Study Area

The research area was situated on the right slope of section No. 2 (YK296 + 300 – YK296 + 580) of the Laibin to Du’an highway, in Nongchang Village, Xincheng County, Laibin City, China. The slope extended over a length of 280 m, with a maximum vertical slope height of 26 m. The area was divided into three development levels, with a slope ratio of 1:1.75 for the first and second levels, and 1:2 for the third level. Active protective nets were employed for protection, along with the use of vines for greening purposes. Xincheng County is situated in a monsoon climate zone that transitions from subtropical to tropical regions, intersected by the Tropic of Cancer. The area experiences abundant sunshine and rainfall, with a mild climate. The terrain features erosion and hills, with ground elevations ranging from 145 to 180 meters and a relative height difference of 35 m. During autumn 2022, the area experienced minimal rainfall and severe drought conditions, as illustrated in Fig. 1. Between 17 September and 31 December 2022, the total rainfall amounted to 162 mm, with October recording only 17.8 mm of rainfall (Fig. 1).

Experimental Species

The experimental species selected were the small shrub Bougainvillea spectabilis (B. spectabilis) and the vine Pueraria lobata (P. lobata), which were planted on a first-level slope in August 2022. B. spectabilis is widely employed as a landscape plant in Guangxi, while P. lobata is commonly utilized to enhance vegetation coverage on highway slopes. The soil was locally backfilled yellow lime soil, with a soil bulk density of 1.29±0.12 g·cm^-3, soil mass moisture content ranging from 6.5% to 8.9%, salinity of 66.33±1.52 ms·m^-1, pH of 6.56±0.23, total phosphorus of 0.23±0.01 g·kg^-1, total potassium of 1.78±0.02 g·kg^-1, and a cation exchange capacity of 4.98±0.06 cmol·kg^-1. The soil was characterized by aridity and barrenness. The sampling took place on 22–23 October 2022. During the dry season, the two plant species exhibit a decelerated growth rate without the manifestation of desiccation-induced leaf abscission or chlorosis.

Fig. 1.

Temperature and rainfall in the study area during the dry season.

Light Curve Determination

Fully unfolded upper leaves were selected and a Li6400 portable photosynthesis instrument (LI-COR, Lincoln, NE, USA) was used from 08:00 to 11:30 to provide varying levels of photosynthetically active radiation through red and blue light sources. Radiation intensity was set to 2000, 1500, 1000, 800, 500, 300, 200, 150, 100, 50, 20, and 0 µmol m^-2s^-1 to assess the P_n of the two plants under different light intensities. During measurements, the leaf temperature was set at 32℃, the gas source CO₂ concentration at 400 µ mol·mol^-1, and the relative humidity at 18%. The experiment was conducted in triplicate, and the average value was reported with an error margin of < 10%. The gas flow rate was set at 500 μmol·s^-1 and the photosynthetically active radiation was measured using automatic measurement mode, with a dwell time of 120–200 s at each set value. The measurement period was from 22 to 23 October 2022.

A/C_i Curve Determination

On a sunny morning, with a light saturation intensity of 1000 μmol/(m²·s) for B. spectabilis and 1250 μmol·m^-2·s^-1) for P. lobata, an airflow velocity of 500 μmol·s^-1and a leaf surface temperature of 30℃, 13 CO₂ concentrations (400, 300, 200, 150, 100, 50, 400, 600, 800, 1000, 1200, 1600, and 2000 μmol·mol^-1) were set for CO₂ response curve measurement. An automatic measurement mode was selected with a dwell time of 60–300 s at each set value. The P_n of the two plants was measured and repeated three times, and the mean value of the three replicates was reported with an error margin of < 10%.

Determination of Meteorological Indicators and Soil Physicochemical Properties

Temperature and rainfall were recorded using a self-recording rainfall collector (Davis 6466M, Davis Instruments, Hayward, CA, USA). Soil pH, bulk density, cation exchange capacity, total phosphorus, and total potassium were measured according to the methods described by Lu [16].

Calculation and Analysis

The light curves of the tested plants were fitted and calculated using Photosynthesis Calculation software version 4.1.1, employing the right-angle hyperbolic modified model (RHMM), exponential model (EM), right-angle hyperbolic model (RHM), and non-right-angle hyperbolic model (NRHM) [17, 18]. Various indicators were calculated, such as the maximum net photosynthetic rate (P_{n max}), light compensation point (I_c), and dark respiration rate (R_d), along with other light response parameters and apparent quantum efficiency (AQE). The A/C_i curve of the tested plants was fitted and calculated using RHM, the Michaelis–Menten model (M-M), and RHMM [13, 14]. This analysis included calculating the initial slope of the CO₂ response curve, known as the initial carboxylation efficiency (α), along with response parameters such as P_{n max}, CO₂ compensation point (Γ), and photorespiration rate (R_p).

The coefficient of determination (R²), mean squared error (MSE), and mean absolute error (MAE) were employed to assess the goodness of fit for each model. An R² value approaching 1 indicates a closer match between the predicted or fitted values and the measured values, reflecting higher model accuracy [19–21]. Photosynthetic Calculation software (version 4.1.1) was utilized to compute R² directly and the following formulas were applied to calculate MSE and MAE:

where y_t and y^t represent the measured and fitted values, respectively. Origin 2020 software was used to statistically analyze and organize the data and draw charts.

Light Response Model Goodness of Fit Evaluation and Photosynthetic Parameters

The fitting results of the RHM, NRHM, RHMM, and EM for the light response of both species are presented in Table 1 and Fig. 2. The results indicate that, aside from the B. spectabilis RHM which exhibited a poor fit (R² = 0.6076), the fitting curves of other models for B. spectabilis and all four models for P. lobata demonstrated excellent agreement with the measured values (R² > 0.96). Except for the MSE and MAE values exceeding 1 for the B. spectabilis RHM, the MSE and MAE values for the other models were relatively small. Notably, EM for B. spectabilis exhibited the smallest MSE and MAE values, while NRHM for P. lobata demonstrated the smallest MSE and MAE values.

However, the methods used to estimate plant light saturation (I_sat) for these four light response models varied, leading to substantially different I_sat (Fig. 3). The I_sat value obtained for B. spectabilis using the NRHM model is notably lower than the measured value (Fig. 3). The I_sat estimated by the RHM and RHMM models closely approximated the actual values. However, the P_{n max} fitted by the RHM was 10, significantly higher than the measured value, whereas that estimated by the RHMM closely matched the observed value. Consequently, the larger the R² value, the higher the fitting degree of the light response model; however, the fitting results will not always match the measured values. Overall, the RHMM had the closest I_sat and P_{n max} to the measured values, with high fitting accuracy and the highest goodness of fit.

Fig. 2.

Four models fitting the photosynthesis response curves of B. spectabilis (I) and P. lobata (II).

Fig. 3.

Methods for estimating irradiance saturation (I_sat) using four photosynthesis response models. a. RHM; b. NRHM; c. RHMM; d. EM assumes that the I_sat is the photosynthetically active radiation (PAR) corresponding to a P_{n max} of 0.9_. The apparent quantum efficiencies (AQE) are the slopes of the linear fitting equations for the light response curve under PAR ≤ 200 μmol·m^-2·s^-1. a and b obtained the I_sat by solving the linear equation P_{n max} = AQE × I_sat – R_d (dark respiration rate).

Similarly, the I_sat obtained for P. lobata using the RHM, NRHM, and EM models were all notably lower than the measured values. In contrast, the P_{n max} fitted by the RHM and NRHM were significantly higher than the measured values. Only the RHMM had an I_sat and P_{n max} close to the measured values (R² = 0.99; MSE and MAE ≤ 0.21). Overall, the RHMM had the highest degree of goodness of fit. Therefore, when evaluating the goodness of fit of various photoresponse models, it is important to consider not only their fitting effects but also their accuracy in fitting the data.

Fig. 4.

CO₂ response curves of B. spectabilis and P. lobata for each model.

Given the optimality of the RHMM model, the data were compared using this model. The P_{n max} of P. lobata was 48% higher than that of B. spectabilis (Table 1). P_{n max} represents the peak photosynthetic capacity of leaves, indicating the maximum assimilation capacity achievable under specific environmental conditions [22]. The AQE of B. spectabilis and P. lobata were 0.0167 and 0.0267, respectively, with 37% higher in P. lobata. The AQE reflects the photosynthetic capacity of plants under low light conditions [18]. Previous studies showed that the range of AQY is from 0.03 to 0.05 in common plants under optimal conditions [10]. The AQE values are similar to those observed in Mangifera indica (0.020– 0.024) [7] and slightly lower than those of Populus euphratica (0.036–0.044) [10], both under severe drought conditions.

For P. lobata, the R_d was 29% higher than that of B. spectabilis, while the I_c was 6.68 times higher and the I_sat was 2.68 times higher than that of B. spectabilis. Overall, compared to B. spectabilis, P. lobata had a stronger ability to utilize both strong light and weak light. The B. spectabilis achieves organic compound preservation by minimizing dark respiration. Finally, P. lobata exhibits faster biomass growth under drought conditions and demonstrates a faster capability to cover exposed rocks on rocky slopes.

CO₂ Response Model Goodness of Fit Evaluation and Photosynthetic Parameters

Fig. 4 shows the fitting results of the RHM, M-M, and RHMM models for the CO₂ responses of both species. The fitting curves of both species using the three models demonstrated agreement with the measured values (R² > 0.98). The P_n values simulated by the RHM and M-M were consistent, with the RHM essentially being equivalent to the M-M model. The MSE and SAE of the RHMM for B. spectabilis were the smallest. From P_{n max}, the simulated values of the B. spectabilis RHMM were closest to the measured values (Table 2), whereas the RHM and M-M were far higher than the measured values, indicating a deviation from actual photosynthesis. The RHMM exhibited the best goodness of fit.

The RHM/M-M for P. lobata had the smallest MSE and SAE. For P_{n max}, the results of the three models were considerably higher than the measured values, with the RHMM displaying the smallest deviation from the measured values. Nevertheless, the fitting curve revealed that the measured values did not reach their maximum. Both the measured and fitted values increased with rising CO₂ concentration, and P_n did not exhibit a decreasing or stable pattern, indicating that the CO₂ saturation point (C_isat) could not be identified.

The α value represents the slope of the line obtained by fitting the response data of C_i ≤ 200 with a linear equation, reflecting both the quantity and enzymatic activity of Rubisco [23]. The α values of B. spectabilis and P. lobata were 0.0168 and 0.0883, respectively, indicating that P. lobata had a higher cooperative ability under low CO₂ conditions. The P_{n max}, Γ, and R_p of P. lobata were 1.89, 0.52, and 2.44 times higher than those of B. Spectabilis (Table 2), respectively. P_{n max} reflects the activity of photosynthetic electron transfer and phosphorylation in plant leaves [24]. The Γ reflects the CO₂ concentration at which the rate of photosynthetic assimilation equals that of respiratory consumption in plant leaves [25]. This suggests that P. lobata had a stronger ability to utilize low and high CO₂ concentrations compared to B. spectabilis. Both the R_p and the overall P_n of B. spectabilis were higher (Table 2). Overall, P. lobata has a greater potential for biomass accumulation during a continuous increase in atmospheric CO₂ concentration.

The C_isat reflects the ability of plants to utilize high CO₂ [26]. In contrast to previous studies, neither the measured nor simulated C_i of P. lobata reached saturation. Zhu et al. [21] found that under different nitrogen conditions, P_n gradually increased and eventually stabilized as the CO₂ concentration increased. However, Arp [27] observed that two Chinese Liriodendron plants grown at high CO₂ concentrations for an extended period did not reach saturation, resulting in significant changes in the C_isat of the plants grown under high CO₂ concentrations. In this study, P. lobata lived under extreme drought conditions for an extended period, preventing it from reaching the C_isat; however, the rate of increase gradually decreased. Photosynthesis under light saturation is constrained by its carboxylation ability at low CO₂ levels and by electron transfer and ribulose bisphosphate regeneration at high CO₂ levels [25].

RHMM is the Best Fitting Model for the Light and CO₂ Response Curves

Researchers have established various response models of photosynthesis, including common light response models such as RHM, NRHM, EM, and RHMM [28]. Previous studies [7, 10] found that the RHMM can better estimate I_sat than other models. This model simulates functions with extreme values and high accuracy, resulting in a fitting effect closer to the actual values compared to other models, particularly applicable to arid desert habitats [10].

The RHMM was also the optimal model for the light and CO₂ response curves in our results. Fitting the light-response curve offers two advantages: firstly, it can directly calculate the I_sat and the P_{n max} of plants. Secondly, it allows for the fitting of light response curves of plants under two conditions: either no decrease in P_n after reaching I_sat or decreased photosynthetic rate. The estimated I_sat and P_{n max} were closest to the measured values [7, 10, 26].

Similarly, during the fitting of the CO₂ response curve, neither the RHM nor the M-M models were able to fully capture the decline in P_n with increasing CO₂, nor could they accurately estimate P_{n max}. The RHMM demonstrated a good fit with an R² value and outperformed the other models in terms of fit. Its photosynthetic parameters, including α, the fitted P_{n max} values, and C_isat, closely approximated the true values and accurately reflected the actual situation.

Light Response and A/Ci Curves of Transpiration

Rate (T_r), WUE, Stomatal Conductance (G_s),

and Intercellular CO₂ Concentration (C_i)

A positive linear relationship was observed between T_r, G_s, and photosynthetically active radiation (PAR). WUE rapidly increased from 0 to 250 μmol·m^-2·s^-1 of PAR and subsequently stabilized. Contrary to the P_n, C_i exhibited a rapid decline from 0 to 250 μmol·m^-2·s^-1 of PAR before stabilizing (Fig. 5). The G_s of B. spectabilis remained relatively stable, whereas WUE exhibited a slight decrease after reaching its peak value (Fig. 5). The rapid decline in C_i indicates a high demand for raw photosynthetic CO₂ and high photosynthetic efficiency. Due to the small increase in G_s under low-light conditions, the increased photosynthetic rate rapidly increased CO₂ consumption, leading to decreased CO₂ levels [5].

Fig. 5.

Photosensitive curves of transpiration rate (T_r), water use efficiency (WUE), stomatal conductance (G_s), and intercellular CO₂ concentration (C_i).

Under weak light conditions (0–250 μmol·m^-2·s^-1), B. spectabilis had higher WUE and G_s compared to P. lobata, resulting in a higher P_n and lower C_i (Fig. 4). As the PAR increased, the T_r, WUE, and G_s of P. lobata rapidly increased, whereas the corresponding rise in B. spectabilis was slower. Consequently, this resulted in a rapid increase in the P_n of P. lobata, surpassing that of B. spectabilis, and a rapid decrease in C_i (Fig. 5). Under drought conditions, the two species exhibit distinct adaptive strategies. B. spectabilis employs a lower organic matter decomposition rate (lower R_d) and maintains stable G_s to reduce organic matter consumption and excessive water transpiration. However, solely enhancing WUE leads to a slight increase in P_n. This can be compared to the strategy of deciduous plants shedding leaves to reduce transpiration and entering a dormant state throughout the entire drought period [29]. In contrast, The P. lobata responds actively to increased light levels by augmenting G_s, T_r, and WUE, thereby enhancing the rate of organic matter accumulation, and exhibiting a performance similar to non-water-stressed state [5].

Fig. 6.

CO₂ response curves of T_r, WUE, and G_s.

WUE in both B. spectabilis and P. lobata was positively correlated with C_i. The T_r and G_s of P. lobata rapidly decreased at C_i levels between 0–250 μmol·m^-2·s^-1, followed by a slower decline (Fig. 6). The higher concentration of CO₂ enables plants to engage in photosynthesis more efficiently, thereby reducing the demand for stomatal opening. By reducing stomatal aperture, plants can effectively regulate water balance and decrease transpiration rates in environments with elevated CO₂ concentrations [30]. Meanwhile, B. spectabilis consistently maintains lower G_s and T_r to minimize water consumption (Fig. 6).

Indicative Significance of P. lobata and B. spectabilis in Response to Environmental Changes

Research indicates that water stress significantly impacts the light response of Cucumis sativus leaves, leading to a reduction in their photosynthetic capacity [21]. Water stress not only reduced gas exchange parameters, such as G_s, T_r, and WUE of Bothriochloa ischaemum, but also reduced the parameters of the P_n-PAR curves, such as P_n, AQE, and I_c. Water serves as the primary limiting factor for photosynthesis in B. ischaemum, and appropriate nitrogen application can enhance its potential photosynthetic capacity under water-deficient conditions [31]. When the relative soil moisture content falls below the optimal range, P_n and plant growth decrease significantly by influencing various cellular and physiological processes [32, 33]. Therefore, under severe drought conditions, the P_{n max} of B. spectabilis (2.38 μmol·m^-2·s^-1) and P. lobata (4.14 μmol·m^-2·s^-1) in this study were lower compared to Bidens pilosa (6.5 μmol·m^–2·s^-1), P. euphratica (8.41– 12.60 μmol·m^-2·s^-1) [10] and Mangifera indica (3.45– 3.97 μmol·m^-2·s^-1) [7].

Under low irradiance, photosynthesis is limited by the electron transfer rate, while under high irradiance, photosynthesis is often limited by Rubisco activity [34]. Consequently, as the light intensity gradually increased, the P_n of B. spectabilis and P. lobata also increased gradually and then stabilized after reaching the I_sat.

Generally, for species with C3 photosynthesis, elevated CO₂ leads to a 50% increase in P_n [35], and a 157% increase in plant biomass and leaf area [36]. Free-air CO₂ enrichment technology developed recently [37, 38] can increase the CO₂ concentration by 150% under field conditions, resulting in a 31% increase in the photosynthesis rate and an approximately 18% increase in crop yield [25, 39–41]. Although short-term exposure to elevated atmospheric CO₂ can lead to a significant increase in photosynthesis in many plants, long-term growth in elevated CO₂ typically results in a smaller increase in photosynthesis owing to reduced photosynthetic capacity [37, 38].

In the face of future frequent, extreme droughts and continuously increasing CO₂ concentrations, P. lobata is expected to exhibit better adaptation to changing water availability and atmospheric CO₂ levels than B. spectabilis; it maintained a higher photosynthesis rate and faster biomass accumulation, rendering it a superior species choice for restoring rocky desertification slopes. However, B. spectabilis is capable of surviving and maintaining a lower growth rate under drought conditions by retaining lower G_s and transpiration rates, thus fulfilling the landscape requirements of highways. Integrating vine and shrub planting in vegetation configuration contributes to soil and water conservation on rocky desertification slopes and enhances ecosystem stability [42]. Measuring the photosynthetic response of plants is a rapid method for screening plants suitable for this purpose. It should be noted that the challenges associated with capturing the extreme arid climate conditions resulted in the small sample size, short study period, and lack of replication of this study at different sites. These limitations could be addressed in future studies by expanding the sample size and conducting studies across multiple sites with similar conditions or using indoor simulations.