In order to probe the effects of light condition altered by canopy gaps on the growth of understory tree seedlings in tropical forest, six native tree species from early- [Dolichandrone cauda-felina (Hance) Benth. et Hook. F. and Radermachera hainanensis Merr.], mid- [Syzygium cumini (L.) Skeels and Sterculia lanceolata Cav.], and late- successional stages (Dillenia turbinate Finet et Gagnep. and Cryptocarya chinensis Hance) were selected from Jianfengling tropical montane rainforest (E 108°46′–109°45′, N 18°23′–18°50′, Hainan, China) (Li, 2002; Fang et al., 2004; Sheng et al., 2012). The seeds for each species were collected from the understory layer under five parent trees during seed dispersal stage in Jianfengling natural reserve and germinated in shallow trays filled with forest topsoil (Yang et al., 2011). The parent trees were selected in habitat without canopy opening to avoid the effect of disturbance on selected seeds. After growing in a nursery with 5% full sunlight, 1 month-old tree seedlings for each species with 20 replicates were removed to three shade-houses from June. The three shade–houses were covered with plastic shade nets of different thicknesses to achieve 0–5% (as control, i.e., the light condition in tropical forest understory without disturbance) (Capers and Chazdon, 2004), 10–15% (SG, as light condition under small canopy gaps caused by heavy defoliation or broken branches) and 40–50% of full sunlight (LG, as light condition under large canopy gaps caused by fallen tree) (Clark et al., 1996; Tobita et al., 2010; Yang et al., 2017), respectively. All the seedlings were received natural rainfall, and sprayed monthly with 100 ml solution of NPK (ammonium, phosphate, and potassium) compound fertilizer (2 g l^–1) and a fungicide solution (50% Carbendazim, Pesticide Technology Development Co., Ltd., Wuhan Scarlett, China) twice during the experiment in order to control fungal infections. In order to calculate the seedlings’ relative growth rates for each species under Control, SG and LG treatments, the averaged initial biomass parameters (e.g., stem height, dry mass in leaves, stem and root) of each species were measured based ten additional replicates.

From July to October, the photosynthetic properties of five seedlings per species under three light treatments were measured from 09:00 to 11:30 a.m. under clear skies, by taking on five fully expanded leaves per plant. Light response curves were generated with a Li-6400 portable photosynthesis system (Li-Cor Inc., Lincoln, NE, USA) using the “Light Curve” automatic program (Yang et al., 2008). Leaves were allowed 10 min to acclimate to light intensity changes before measurements at light levels of 2,000, 1,500, 1,000, 500, 200, 100, 50, 20, 10, and 0 μmol m^–2⋅s^–1. Ambient temperature ranged from 24 to 28^°C, and leaf chamber temperature was about 25^°C. The leaf chamber environment was maintained at 370 mmol m^–2 s^–1 CO₂, 28 ± 2^°C leaf temperature and 65 ± 5% relative air humidity in the measuring chamber, respectively. Light response curves (A/PAR) were fit using non-rectangular hyperbola least square curve fitting procedure (Equation 1, Eq. 1; Lambers et al., 1998):

where AQE is the apparent quantum efficiency; PAR is photosynthetic available radiation; P_max (μmol CO₂ m^–2 s^–1) is the light-saturated rate of CO₂ assimilation; R_d (μmol CO₂ m^–2 s^–1) is the dark respiration rate; and k is the convexity or curvature factor. The light compensation point (LCP) was determined to be the light intensity on the light curve where the rate of photosynthesis exactly matches the rate of cellular respiration. The light saturation point (LSP) was calculated by the same equation, considering that LSP is the value when net photosynthetic rate (P_n) reaches 90% of P_max (Quero et al., 2006).

After 12 months of experimental duration in shade houses to create canopy conditions caused by disturbances, 10 seedlings per species from three treatments of light conditions (i.e., Control, SG, and LG) were harvested for relative growth rate in mass (RGR_m, Eq. 2). Before being dried at 72^°C in a forced air oven for 48 h for dry mass measurement, seedlings were separated into roots, stems and leaves, and washed for morphological characteristics determination, including specific leaf area (SLA, Eq. 3), leaf mass ratio (LMR, Eq. 4), stem mass ratio (SMR, Eq. 5), root mass ratio (RMR, Eq. 6), and leaf area ratio (LAR, Eq. 7).

where W₂ and W₁ are the final and initial total dry weights per plant; and T₂–T₁ is the growth time interval (i.e., 12 months, Yang et al., 2011).

where leaf total area of each seedling was measured using a Li-3000 leaf area meter (Li-Cor Inc., Lincoln, NE, USA, An et al., 2010).

The mean effect size of light condition changed by SG and LG on morphological and photosynthetic properties on tree seedlings from early-, mid-, and late-successional stages were calculated as Eq. 8–12 (Hedges et al., 1999):

where, i is the number of species of tree seedlings from early-, mid-, and late-successional stages; XS⁢G⁢o⁢r⁢L⁢G¯ and XC⁢K¯,n_{SG or LG}, and n_CK, S_{SG or LG} and S_CKare the mean value, number of replication and standard deviation of morphological and photosynthetic properties for tree seedlings in SG or LG, and control treatment, respectively. A significant mean effect size of light condition changed by SG and LG on morphological and photosynthetic properties was considered only when the 95% confidence interval (CI) did not overlap with zero (Zhou et al., 2014).

The importance of morphological and photosynthetic properties on seedlings’ RGR_m was expressed as %IncMSE (percent increase in mean squared error) using the package “RandomForest” (version 4.6-12, Liaw and Wiener, 2002) in R (R Core Team, 2015). The relationships among predictors with RGR_m were analyzed by Pearson correlation analysis. The effect of light conditions (Control, SG, and LG) and successional stage (early-, mid-, and late-) on the variables were examined by analysis of variance (ANOVA). The path analysis was performed using the “lavaan” package (version 0.6-12, Rosseel, 2010) in R to examine the effect of light conditions and successional stages on RGR_m through morphological and photosynthetic properties.

In the control groups with ambient light intensity, seedlings from different successional stages exhibited significant differences in the relative growth rate (RGR_m, p < 0.0001), with the lowest values of –8.61 (± 0.41 standard error, s.e.) mg g^–1d^–1 at early-successional stage and the highest value of 5.61 (± 0.80 s.e.) mg g^–1d^–1 at late-successional stage (Supplementary Table 1 and Figures 1A–D). Relative to control, both SG and LG induced significant increment of RGR_m for seedlings (P < 0.05, Supplementary Table 1 and Figure 1E). Improved light condition caused greater enhancement of growth for seedlings from both early- and mid-successional stage than that from the late-one, that SG and LG increased RGR_m of seedlings from both early- and mid-successional stage significantly, but not always of the seedling RGR_m of late- successional tree species (Figures 1A–C). Canopy gaps could affect the structure of understory community due to these distinct positive effects of increased light intensity on relative growth rates among seedlings from different successional stages (Figure 1).

Improved light intensity (i.e., in SG and LG) caused higher photosynthesis rates (P_n) relative to that in control, inducing 24.6–212.1, 35.0–107.8, and 17.8–50.2% increase of mass- and area-based light-saturated rate of CO₂ assimilation (A_mass and A_area) for species at early-, mid-, and late-successional stage, respectively (P < 0.05, Figures 2A–F). Leaf area (LA) and light saturation point (LSP) in both SG and LG were greater than that in control (P < 0.05, Supplementary Tables 1, 2), while the pattern of specific leaf area (SLA) was contrary (Figures 2G–I and Supplementary Table 2). Increased light intensity induced larger positive effects on both LA and LSP for seedlings from early-successional stage than that from mid- and late-ones (Figure 2). Relative to SG, LG induced greater effect on seedlings’ photosynthetic properties, especially for species from early-successional stage (Figure 2 and Supplementary Table 1). Both dark respiration (R_d) and light compensation point (LCP) displayed positive response to LG (P < 0.05) but not to SG (P > 0.05, Figures 2G–I, and Supplementary Table 2).

Both SG and LG caused positive effect on root: shoot ratio (R/S) and root mass ratio (RMR) for seedlings from all three successional stages, while only LG significantly decreased seedlings’ leaf area ratio (LAR) due to the dependence on both light condition and successional stages (P < 0.05, Figures 2D–F and Supplementary Tables 1, 3). Species at early-successional stage displayed the greatest decrease of leaf mass ratio (LMR) and increase of RMR in comparison with that at mid- and late-successional stage, especially for those at late- stage, seedlings displayed no significant decrease of LMR under both SG and LG (P < 0.05, Figure 2F and Supplementary Figure 1). Seedlings from early-stage displayed greater positive responses of both photosynthetic and morphological properties to enhanced light intensity, suggesting a larger advantage in growth in early-stage tree species over mid- and late-successional tree species (Figures 2, 3).

Among the predictor variables, we confirmed that light condition was the most important factor for seedlings’ RGR_m, followed by SLA, LA, LAR, successional stage, and LSP (Figure 1F). Light condition increased LA, RMR, R/S, R_d, LCP, LSP and A_area, but decreased SLA (Figures 2, 3, 5), and displayed significant interactions with successional stage on SLA, R_d, LCP and A_area (P = 0.021, 0.014, 0.012 and 0.048, Supplementary Table 1). RGR_m displayed positive correlations with A_area, LA, LSP and R/S (R² = 0.51, 0.39, 0.46, and 0.24, respectively, P < 0.05, Figures 4A–C, F), while exhibited negative relationship with SLA and LAR (R² = 0.52 and 0.45, P < 0.01, Figures 4D, E). The positive effects of improved light condition on RGR_m included direct and indirect pathways (through SLA and A_area, Figure 5). The regulation of species’ successional stage on RGR_m was through the path of SLA. Light condition, A_area and SLA all had direct path to affect RGR_m, and jointly explained 93% of seedlings’ RGR_m in total (P = 0.319, Figure 5). As one of crucial traits of plants, SLA represented the important regulation of successional stages on seedlings’ RGR_m in responding to enhanced light intensity after disturbance-induced canopy gaps (Figures 1F, 5).

Canopy opening readily alleviate the light limitation for understory plants in a forest (Lin et al., 2003; Lee et al., 2017). In our study, seedlings grown with light intensity to mimic small (SG) and large canopy gap (LG) displayed significantly higher RGR_m than that with ambient light intensity (Control, P < 0.05, Figure 1 and Supplementary Table 1). Moreover, we found that the development of leaf area (LA, Figures 2, 4) was one of the most important mechanisms to improve photosynthetic assimilation for plant individuals at enhanced light availability (Evans and Poorter, 2001; Tang et al., 2021). The formation of canopy gaps was verified to be beneficial for the understory regeneration in Jianfengling tropical montane rainforest (Feldmann et al., 2020), without considering the impacts on the direction of forest succession.

Exposed to increased light intensity, seedlings at early successional stage with a negative RGR_m in ambient light condition, exhibited the greatest improvement in their RGR_m (Figure 1), altering the survival rate, competition and dynamics of seedling community in the tropical forest understory subsequently (Valladares et al., 2000). In contrast, relative to seedlings from early- and mid- successional stage, those from the late-successional stage displayed a lower increment in RGR_m at increased light intensity (Figure 1), potentially confirming to the narrow niche breadth of species from the late-successional stage (Parrish and Bazzaz, 1982; Carscadden et al., 2020). Therefore, the canopy gaps due to defoliation (with the equivalent light intensity in SG) and tree fall (with the equivalent light intensity in LG) would trigger the changes of understory productivity, biodiversity and successional trajectory in disturbed patches, altering the structural heterogeneity in tropical forest (Takafumi et al., 2010; Lee et al., 2017).

The different response of RGR_m among seedlings from three successional stages could be firstly reflected by corresponding changes in leaf-scale photosynthetic properties (Figures 1, 2, 4), which is the foundation for regulating plant carbon assimilation to take the advantage from enhanced light due to temporary canopy gaps (Kneeshaw and Bergeron, 1998; Yao et al., 2015). Especially for the greatest improvement of RGR_m for seedlings at early successional stage could be ascribed to their changes in photosynthetic properties, e.g., a significant enhancement of light saturated point (LSP, Figures 2, 4). We suspect that greater LSP may have benefited the seedlings to achieve a higher saturated photosynthetic rate, such as the light–demanding species from the early stage of succession (Ackerly, 1996; Qi et al., 2004). Especially for light condition in LG, seedlings from early-successional stage displayed the largest improvement in both mass- and area-based light-saturated rate of CO₂ assimilation (A_mass and A_area, Figure 2), thus the more light penetrated into forest understory the more advantage would be found for seedlings from early-successional stage (Baul et al., 2022). Beside changes in LSP, light compensation point (LCP) and dark respiration (R_d) in seedlings from early successional stages may also have enhanced due to increased growth respiration driven by accelerated development of leaf and total biomass (Figures 1, 2; Marcelis et al., 1998; Inoue et al., 2022). Furthermore, seedlings from early- and mid- successional stages declined their specific leaf area (SLA, Figure 2) as a response to improved light conditions, especially in LG treatment, while that from the late- with an original successional stage lower SLA had no significant changes in SLA (Figure 2 and Supplementary Table 3), displaying the evolutionary conservatism in this functional trait among these plant species (Schweizer et al., 2013; Letcher et al., 2015). The intensity and range of disturbance would be crucial for species composition and successional process in understory patches in tropical forest, the larger canopy gap might brought more advantage for seedlings from early- and mid-successional stages (Yamashita et al., 2000; Liu B. et al., 2012).

Different from other photosynthetic properties with positive or negative responses, the apparent quantum yields (AQE) didn’t change even for the seedlings from the early-successional stage (Figure 2 and Supplementary Table 1), which agreed with the correlated results in comparing AQE between shade-tolerant species (i.e., from the late-successional stage) and pioneer species (i.e., from the early-successional stage) previous studies (e.g., Ramos and Grace, 1990; Marenco et al., 2001; Tsvuura et al., 2010).

Apart from the regulation in leaf-scale photosynthesis properties, seedlings would also alter allocation strategy of photosynthates to the organs acquiring the resource that strongly limits seedling growth (McCarthy and Enquist, 2007; Zhou et al., 2020). In this study, the sensitivity of root mass ratio was greater than that of leaf and stem mass ratio under improved light intensity (Figure 3; Poorter and Nagel, 2000). The adequate root development was considered to be crucial for improving seedlings’ survival in forest understory, since more belowground sources, e.g., nutrients and water, could support seedlings with an accelerated growth and a greater competitiveness in understory community (Landhausser and Lieffers, 2001; Myers and Kitajima, 2007). Correspondingly, the vertical biomass allocations between below- and above- ground biomass, reflected in root: shoot ratio (R/S) in these seedlings were altered by improved light availability (Figure 3; Poorter and Nagel, 2000). However, the significant positive response of R/S to both SG and LG exhibiting an increasing trend from late- to early-successional stage (Figure 3), that the seedlings from early- and mid- successional stages displayed a greater morphological plasticity than that from late- stage (Yan et al., 2006). The greater increment of R/S for seedlings from early- and mid- successional stages would be more beneficial for root nutrient absorption, particularly for phosphorus capture, in tropical montane forests, such as our study site (Liu et al., 2010, Liu F. et al., 2012). Thus, seedlings from early- and mid- could successional stages take advantage of the transitory opportunity of improved light condition caused by anthropogenic disturbances, regulating seedlings’ niche-partitioning and successional process in the understory of tropical forest (Dupuy and Chazdon, 2006; Yang et al., 2011).

In contrast to the positive correlation between R/S and RGR_m, leaf area ratio (LAR) exhibited a negative relationship with RGR_m (Figure 4). Under higher light intensity, per unit leaf area or shoot mass could provide greater productivity for plant individual than ambient light condition in understory (Kong et al., 2016), that seedlings with lower LAR and higher R/S potentially had a sufficient functions of above-ground organs, e.g., adequate photosynthetic capacity under increased light intensity (Shafiq et al., 2021). Therefore, canopy gap-induced greater light intensity could drive the trade-off between above- and below-ground biomass allocation, regulating morphological properties of seedlings for better resource acquisition and growing (Freschet et al., 2013; Baez and Homeier, 2018).

Our results show that the seedlings in the understory from different successional stages are characterized by diverse strategies in responding to improved light condition caused by canopy gap after disturbance (Figures 1–3). In ambient light condition of understory, seedlings from the late-successional stage had a higher RGR_m, ensuring a greater competitiveness relative to seedlings from early- and mid-successional stages in the understory tree community (Figure 2; Gao et al., 2016). However, greater light availability due to canopy gaps provided a potential opportunity for the seedlings from early- or mid- successional stages to offset the competitive advantage of late-successional plants (Abbas et al., 2020).

Seedlings’ RGR_m in the understory of tropical forest performed a dependence on the light condition, while the successional stage of species displayed a significant regulation on changes of RGR_m through effects on A_mass, LAR and SLA (Figure 5). Thereinto, SLA was a critical traits connected the effects of light condition and successional stage on seedlings’ growth, and playing the most important role for RGR_m after light condition (Figures 1F, 5; Baez and Homeier, 2018). For different seedlings, interspecific functional trait differentiation had been indicated to mainly associate with SLA, that at early-successional stage are generally characterized by a higher SLA relative to that at mid- and late- stages (Supplementary Table 3; Wang et al., 2012). Light condition and successional stage displayed a significant interaction on seedlings’ SLA, jointly affected seedlings’ RGR_m in the understory of tropical forest (Supplementary Table 1 and Figure 5; Kitao et al., 2000). Under the more frequency and intensity disturbances, the advantage of seedlings from late-successional stage would be weaken much more than that from early- and mid-successional stages, slowing down the process of forest forward succession.