai, China) after an anthracenone-sulfuric acid reaction. The concentrations of soluble sugars and starch (measured as glucose equivalents) were calculated on dry mass basis (mg g^–1).

Gas-exchange characteristics of seedlings were measured before each harvest. The maximum photosynthetic rate (A), transpiration rate (E), intercellular carbon dioxide concentration (C_i), and stomatal conductance (G_s) of fully expanded leaves from four replicates (one leaf per seedling) for each measurement were measured using a portable gas exchange measurement system (Li-6800, Li-Cor, Lincoln, NE, United States). These measurements were conducted between 9:00 and 12:00 h on sunny days. During the measurements, photosynthetically active radiation, temperature, relative humidity, and CO₂ concentration inside the leaf chamber were controlled at 1000 μmol m^–2 s^–1, 28°C, 50%, and 400 ppm, respectively. The 1000 μmol m^–2 s^–1 PAR was high enough to obtain the maximum photosynthetic rate according to a pre-experiment. Since all the upper half leaves were defoliated, we measured the gas exchange parameters of mature leaves in the middle position of the individuals in all treatments at days 1 and 10. Due to the reflushing leaves were fully expanded at day 30 after defoliation, we measured the gas exchange parameters on newly mature leaves in all treatments at days 30 and 70. Instantaneous water use efficiency (iWUE) was calculated as follows:

Specific leaf area (SLA) of four replicates per treatment was also measured after each harvest. Fully expanded and healthy composite leaves (fifty leaflets in total per seedling) from four seedlings in each treatment were scanned, and images were analyzed with the WinFOLIA Pro 2009a software (Regent Instruments, Inc., Quebec, QC, Canada) to determine leaf area. After scanning, these leaves were oven-dried for 72 h at 75°C and weighed. SLA was calculated as leaf area/leaf dry mass. The collection of leaf samples was the same as the gas exchange measurement.

Some other leaf morphological and physiological traits were also measured after 70 days of treatment. Four healthy and fully expanded leaves from the upper reflushing plant part (one leaf per seedling) were selected to determine leaf chlorophyll concentration in each treatment in both species. After extraction with 95% ethanol (v/v), chlorophyll concentration was determined using the spectrophotometric method (Lichtenthaler and Wellburn, 1983). The absorbance (A) was measured at 645 and 649 nm (UV-9000S, Metash, Shanghai, China), Chlorophyll a concentration (Chl a), Chlorophyll b concentration (Chl b), the ratio of chlorophyll a/b (Chl a/b), total Chlorophyll concentration (Chl_total) were calculated as follows:

where V is ethanol volume, and FW is fresh weight.

Leaf thickness was measured with an electronic digital micrometer for four replicates per treatment and was averaged from 20 fresh leaves per seedling. Leaf lignin concentration was estimated by a spectrophotometric procedure using acetyl bromide, and absorbance (A) was measured at 280 nm (UV-9000S, Metash, Shanghai, China). The calculation of the lignin concentration was according to the standard curve (Iiyama and Wallis, 1990).

Three-way analysis of variance (ANOVA) was applied to detect the effects of light treatment, defoliation treatment and sampling time for plant traits in each species. One-way ANOVA was used to detect difference among treatments for each species, and Duncan’s multiple comparison tests at α = 0.05 were followed when significant differences were observed. Before ANOVA, the data was checked for normality (Shapiro–Wilk test) and homogeneity of variance (Levene test). ANOVA and Spearman correlation was performed using IBM SPSS Statistics 23.0 (IBM Corp., Armonk, NY, United States), and the figures were illustrated using OriginPro 2016 (Originlab Co., Northampton, MA, United States). To evaluate the relationship between plant traits and experimental treatments, redundancy analysis (RDA) was carried out using the vegan package in R Statistical Software v.4.0.3 (Oksanen et al., 2019).

Seedling height, basal diameter and total biomass were affected by time and light treatments, but the interaction of light and defoliation had no significant effect on the growth indexes in both species (Supplementary Tables 1, 2). Owing to the defoliation of upper leaves, the total biomass of the defoliation treatments was significantly lower than control treatments at day 1 in both species (Figures 1E,F). Starting from day 10, the plant height and basal diameter in L+ treatments were higher than L− treatments in both species (Figures 1A,B), and total biomass in defoliation treatments could catch up the seedlings in control in R. pseudoacacia (Figure 1E). After 30 days, the total biomass in defoliation treatments could recover to the control level in A. fruticosa (Figure 1F). After 70 days, the seedling height, basal diameter and total biomass in L + treatments were all significantly higher than L− treatments in both species (Figure 1).

Defoliation treatment significantly affected the LMR and SMR in both species, light treatment significantly affected the biomass allocation index in A. fruticosa (Supplementary Tables 1, 2). The LMR in the L−DE treatment was higher than L−CK treatment at day 30 in R. pseudoacacia but not in A. fruticosa (Figures 2A,B). Starting from day 30 day, the RMR and R/S in the L+ treatments were higher than L− treatments in A. fruticosa but not in R. pseudoacacia (Figures 2E–H).

Light treatment significantly affected the SLA in both species (Supplementary Tables 1, 2). Defoliation treatment and the interaction of light and defoliation treatment significantly affected the SLA in R. pseudoacacia but not in A. fruticosa (Supplementary Tables 1, 2). Compared with the L−CK treatment, the SLA of the L−DE treatment increased significantly from day 10–30 in R. pseudoacacia (Figure 3A). Compared with the L + DE treatment, the SLA in the L−DE treatment increased significantly during day 10–70 in both species (Figures 3A,B).

After 70 days, the Chl a/b and leaf thickness in L− treatments were lower than L+ treatments, while the total chlorophyll concentration in L− treatments was higher than L+ treatments in both species (Figures 4A–F). The leaf lignin concentration of L−DE treatment was significantly lower than L + DE treatment in both species (Figures 4G,H).

Light and time treatments significantly affected the gas exchanges in both species (Supplementary Tables 1, 2). The maximum photosynthetic rate of the L− treatments was lower than L+ treatments starting from day 30 in both species (Figures 5A,B). After 70 days, there was no significant difference in transpiration rate and stomatal conductance among treatments in both species (Figures 5C–F). The iWUE of L− treatments was significantly lower than that of L+ treatments after 70 days in A. fruticosa but not in R. pseudoacacia (Figures 5G,H).

Light treatment significantly affected the concentration of seedling NSC, stem SS, root SS, leaf ST, and stem ST in both species (Supplementary Tables 1, 2). Defoliation treatment significantly affected the root ST concentration in R. pseudoacacia but not in A. fruticosa (Supplementary Tables 1, 2). The interaction of light and defoliation treatment significantly affected the root SS and root ST concentration in A. fruticosa but not in R. pseudoacacia (Supplementary Tables 1, 2).

For R. pseudoacacia, shading treatment significantly decreased the total NSC concentration in the later stage (days 30–70), but not in the early stage (days 1–10) (Figure 6G). Before the reflushing leaves matured (days 1–10), the leaf ST concentration in defoliation treatment decreased, and the leaf SS concentration remained unchanged (Figures 7A,B). After the new leaves matured (days 30), the leaf NSC concentration in defoliation treatment remained unchanged, the stem NSC concentration was significantly reduced, and the root NSC concentration was significantly increased (Figures 6A,C,E). After 70 days, the stem SS and ST concentration, root SS and ST concentration in L−CK treatment were significantly lower than L + CK treatment (Figures 7E,F,I,G).

For A. fruticosa, defoliation and light treatment significantly decreased seedling total NSC concentration at day 1 (Figure 6H). After 10 days, there was no significant difference in leaf, stem and root NSC concentration between L + CK treatment and L−CK treatment (Figures 6B,D,F). After 30 days, the leaf and stem ST concentration in L−CK treatment were significantly lower than L + CK treatment, the leaf and stem SS concentration remained unchanged (Figures 7C,D,G,H). After 70 days, the root SS and ST concentration in L−CK treatment were significantly lower than L + CK treatment (Figures 7K,L).

Redundancy analysis was performed for both species after 70 days. The first two axes explained 67.48% of the variation in R. pseudoacacia (Figure 8A), and 60.53% of the variation in A. fruticosa (Figure 8B). SLA was positively correlated with total chlorophyll concentration, while negatively correlated with the leaf thickness and leaf lignin concentration. Low light treatment was positively correlated with SLA and total chlorophyll concentration, but negatively correlated with maximum photosynthetic rate, leaf thickness, leaf lignin concentration, and growth parameters in both species. Defoliation treatment was negatively correlated with RMR and R/S in R. pseudoacacia, while positively correlated with stomatal conductance (Figure 8).

After 70 days of recovery, compensatory growth in response to defoliation was observed in both species, with a greater compensatory total biomass growth in high light versus low light availability treatments. Our results corroborate previous research in other species, which showed that moderate (50–66%) defoliation did not affect growth in Nothofagus solandri var. cliffortioides seedlings (Mikola et al., 2000). However, low light treatments after 70 days significantly altered growth, and leaf morphological and physiological parameters of both species. Plant height, basal diameter, total biomass, maximum photosynthetic rate, Chlorophyll a/b, and leaf thickness all decreased in low light treatments, while SLA and total leaf chlorophyll concentration increased (Supplementary Tables 1, 2 and Figures 1, 3–5). These results showed that plants can undergo plastic changes in morphology and physiological characteristics to adapt to changing light conditions, which is also an important way for plants to adapt to environmental heterogeneity (Õunapuu-Pikas and Sellin, 2020).

In our study, SLA was positively correlated with total chlorophyll concentration in both species (Figure 8). SLA is a major trait in the worldwide leaf economics spectrum, which reflects nutrient and dry mass investment in leaves (Wright et al., 2004). A higher SLA and total chlorophyll concentration grown under low-light conditions could help plants increase the efficiency of light capture and maximize carbon gain by reducing the diffusion resistance of CO₂, leaf construction and maintenance costs and the self-shading of the chloroplast (Rijkers et al., 2000; Xu et al., 2009; Gommers et al., 2013; Takahashi and Obata, 2014; Freschet et al., 2015). Other studies suggest that low chlorophyll a/b is considered to be a characteristic of plant’s shade tolerance, which also could improve the light-capture efficiency in low light environment (Hansen et al., 2002). The plastic response of SLA to shading may result in thinner, and relatively larger, leaves (Liu et al., 2016). In our study, results in low light treatment were consistent with leaf cost-benefit strategies (Zhu et al., 2016). Seedlings reduced leaf structural investment by reducing leaf thickness and leaf lignin concentration, thereby increasing SLA and total chlorophyll concentration to increase the investment in light efficiency capture (Figures 3, 4), which suggested that plants could achieve high carbon gain at a low leaf carbon cost under low light conditions. The RDA analysis also proved that there was a trade-off between leaf structural investment and light capture efficiency during low light treatments (Figure 8). The increase in SLA under low light conditions was concurrent with declines in leaf thickness and leaf lignin concentration, which was also an adaptation strategy in low light conditions to achieve high carbon gain.

This experiment demonstrated that plant growth and NSC concentration were reduced in L−DE treatment compared with L + DE treatment after 70 days in R. pseudoacacia (Figures 1, 6). The result supports our first hypothesis that seedling growth after defoliation would be significantly suppressed under low light conditions, as the production of carbohydrate is not enough. However, in the early period, plant growth was also reduced in L−CK treatment compared with L + CK treatment, although the NSC concentrations were maintained at control levels after treatments for 10 days in R. pseudoacacia (Figures 1, 6). The results showed that growth suppression might not be caused by insufficient supply of NSCs. The previous study showed that plant growth was more sensitive to stress than photosynthesis, resulting in NSC accumulating (Weber et al., 2019). Thus, the decrease of NSC concentration as an indicator of C limitation is not accurate.

In the early experiment (days 1–10), we found that stems soluble sugar concentration decreased significantly after defoliation, while the leaves soluble sugar concentration remained stable in R. pseudoacacia at day 1 (Figure 7). After removing the upper leaves, the production of new leaves requires a lot of carbohydrates, thereby remobilizing carbon stored in the stems. After 10 days, the leaves starch tended to be more depleted in defoliation treatment compared with controls in both species (Figure 7), indicating that leaf flush relies on stored NSC reserves (Merryn et al., 2018). Meanwhile, under low light treatment, defoliation significantly decreased LMR and increased SLA, but did not significantly affect RMR in R. pseudoacacia at day 10 (Figures 2, 3), suggesting that seedlings adapt to defoliation and light stress in the early stage by changing leaf morphology and physiological characteristics rather than root growth. From the gas exchange characteristics, A. fruticosa was a species with high rate of photosynthesis and transpiration. The transpiration rate and stomatal conductance of L− treatments were higher than L+ treatments, but there was no significant difference in the maximum photosynthetic rate between L+ treatments and L− treatments at day 10 (Figure 6). Previous study had shown that shade-tolerant species often retained open stomata in low light conditions (Hetherington and Woodward, 2003). The carbon limitation caused by L + DE or L−CK did not change the NSC concentration in A. fruticosa, but the L−DE treatment reduced the NSC concentration. This indicated NSC levels would not decrease as growth declined under moderate carbon limitation, but as limitation became severe, NSC levels would decline (Erin et al., 2013), which also showed that defoliation had a greater impact on seedlings under low light conditions.

In the late phase of the experiment (days 30–70), growth and carbon allocation recovered rapidly after defoliation under high light conditions but not under low light conditions in both species. With the prolongation of the shading stress, the plant NSC concentration, height, basal diameter, total biomass, and maximum photosynthetic rate in the L− treatments was lower than that of L+ treatments in both species at day 30. Thus, shade stress can inhibit photosynthesis and cause insufficient carbon supply, resulting in carbon limitation, and limited plant growth (Sala et al., 2012), which was consistent with previous study that plants reduced their photosynthetic capacity in response to low light availability (Rossatto et al., 2010). In addition, the NSC concentration in different organs was consistent with their respective functions (Figure 6). Under moderate defoliation, the reduction in growth showed significant trend of recovery over time. However, under low light treatments, the plant height, basal diameter and total biomass were smaller in both species after 70 days compared to the high light treatment (Figure 1). The results showed that insufficient carbon supply would inhibit plant growth and carbon storage (Ida et al., 2012). However, there was no significant difference in the total NSC concentration in L−CK treatment compared with the control after 70 days in A. fruticosa, but the growth parameters of L−CK treatment was significantly lower than control group. As an adaptation to increasing carbon limitation, plants can prioritize maintaining NSC concentration at the expense of growth, which could help plants survive long periods of minimal C gain (Erin et al., 2013; Elise and Jason, 2020).

Previous studies showed that herbivore damage could vary significantly between the contrasting light habitats (open vs. shade) on various species (Miljković et al., 2018). Under high light conditions, defoliation treatment had a significant impact on physiological indicators and carbon allocation during the early recovery period (days 1–10), but in the later recovery period (days 30–70), defoliation had little effect on plant growth and physiological characteristics. Under low light conditions, defoliation treatment first reduced the growth parameters, SLA, and then reduced the plant’s gas exchange parameters and total NSC levels. In the early stage of the experiment, defoliation and light treatment had a significant impact on leaf physiological and morphological traits, and less on the root growth. As the stress continued, the root growth was also severely restricted in the later stage of the experiment. Previous studies have shown that adaptation of plant to variation in light during herbivore activity depends on plant resource allocation (Piotr et al., 2010).

It has been shown that R. pseudoacacia seedlings can recover their total biomass in about 10 days after defoliation under high light condition, but A. fruticosa seedlings could catch up their total biomass with control within 30 days (Figure 1). These results showed that the R. pseudoacacia seedlings recovered faster than A. fruticosa seedlings after defoliation. Low light treatments significantly altered the growth and biomass allocation strategy of A. fruticosa. LMR in L− treatment was significantly higher than the L+ treatment at day 30, while RMR in L− treatment was significantly lower than the L+ treatment in A. fruticosa (Figure 2), which indicated that plant invested more carbohydrate into the growth of aboveground organs than of roots, consistent with the functional equilibrium hypothesis that resources are allocated to organs which are responsible for acquiring the most limiting resource (Poorter et al., 2012).