Investigating the Impact of Species Diversity on the Carbon Sequestration Potential of Forest Vegetation Based on Solar-Induced Chlorophyll Fluorescence

Highlights

What are the main findings?

• Species diversity has a positive effect on vegetation carbon sequestration potential at the national scale.
• Forest origin significantly modulates this relationship, with natural forests showing a stronger effect of species diversity.
• The contribution of species diversity increases with forest succession. These findings highlight forest origin and succession as critical factors shaping the biodiversity–ecosystem functioning relationship.

What are the implications of the main findings?

• Our study provides a scientific basis for conserving natural forests, promoting the ecological transformation of plantation forests.
• Managing carbon sinks in alignment with successional dynamics.

Abstract

Forest species diversity plays a critical role in regulating vegetation carbon sequestration potential. However, the mechanisms by which species diversity influences carbon dynamics under varying forest conditions are not yet fully understood. In this study, we used solar-induced chlorophyll fluorescence (SIF) as a proxy for carbon sequestration potential and integrated nationwide forest plot data collected through standardized protocols. A random forest model was employed to examine the influence of species diversity on carbon sequestration potential across forests differing in origin and succession. The results revealed that: (1) species diversity has a positive effect on vegetation carbon sequestration potential at the national scale (r ≈ 0.15); (2) forest origin significantly modulates this relationship, The importance index of species diversity in natural forests (0.409) was significantly higher than that in planted forests (0.043), with a relative contribution exceeding that of planted forests by approximately 20% and (3) the contribution of species diversity increases with forest succession. These findings highlight forest origin and succession as critical factors shaping the biodiversity–ecosystem functioning relationship. Our study provides a scientific basis for conserving natural forests, promoting the ecological transformation of plantation forests, and managing carbon sinks in alignment with successional dynamics.

1. Introduction

Forests in China are assuming growing significance within the global ecological framework. Statistics show that China holds the fifth-largest forest area and the sixth-largest timber stock worldwide, while boasting the largest planted forest area globally. As the predominant carbon pool in terrestrial ecosystems, forests outperform all other land ecosystems in productivity, accounting for more than two-thirds of the annual carbon fixed by terrestrial ecosystems [1]. Nonetheless, forests worldwide are confronted with compounded pressures from excessive logging, persistent degradation, and climate change, endangering nearly half of all tree species globally [2]. Driven by human activities and global change, accelerated species extinction is precipitating a biodiversity crisis and substantially undermining ecosystem functioning and sustainability [3,4]. These impacts are reflected in the disrupted structure of native plant communities, diminished self-recovery capacity of ecosystems, impaired provision of ecosystem services, and fragmented connectivity of ecological corridors [5]. Consequently, there is heightened ecological concern and increased scientific focus on biodiversity and forest productivity.

Species diversity serves as a key driver of ecosystem productivity, yet the underlying mechanisms remain complex. Although numerous studies report a positive correlation between species diversity and productivity [6,7,8], negative [9], hump-shaped [10], and nonsignificant relationships [11] have also been documented. These divergent patterns are commonly explained by the interplay of two principal mechanisms: complementarity effects and selection effects [12,13,14,15,16,17]. Complementarity effects arise from niche differentiation, whereby species evolve distinct ecological adaptations, leading to divergent resource-use strategies and reduced interspecific competition. Additionally, diversity confers resilience through an insurance effect: under environmental variability or disturbance, species with differing functional traits can compensate for one another, thereby maintaining continuous ecosystem functioning. In contrast, selection effects enhance biomass primarily by favoring dominant species with high productivity. For example, in the South subtropical region, forests with high species richness tend to have longer phenological periods, which helps reduce interspecific competition and boost photosynthetic capacity [18]. Furthermore, diverse plant communities benefit from variations in photosynthetic mechanisms and physiological traits, allowing competitively superior species to utilize resources such as light, water, and nutrients more efficiently. This leads to a marked increase in gross primary productivity (GPP) at the ecosystem level [19,20]. The rise in GPP not only augments vegetation biomass directly but also stimulates litter production and accumulation. As a key phase in plant life cycles, litter supplies microorganisms with ample substrates for decomposition, thereby accelerating the transfer of carbon into the soil. Through microbial processing, carbon is effectively stabilized within the soil matrix, enhancing soil organic carbon storage and fertility [21]. This interconnected series of processes reinforces the carbon sequestration capacity of ecosystems. Hence, conserving and restoring species diversity is imperative not only for ecological stability but also for advancing sustainable development goals. Strengthening research on the biodiversity–ecosystem function relationship will improve our understanding of ecosystem complexity and support evidence-based conservation strategies.

GPP denotes the total carbon fixed via photosynthesis per unit time and area within an ecosystem, forming the foundational substrate for all subsequent carbon allocation and storage processes. Consequently, the accurate assessment of GPP is essential for understanding the global carbon cycle and projecting future climate change. Advances in remote sensing technology have enabled the emergence of solar-induced chlorophyll fluorescence (SIF) as a promising tool for the direct, large-scale estimation of forest GPP [22,23]. Satellite-retrieved SIF constitutes optical signals emitted by chlorophyll in leaves upon absorption of photosynthetically active radiation, typically within the 650–800 nm wavelength range under solar illumination. The absorbed energy in leaves is predominantly partitioned among three competing pathways: photosynthesis, fluorescence, and thermal dissipation [24]. This energy partitioning mechanism provides a physical basis linking SIF to the photochemical efficiency of Photosystem II (PSII) and the rate of photosynthetic electron transport. The yield of SIF thus reflects the energy allocation status within PSII [25]. Owing to this intrinsic link, SIF strength is closely related to photosynthetic capacity and can be employed as a direct probe for photosynthetic activity. In contrast to traditional satellite-derived vegetation indices based on canopy greenness (e.g., the Normalized Difference Vegetation Index, NDVI; the Enhanced Vegetation Index, EVI), SIF offers distinct advantages. It is capable of detecting subtle physiological variations within the canopy, demonstrates higher sensitivity to short-term vegetation dynamics [26,27,28], and responds more rapidly to external environmental stressors. This advantage is supported by a growing body of research. For example, Hou et al. [29] evaluated the capacity of vegetation indices (VIs) and SIF to track GPP variations across seasonal scales and during droughts, demonstrating that SIF was superior in capturing these changes compared to the tested VIs. Similarly, Zhou et al. [30] compared SIF and NDVI for monitoring GPP and crop yield across various land cover types and found SIF to be more effective. Furthermore, in an analysis comparing NDVI, EVI, and SIF for assessing vegetation productivity dynamics in polar regions, Qiu et al. [31] reported that traditional vegetation indices were more susceptible to environmental interference than SIF.

Since both solar SIF and GPP are theoretically governed by absorbed photosynthetically active radiation and photosynthetic efficiency, and share a fundamental linear relationship [32], satellite-derived SIF provides a robust tool for global monitoring of GPP through remote sensing [33]. Numerous studies have demonstrated strong linear correlations between SIF and GPP across a wide range of ecosystems [34,35,36,37]. Globally, SIF performs reliably under diverse environmental conditions, including regions such as Asia [38], North America [39], and tropical zones [40,41]. under varying environmental conditions [42,43], for different vegetation types [34], and over multiple temporal scales [44]. These findings demonstrate that SIF is effective in capturing ecosystem photosynthetic activity and serves as a key indicator of GPP [45,46,47].

Research across various forest types further elucidates the relationship between species diversity and carbon sequestration. Osuri et al. [48] evaluated carbon capture stability by comparing short-rotation commercial monoculture plantations with species-rich natural forests. Their findings indicated that natural forests maintained higher temporal stability in carbon capture across both dry and wet seasons. Moreover, species-poor plantations demonstrated weaker resilience to climate variability in their carbon sequestration capacity. In a study of two urban forests in Nigeria, Agbelade et al. [19] reported that the forest with a higher Shannon diversity index possessed greater total carbon stocks, alongside higher volume yield and biomass. Examining humid and semi-humid forests in Mexico, Martínez-Sánchez et al. [49] identified positive linear correlations of carbon stocks with forest age and with indices combining biomass and species metrics. Complementing this, Forrester et al. [50] highlighted that the positive effect of species diversity on forest productivity is particularly pronounced under resource-limited or harsh climatic conditions. Other research notes a divergence in carbon sequestration capacity between natural and planted forests [51], and that GPP typically increases with stand age and successional stage before declining in mature forests [52,53]. Collectively, these studies underscore that the impact of species diversity on the carbon sequestration potential of forest vegetation is contingent upon factors such as forest origin, stand age, and successional stage.

Despite these findings, the physiological mechanisms linking species diversity and photosynthesis remain poorly understood. Cao et al. [54] employed SIF to examine this relationship and reported a global positive correlation between species richness and photosynthetic rates, with the strongest associations observed in tropical forests. Their study also found that the effect of species diversity on photosynthesis was second only to temperature. Mechanistically, species diversity was shown to influence maximum photosynthetic capacity rather than extend the growing season. However, how these mechanisms vary across forests of different origins and at different stages of forest succession remains unclear. Further research is needed to determine whether GPP responses to environmental factors in forests with contrasting origins and succession histories modify the relationship between species diversity and GPP.

This study integrates data from 7800 forest plots surveyed using standardized protocols with a high spatiotemporal resolution SIF dataset to investigate the influence of species diversity on vegetation carbon sequestration at the national scale. Unlike model-derived GPP estimates, SIF is directly linked to photosynthetic processes, allowing assessment of species diversity effects on GPP potential at the physiological level. Therefore, SIF was adopted as a proxy for GPP in this study. We propose the following hypotheses: (1) Species diversity is positively correlated with the carbon sequestration potential of forest vegetation; (2) The effect of species diversity on carbon sequestration potential differs between natural forests and plantations, with a stronger effect in natural forests; (3) This effect varies across successional stages. Based on prior studies and data availability, the following variables were selected as predictors in the random forest model: Shannon Index (SI), mean annual temperature (MAT), mean annual precipitation (MAP), humidity index (P/PET), soil moisture (SM), root to shoot ratio (R/S), litterfall, and canopy cover. Improving understanding of the relationships among these factors and validating their relevance under varying environmental conditions contributes to a theoretical foundation for ecosystem function analysis. This knowledge also supports evidence-based policy development and promotes biodiversity conservation and sustainable land management.

2. Materials and Methods

2.1. Data Source

The forest plot dataset used in this study was derived from nationwide field surveys carried out between 2011 and 2015 under a systematic, stratified, and standardized protocol, with surveys conducted mainly from June to September each year [55]. Based on China’s 1:1,000,000 vegetation map, the territory was divided into grids of different sizes: 100 km² in species-rich regions, and 400 or 900 km² in other areas. Each plot consisted of a 1000 m² main area (600 m² for some plantation plots), subdivided into ten 10 × 10 m subplots. After plot establishment, a systematic tree survey was performed. Survey teams also collected uniform geographic information and completed standardized records for each plot, including: (1) forest origin (natural or planted) and (2) species identity of every individual tree. Data from individual plots were integrated using area-weighted averaging and further validated through machine-learning-based spatial modeling. This standardized design covers China’s major forest types and climatic gradients, offering a consistent and reliable observational basis for estimating forest carbon stocks at the national scale.

SIF data were obtained from the China regional SIF dataset developed by Tao et al. [56]. This dataset was produced using a weighted stacking algorithm that integrates CatBoost, random forest, and gradient-boosted decision tree models, alongside multi-source remote sensing inputs. These inputs include MODIS vegetation indices, land surface temperature, the fraction of absorbed photosynthetically active radiation (fPAR), ERA5-Land meteorological variables, and SRTM topographic features. The resulting dataset offers a high spatiotemporal resolution of 500 m and 8 days, spans the period from 2000 to 2022, and effectively captures spatial and temporal variations in vegetation photosynthetic efficiency and carbon sequestration potential.

Meteorological data used in this study included MAT and MAP for the period 2011 to 2015. These data were obtained from the national ecosystem observation and research network (www.cnern.org.cn).

2.2. Research Methods

2.2.1. Meteorological Data

For each plot, values were extracted from the nearest meteorological station based on geographic coordinates, and elevation correction was applied to better represent local climatic conditions.

2.2.2. Stages of Successional Development

Following the standard ‘Classification of Age Classes and Age Groups for Major Tree Species’ (LY/T 2908-2017) [57] issued by the national forestry administration, this study categorized plots into three forest succession stages: early, middle, and late stages. This classification system takes into account both species-specific growth characteristics and regional climate conditions. For example, fast-growing species typically reach the late stage earlier than slow-growing species, and the same species tends to grow faster in warmer regions than in colder ones, thereby entering later stages of succession more quickly. The succession stage of each plot was determined based on the average age of its dominant tree species.

2.2.3. Calculations of Humidity Index

The moisture index is calculated based on annual precipitation and annual potential evapotranspiration (PET) [58]. The PET formula is:

$$PET=0.1651\times D\times V_d\times K\times 365$$

(1)

In the equation, D represents the duration from sunrise to sunset in multiples of 12 h, varying with the date, latitude, slope, and aspect of the watershed (if slope and aspect effects are ignored, the average daily D for the entire year is 1); $V_d$ denotes the saturated vapor density at the annual mean temperature (g·m⁻³), $V_d=216.7\ast V_s/(T+273.3)$, where $V_S$ is the saturated vapor pressure. $V_s=6.108\ast \mathrm{e}\mathrm{x}\mathrm{p}[17.26939\ast {\displaystyle \frac{T}{T+273.3}}]$; K is the correction factor that adjusts the PET calculated by the Hamon method to its actual value.

2.2.4. Calculations of Shannon Index

The SI is calculated using the following formula [59]:

$$SI={\sum }_{i=1}^S-p_i\times ln(p_i)$$

(2)

In the formula, SI denotes the Shannon index, S represents the total number of species or categories in the sample; $p_i$ is the relative abundance of the ith species, $p_i$ = ${\displaystyle \frac{n_i}{N}}, n_i$ is the number of individuals of the ith species, and N is the total number of individuals of all species.

2.2.5. Calculations of SIFmean and SIFmax

Based on the geographic coordinates of each plot, SIF data for the corresponding locations from 2011 to 2015 were extracted. For each year, observations from June to September were selected, and the mean and maximum SIF values during this period were calculated separately. These seasonal values were then averaged across the five-year period to derive the mean (SIFmean) and maximum (SIFmax) SIF values for each plot. SIFmean represents the average activity level of the photosynthetic light reactions during the sampling period and reflects overall photosynthetic intensity. In contrast, SIFmax captures the maximum photosynthetic potential under optimal environmental conditions, indicating the upper limit of carbon sequestration potential during this time.

2.2.6. Correlation Analysis and Random Forest

This study employs Pearson’s correlation coefficient to assess the relationship between factors. The calculation formula is as follows:

$$r={\displaystyle \frac{{\sum }_{i-1}^n(X_i-\overline{X})(Y_i-\overline{Y})}{\sqrt{\sum _{i-1}^n{(X_i-\overline{X})}^2} \sqrt{\sum _{i-1}^n{(Y_i-\overline{Y})}^2}}}$$

(3)

here, r denotes the correlation coefficient, X_i and Y_i represent the ith sample of X and Y respectively, $\overline{X}$ and $\overline{Y}$ denote their respective sample means, and n is the sample size.

In building and analyzing the random-forest model, we used permutation importance to assess the relative contribution of each predictor. This metric quantifies a feature’s importance by randomly shuffling its values in the test set and measuring the corresponding change in model performance, offering an intuitive measure of its influence on the model’s predictions. Additionally, A random forest model was established to evaluate the impact of species diversity on SIF under different forest conditions. Model hyperparameters were optimized using five-fold cross-validation combined with grid search, considering the following key parameters: number of decision trees (100, 200, 500, 1000, 2000), maximum tree depth (unconstrained, 5, 10), and minimum sample size required for node splitting (2, 5, 10). For each hyperparameter combination, the coefficient of determination (R²; higher values preferred) and root mean square error (RMSE; lower values preferred) were calculated. These performance metrics were ranked, weighted, and aggregated to identify the optimal model configuration. All analyses were conducted using Python (3.14), R (4.5.0), and Origin (2026) software.

3. Results

3.1. Spatial Distribution Patterns of Species Diversity and SIF

Figure 1 illustrates the spatial distribution characteristics of SI across China The results show that pure forests (SI = 0) do not exhibit a significant spatial pattern at the national scale. In contrast, SI values in mixed forest areas display substantial spatial variation. Overall, SI tends to be lower in the inland northwest and relatively higher in the southeastern coastal regions. Both SIF and SI exhibit significant latitudinal gradient changes, decreasing markedly with increasing latitude. (p < 0.01). SIFmax showed a moderate positive correlation with latitude (r = 0.25), whereas SIFmean and SI were only weakly correlated (r = −0.07, r = −0.13). Both SIF and SI showed a significant increasing trend with longitude (p < 0.01). SIFmax exhibited the strongest correlation with longitude (r = 0.64), followed by SIFmean (r = 0.58), while SI showed the weakest correlation with longitude (r = 0.17).

3.2. The Overall Relationship Between Species Diversity and SIF

Figure 2 presents the person r between SIF and various environmental and ecological factors. Both SIFmean (r = 0.14) and SIFmax (r = 0.13) showed significant positive correlations with SI at the 0.01 significance level. In addition to SI, SIFmean exhibited significant positive correlations (p < 0.01) with climatic factors including MAT, MAP, and P/PET, as well as with litterfall, canopy cover, and SM. It also showed a significant negative correlation (p < 0.01) with the R/S. In contrast, SIFmax was significantly positively correlated (p < 0.01) with climatic factors, litterfall, and canopy cover, but showed no significant correlation with SM or R/S. Overall, SIFmean demonstrated stronger correlations with environmental and ecological factors compared to SIFmax.

3.3. Relationship Between SIF and Species Diversity Under Different Forest Conditions

3.3.1. Characteristics of Factors Influencing Forest Origins

Table 1. Statistical Table of Relevant Characteristics of Natural and Plantation Forests.

	Natural Forests	Plantation Forests
SI	0.973 * (0.017)	0.320 * (0.010)
SIFmean (mW m−2 nm−1 sr−1)	0.322 * (0.002)	0.382 * (0.002)
SIFmax (mW m−2 nm−1 sr−1)	0.467 * (0.002)	0.514 * (0.002)
MAT (°C)	9.293 * (0.127)	13.829 * (0.125)
MAP (mm)	1025.643 * (9.112)	1164.117 * (10.274)
P/PET	1.459 * (0.007)	1.320 * (0.008)
SM (%)	26.971 * (0.098)	26.419 * (0.143)
R/S	0.257 * (0.001)	0.245 * (0.002)
Litterfall (t ha−1 a−1)	4.634 * (0.050)	4.129 * (0.052)
Canopy cover (%)	65.438 * (0.313)	63.230 * (0.447)
Altitude (m)	1392.91 * (21.61)	554.60 * (13.91)

Note: * Values are means with standard errors in parentheses.

summarizes the mean values and standard errors of various factors for different forest origins. All factors showed significant differences between natural and plantation forests (p < 0.05). In terms of environmental conditions, natural forests were predominantly distributed at higher elevations (1392.91 m), characterized by lower MAT (9.30 °C), lower MAP (1025.64 mm), and relatively higher P/PET (1.46). In contrast, plantation forests were concentrated at lower elevations (554.60 m), with warmer climates (MAT = 13.83 °C) and higher precipitation (MAP = 1164.12 mm), but lower P/PET (1.32). Regarding carbon sequestration potential, the SIFmean (0.322 mW m⁻² nm⁻¹ sr⁻¹) and SIFmax (0.467 mW m⁻² nm⁻¹ sr⁻¹) of natural forests were significantly lower than those of plantation forests (SIFmean = 0.382 mW m⁻² nm⁻¹ sr⁻¹; SIFmax = 0.514 mW m⁻² nm⁻¹ sr⁻¹). For species diversity, the SI was significantly higher in natural forests (0.97) than in plantation forests (0.32), indicating greater species richness. The mean R/S did not differ significantly between forest origin (0.26), while canopy cover was significantly higher in natural forests (65.44%) than in plantation forests (63.23%).

Figure 3 illustrates the relationship between SI and SIF under different forest origins. In natural forests, the relationship between SI and SIF displays a clear nonlinear pattern, with a significant quadratic trend. Both SIFmean and SIFmax reach their peak values when SI is approximately 2.3. In contrast, the relationship between SI and SIF in plantation forests shows no significant curvature. SIFmean exhibits a slight decreasing trend as SI increases, while SIFmax increases gradually with rising SI.

3.3.2. Factors Influencing SIF Across Forest Origins

Figure 4 presents the factor importance rankings from the random forest models for SIF in natural and plantation forests. The corresponding importance indices and relative contributions are summarized in

Table 2. Importance Indices and Relative Contributions of SIF-Related Features in Natural and Plantation Forests.

	Natural Forests		Plantation Forest
	SIFmean	SIFmax	SIFmean	SIFmax
SI	0.336	0.482	0.047	0.040
	(22.7%)	(31.1%)	(4.2%)	(2.4%)
MAT (°C)	0.299	0.390	0.368	0.572
	(20.2%)	(25.2%)	(33.4%)	(34.3%)
MAP (mm)	0.402	0.188	0.392	0.594
	(27.2%)	(12.2%)	(35.6%)	(35.3%)
P/PET	0.080	0.094	0.127	0.247
	(5.4%)	(6.1%)	(11.5%)	(14.8%)
SM (%)	0.171	0.146	0.059	0.061
	(11.5%)	(9.4%)	(5.3%)	(3.7%)
R/S	0.085	0.108	0.037	0.089
	(5.7%)	(7.0%)	(3.3%)	(5.4%)
Litterfall (t ha−1 a−1)	0.048	0.075	0.035	0.033
	(3.2%)	(4.9%)	(3.1%)	(2.0%)
Canopy Cover (%)	0.059	0.066	0.038	0.029
	(4.0%)	(4.3%)	(3.4%)	(1.7%)

Note: Values are importance indices; numbers in parentheses indicate relative contribution.

. The models performed well in both forest types, with higher explanation observed in natural forests. Specifically, the R² values for SIFmean and SIFmax in natural forests were 0.902 and 0.887, respectively, slightly exceeding those in plantation forests (R² = 0.851 and 0.882). In natural forests, the most important predictors of SIFmean were MAP (importance = 0.402, contribution = 27.2%), SI (0.336, 22.7%), and MAT (0.299, 20.2%). Vegetation structural factors, including R/S, canopy cover and litterfall, jointly accounted for 12.9% of the total contribution. For SIFmax, SI was the dominate factor (0.482, 31.1%), followed by MAT (0.390, 25.2%) and MAP (0.188, 12.2%). The combined contribution of vegetation structural factors increased to 16.2%.

Changes in SIFmean for plantation forests were primarily influenced by climatic factors, with relatively minor contributions from species diversity and vegetation structure. Specifically, MAP (0.392, 5.6%), MAT (0.368, 33.4%), and P/PET (0.127, 11.5%) were the main influencing factors. The importance index for SI was 0.047, with a relative contribution rate of 4.2%, while the combined contribution rate of vegetation structure factors (root-crown ratio, canopy cover, litterfall volume) was 9.8%. For SIFmax, the primary influencing factors were MAP (0.594, 35.3%), MAT (0.572, 34.3%), and P/PET (0.247, 14.8%). Among secondary factors, SI contributed less, with an importance index of 0.040 and a relative contribution of 2.4%.

In plantation forests, changes in SIFmean were primarily influenced by climatic factors, with relatively minor contributions from species diversity and vegetation structure. Factors with higher influence were MAP (0.392, 35.6%), MAT (0.368, 33.4%), and P/PET (0.127, 11.5%). The SI had a low importance value (0.047) and a relative contribution of 4.2%. Vegetation structural factors, including R/S, canopy cover and litterfall, collectively accounted for 9.8% of the contribution. For SIFmax, SI was even less influential, with an importance index of 0.040 and a relative contribution of 2.4%.

The regulatory role of species diversity in vegetation carbon sequestration potential varies markedly between different forest origins. In natural forests, the SI is among the primary predictors of both SIFmean and SIFmax, together with MAT and MAP. These three factors collectively account for approximately 70% of the total contribution to SIF. Notably, SI has the highest importance index for SIFmax, indicating a particularly strong influence under optimal photosynthetic conditions. In contrast, the influence of SI is substantially reduced in plantation forests, contributing less than 5% to both the SIFmean and SIFmax models. In these forests, climatic factors become dominant, together explaining more than 80% of the variation in SIF. In both forest origins, vegetation structural factors play a secondary role, contributing between 9.1% and 15.3%.

Table 3. Importance Indices and Relative Contributions of SIF-Related Features in Pure versus Mixed Forests.

		Pure Forests			Mixed Forests
		All	Natural Forests	Plantation Forests	All	Natural Forests	Plantation Forests
SIFmean	SI	/	/	/	0.119 (8.20%)	0.223 (13.71%)	0.038 (3.73%)
	Climatic and environmental factors	1.204 (88.2%)	0.945 (88.24%)	0.977 (91.56%)	1.100 (75.60%)	1.137 (69.79%)	0.879 (85.38%)
	Vegetation structure factors	0.161 (11.80%)	0.126 (11.76%)	0.090 (8.43%)	0.237 (16.30%)	0.269 (16.50%)	0.112 (10.89%)
SIFmax	SI	/	/	/	0.117 (8.20%)	0.275 (17.83%)	0.021 (2.0%)
	Climatic and environmental factors	1.050 (80.50%)	1.204 (88.20%)	1.166 (90.88%)	1.100 (77.20%)	0.994 (64.37%)	0.927 (89.26%)
	Vegetation structure factors	0.255 (19.50%)	0.161 (11.80%)	0.117 (9.12%)	0.208 (14.60%)	0.275 (17.79%)	0.091 (8.76%)

Note: Values are importance indices; numbers in parentheses indicate relative contribution.

summarizes the importance indices and relative contributions of variables to SIF in the random forest models for monoculture and mixed-species stands. The results show that climatic and environmental factors accounted for over 80% of the contribution to both SIFmean and SIFmax in monocultures. Specifically, their contributions in natural monocultures were 88.24% and 88.20%, respectively, which were notably lower than those in planted monocultures (91.56% for SIFmean and 90.88% for SIFmax). In contrast, vegetation structural factors contributed relatively little, accounting for 11.80% and 19.5% across all monocultures. In natural monocultures, their contributions were 11.76% and 11.80%, respectively—higher than those in planted monocultures (8.43% for SIFmean and 9.12% for SIFmax).

In mixed-species stands, climatic and environmental factors remained the dominant predictors, though their relative contributions were lower than in monocultures. The Shannon index contributed 8.2% to both SIFmean and SIFmax in mixed stands. In natural mixed stands, its contribution to SIFmax (17.83%) was slightly higher than to SIFmean (13.71%), yet both values were substantially greater than those in planted mixed stands (3.73% for SIFmean and 2.0% for SIFmax). The contribution of vegetation structural factors was markedly higher in mixed stands than in monocultures, reaching 16.30% and 14.60% across all mixed stands. Within natural mixed stands, these values were 16.50% and 17.79%, compared with 10.89% and 8.76% in planted mixed stands.

The drivers of SIF shift systematically from predominantly climate-environmental controls in monocultures to a multi-factor synergy involving climate, vegetation structure, and species diversity in mixed-species stands. This transition is particularly evident in natural forests, offering a quantitative basis for elucidating functional differences across forest ecosystems with varying levels of species diversity.

3.3.3. Factors Influencing SIF by Successional Stage

Table 4. Statistical Measures of Relevant Characteristics Across Different Successional Stages.

	Early Stages	Middle Stages	Late Stages
SI	0.513 (0.020)	0.767 (0.026)	0.726 (0.024)
SIFmean (mW m−2 nm−1 sr−1)	0.371 (0.002)	0.343 (0.002)	0.338 (0.003)
SIFmax (mW m−2 nm−1 sr−1)	0.506(0.003)	0.486 (0.002)	0.474 (0.003)
MAT (°C)	12.259 (0.192)	10.700 (0.130)	11.272 (0.197)
MAP (mm)	1125.603 (12.855)	1051.502 (9.788)	1112.835 (14.034)
P/PET	1.400 (0.009)	1.386 (0.008)	1.433 (0.011)
SM (%)	27.156 (0.188)	26.428 (0.113)	26.995 (0.159)
R/S	0.258 (0.002)	0.253 (0.001)	0.245 (0.002)
Litterfall (t ha−1 a−1)	3.130 (0.064)	4.467 (0.047)	5.446 (0.078)
Canopy Cover (%)	61.360 (0.603)	64.974 (0.351)	66.104 (0.507)

Note: Values are means with standard errors in parentheses.

presents the statistical characteristics of various factors across different successions (early, middle, and late). As forest succession progresses, the SI gradually increases, while SIF values decline significantly. Climatic conditions also differ across stages, while forests in early stages are generally warmer and more humid, middle stages are cooler and drier, and late stages tend to be more temperate. These climatic patterns may not solely result from succession processes but are likely influenced by differences in plot distribution and anthropogenic disturbance. For example, natural forests in the late stages are often located in high-altitude regions with limited human impact. In addition, the importance of canopy cover and litterfall increases consistently along the forest succession, whereas the P/PET, SM, and R/S remain relatively stable across all stages.

Table 5. Importance Indices and Relative Contributions of SIF-Related Features Across Different Successional Stages.

		SIFmean			SIFmax
		Early Stages	Middle Stages	Late Stages	Early Stages	Middle Stages	Late Stages
SI	Importance Index	0.144	0.206	0.235	0.088	0.144	0.146
	Relative Contribution	9.35%	13.77%	18.29%	6.17%	8.61%	12.09%
Climatic and environmental factors	Importance Index	1.130	1.049	0.929	1.11	1.182	1.005
	Relative Contribution	73.54%	70.20%	72.25%	77.76%	70.55%	76.74%
Vegetation structure factors	Importance Index	0.263	0.239	0.122	0.229	0.349	0.146
	Relative Contribution	17.11%	16.03%	9.46%	16.07%	20.84%	11.17%

Note: SI in the table refers to the Shannon index; climatic and environmental factors include MAT, MAP, P/PET, and SM; vegetation structure factors include R/S, litter, and canopy cover. Note: Values are importance indices; numbers in parentheses indicate relative contribution.

presents the importance indices and relative contributions of different factors in the random forest models across forest succession. The models fit both SIFmean and SIFmax well at all stages. Except in early stages, where the R² values for SIFmean and SIFmax are 0.58 and 0.56, respectively, all other models achieved R² values exceeding 0.60. As succession progresses, the importance and relative contribution of the SI increase consistently. In early stages, the importance indices of SI for SIFmean and SIFmax are 0.144 and 0.088, with relative contributions of 9.35% and 6.17%, respectively. At this stage, SIF is primarily driven by climatic and environmental factors, which contribute approximately 75%, while vegetation structural factors account for 16.5%. In middle stages, the importance of SI increases with indices of 0.206 (SIFmean) and 0.144 (SIFmax), and contributions of 13.77% and 8.61%, respectively. The contribution of climatic and environmental factors decreases slightly to around 70%, while the contribution from vegetation structural factors remains stable at approximately 17%. In late stages, the importance of SI further increases with importance indices of 0.235 (SIFmean) and 0.146 (SIFmax), corresponding to relative contributions of 18.29% and 12.09%. Meanwhile, the contribution of climatic and environmental factors rises slightly to about 74%, while vegetation structure factors decline to 10%.

3.3.4. The Combined Effects of Forest Origin and Successional Stage on SIF

To further clarify the combined effects of forest origin and succession on SIF,

Table 6. Importance Indices and Relative Contributions of SIF-Related Features by Forest Origin and Successional Stage.

		Natural Forests			Plantation Forests
		Early Stages	Middle Stages	Late Stages	Early Stages	Middle Stages	Late Stages
SIFmean	SI	0.272 (22.36%)	0.267 (21.55%)	0.388 (33.20%)	0.016 (1.35%)	0.063 (5.21%)	0.005 (0.98%)
	Climatic and environmental factors	0.568 (46.68%)	0.725 (58.57%)	0.673 (57.49%)	1.072 (91.32%)	1.026 (85.30%)	0.452 (81.74%)
	Vegetation structure factors	0.377 (30.96%)	0.246 (19.88%)	0.109 (9.31%)	0.086 (7.34%)	0.114 (9.49%)	0.095 (17.28%)
SIFmax	SI	0.210 (12.96%)	0.314 (22.92%)	0.322 (35.61%)	0.032 (4.62%)	0.045 (3.25%)	0.010 (0.98%)
	Climatic and environmental factors	0.971 (60.02%)	0.762 (54.57%)	0.470 (52.02%)	0.546 (79.05%)	1.244 (90.40%)	0.878 (89.24%)
	Vegetation structure factors	0.437 (27.02%)	0.320 (22.92%)	0.112 (12.38%)	0.113 (16.33%)	0.087 (9.49%)	0.096 (9.78%)

Note: Values in the table represent importance indices, with relative contributions indicated in parentheses; SI denotes Shannon index; climatic and environmental factors include MAT, MAP, P/PET, and SM; vegetation structure factors include R/S, litter, and canopy cover.

presents the importance indices and relative contributions of various factors for SIFmean and SIFmax. The results reveal that the importance of the SI varies markedly between forest origins and across succession. In natural forests, SI was a key predictor of both SIFmean and SIFmax, with its influence increasing steadily along the forest succession. The relative contribution of SI to SIFmean rose from 22.36% in early stages to 33.20% in late stages. For SIFmax, the trend was even stronger, with SI contribution increasing from 12.96% to 35.61%. In contrast, SI had a consistently minor role in plantation forests, with relative contributions below 6% for both SIFmean and SIFmax across all stages. Although a slight increase was observed in middle stages, the overall influence of SI remained substantially lower than in natural forests. Beyond SI, the contribution of climatic and environmental factors increased progressively with succession in natural forests but declined in plantation forests. Vegetation structural factors generally contributed more in natural forests than in plantation forests and exhibited a decreasing trend with succession.

4. Discussion

4.1. Species Diversity and Forest Productivity Indicators

Forest productivity is a continuum from instantaneous photosynthesis to long-term carbon sequestration, requiring comprehensive assessment through multi- spatiotemporal observation indicators. Commonly used metrics mainly include SIF, GPP, Net Primary Production (NPP), and biomass. SIF serves as a direct optical probe for photosynthetic photochemical activity, reflecting the overall carbon sequestration potential of vegetation. GPP represents the total carbon fixed by vegetation through photosynthesis per unit time, serving as the primary energy input source for ecosystems. NPP is calculated as GPP minus autotrophic respiration, representing the carbon available for vegetation growth. Biomass represents the long-term net accumulation of organic carbon derived from NPP, after accounting for losses due to litterfall, mortality, and herbivory, and thus directly reflects the size of the ecosystem’s carbon pool.

Former studies indicate that various carbon sequestration indicators were different in sensitivity and response intensity under the influence of biodiversity. Using GPP as a proxy for forest productivity, Zhang et al. [60] found that tree species richness was a significant positive predictor of GPP in tropical forests, with its direct effect strength being second only to stand structure. Fischer et al. [61] simulated tropical forest succession under varying numbers of plant functional types. They found that GPP was the most sensitive indicator to diversity changes, which was followed by a weaker response in NPP, whereas aboveground biomass showed almost no response once it reached long-term stability. The variation in sensitivity arises because the high photosynthetic capacity of pioneer species is a direct driver of GPP, but this signal is attenuated by respiratory costs and carbon turnover as it propagates to NPP and biomass. Therefore, ignoring diversity would result in a severe underestimation of GPP. Craven et al. [62] underscored the central role of NPP n the diversity–productivity relationship and highlighted the strong scale dependency of the association between species richness and NPP. This relationship becomes significant only at larger spatial scales. Meanwhile, biomass represents the accumulated expression of NPP and also indirectly regulates species richness by modifying competitive interactions and light availability. Recent research has established a significant positive correlation between tree species richness and ecosystem photosynthetic rates [54]. This suggests that biodiversity can directly enhance ecosystem productivity via mechanisms including interspecific facilitation. Together, these results demonstrate that different carbon sequestration metrics differ in their sensitivity and response intensity to diversity regulation. To explore how species diversity regulates the acquisition and utilization efficiency of light energy from the source of ecosystem photosynthesis, this study used SIF as a direct indicator of photochemical activity. The results show that species diversity significantly promotes the carbon sequestration potential of vegetation in forest ecosystems, especially in natural forests.

4.2. The Promoting Effect of Species Diversity on Vegetation Carbon Sequestration Potential

This study demonstrates that species diversity is a key driver of forest vegetation’s carbon sequestration potential. The data show a significant positive correlation between the Shannon index and canopy closure (r = 0.25). Similarly, a significant positive correlation is observed between canopy closure and SIF (r = 0.09) (Figure 2). A key finding is that greater species diversity directly promotes the development of more complex and denser canopy structures. We further compared structural diversity and carbon sequestration potential across varying levels of species diversity (

Table A1. Importance Index and Relative Contribution of SIF-Related Characteristics in Pure and Mixed Forests.

	Pure Forests		Mixed Forests
	SIFmean	SIFmax	SIFmean	SIFmax
SI	/	/	0.119 (8.2%)	0.117 (8.2%)
Climatic and environmental factors	1.204 (88.2%)	1.050 (80.5%)	1.100 (75.6%)	1.100 (77.2%)
Vegetation structure factors	0.161 (11.8)	0.255 (19.5%)	0.237 (16.3%)	0.208 (14.6%)

). The results show that mixed-species stands exhibit significantly higher canopy closure (67.24%) than monocultures (58.31%), along with a greater mean carbon sequestration potential. This indicates that richer species composition promotes the development of more complex canopy structures. A further comparison between the top 10% of samples by Shannon index and the broader set of mixed-species stands revealed that the former group exhibited higher canopy closure (75.40%) and a greater mean carbon sequestration potential than the average across all mixed-species stands. This demonstrates that elevated species diversity substantially promotes both canopy structural complexity and carbon sequestration potential. This diversity-driven structural optimization operates mainly by augmenting total canopy leaf area and spatial heterogeneity, which lowers overall canopy albedo and consequently improves forest light harvesting capacity for photosynthetically active radiation. Ultimately, the increased light-use efficiency translates into greater photosynthetic potential at the ecosystem scale, which is directly observed in this study as enhanced SIF.

The findings of this study support theoretical mechanisms such as the selection effect and complementarity effect by which species diversity enhances carbon sequestration potential and are consistent with related research demonstrating that optimized canopy structure improves light resource utilization. Xu et al. [63] introduced a canopy complementarity index and showed that structural optimization of the canopy directly enhances light-use efficiency via mechanisms such as vertical stratification, crown differentiation, and complementarity of functional traits. As for the inherent ecological functions of structural diversity, Cheng et al. [64] showed that it constitutes the physical basis for spatial niche differentiation and efficient light-use efficiency. The findings are consistent with the core mechanism revealed by Cao et al. [54] at a global scale: species richness enhances ecosystem photosynthesis by increasing community structural complexity. This study offers evidence supporting this mechanism at the national scale. In our random forest model, the Shannon index shows a substantial relative contribution (approximately one-fourth) to SIF in natural forests, which aligns with the finding reported by Cao et al. [54] that species richness is the second-most important factor (about 17%) explaining spatial variation in photosynthesis. Collectively, these studies underscore diversity’s pivotal role in driving light capture processes.

4.3. The Moderating Effects of Forest Type and Successional Stage

This study demonstrates that the effect of species diversity on forest carbon sequestration potential is strongly modulated by forest type and successional stage. The relative contribution of species diversity is markedly higher in natural forests than in plantations (Table 2), with an approximate 20% greater contribution compared to planted forests. This can be attributed to the stringent filtering of species composition by natural conditions like climate and topography in natural forests, which favors the dominance of complementary effects and thereby heightens the critical role of species diversity. The plantation plots analyzed in this study are primarily situated in areas that were reforested following anthropogenic disturbance in the mid-to-late 20th century. These sites generally benefit from favorable hydrothermal conditions, and the reforestation efforts predominantly relied on single or limited numbers of fast-growing, high-yield species [55]. Consequently, productivity in these plantations is therefore governed primarily by species growth traits rather than richness, where competitive exclusion tends to diminish positive diversity effects. Studies on forest origin have shown that plantations exhibit weaker diversity effects, which aligns with their typically uniform management regimes. As noted by Onyekwelu et al. [65], tropical plantations typically feature monocultures of fast growing species, with productivity driven primarily by species selection, genetic improvement, site management, fertilization, and stand density control rather than by species diversity. Based on Chinese forest plot data from 2007 to 2013, Gao et al. [66] systematically compared plantations and natural forests across metrics including biomass, productivity, and diversity. Their results demonstrated a significant positive and mechanistically complex relationship between productivity and species richness in natural forests, a pattern that was absent or extremely weak in plantations, which aligns with the findings of the current study. This reflects a fundamental divergence in the mechanisms underlying natural and plantation forests.

In natural forests, the importance of both diversity and productivity increases with successional stage (Table 5), with their relative contribution rising from about 18% in early stages to over 30% in later stages. This likely occurs because the relatively simple community structure in early-successional natural forests is primarily driven by a few pioneer species, resulting in low diversity that contributes only marginally to overall productivity. As illustrated in Appendix A Figure A1, even early successional plots with low species diversity indices (SI) can maintain high maximum photosynthetic potential (SIFmax). This pattern is largely due to dominant species that drive carbon sequestration through rapid growth and high photosynthetic capacity, reinforcing the characteristic dominance of a few species over early-stage productivity. The relationship between SI and SIFmax is notably scattered during this stage, indicating the unstable nature of diversity effects. As succession proceeds, the system develops complex vertical structures and multiple ecological niches. This structural complexity enables more efficient resource use through spatiotemporal complementarity in light capture and organism-generated environmental heterogeneity, thereby significantly elevating the importance of diversity in mid to late successional stages. During mid to late succession (Figure A1), the relationship between SI and SIFmax shows a stronger and more stable positive correlation, reflecting that as structural complexity and functional complementarity increase, diversity’s enhancement of carbon sequestration potential stabilizes. This dynamic pattern aligns with broader ecological observations. For example, long-term monitoring in Costa Rican tropical rainforests by Lasky et al. [67] revealed that the positive relationship between biodiversity and changes in aboveground biomass shifts significantly as forests age, mirroring the successional trend identified in our study. Cheng et al. [64] identified the key mechanism behind this pattern, showing that stand age is the strongest predictor of structural diversity, as forest communities gradually self-organize—through natural growth, competition, and regeneration—into increasingly complex vertical structures. This temporal development of structural complexity establishes the physical basis for more efficient complementary use of light and other resources among species.

4.4. Limitations and Future Directions

This study applied SIF to examine how species diversity affects vegetation carbon sequestration potential, thereby addressing a key limitation of conventional indicators, which often fail to capture dynamic changes in plant physiological status. SIF is highly sensitive to photosynthetic processes, allowing for a more precise representation of ecosystem photosynthetic activity and carbon sink strength [27,28]. By incorporating standardized plot-survey data, this study provides macroscale evidence for the coupling between species diversity and vegetation carbon sequestration potential, while demonstrating the effective use of SIF in this field of research. Nevertheless, this study has limitations, including the omission of functional traits such as specific leaf area, which likely constrained a more detailed analysis of selection effects. Future studies should combine SIF with functional diversity indices and long-term monitoring data to better reveal the underlying mechanisms and spatiotemporal dynamics of how species diversity functions.

5. Conclusions

Based on nationwide plot surveys and SIF data, this study elucidates the pivotal role of species diversity in regulating the carbon sequestration potential of forest ecosystems and delineates how this role varies with forest origin and successional stage. The results demonstrate that the positive effect of species diversity on carbon sequestration potential is substantially stronger in natural forests than in plantations. This effect intensifies throughout succession, with the most significant enhancement in maximum photosynthetic capacity observed during late-successional stages. Furthermore, this research confirms the efficacy of SIF in characterizing biodiversity-ecosystem functioning biodiversity and ecosystem functioning experiment relationships at a macro scale and underscores its unique utility in quantifying these associations.

These findings highlight the critical importance of conserving natural forests and implementing near-natural management practices in planted forests to enhance carbon sink functionality. Consequently, protecting natural forests and promoting close-to-nature silviculture are not only effective strategies for augmenting forest carbon sequestration but are also vital for bolstering ecosystem resilience and ensuring sustainable forest management. Collectively, this work provides a robust scientific foundation for informing sustainable forest management and climate adaptation policies in an era of global change.