Factors influencing natural regeneration of Fagus hayatae

Study site

The Sichuan Micangshan Nature Reserve (32°29′–32°41′N, 106°24′–106°39′E) occupies the northern margin of the Sichuan Basin. This protected area features a subtropical humid monsoon climate with elevational ranges spanning 1,500–2,000 m. Key climatic parameters include a 260-day frost-free period, mean annual temperature of 13 °C, annual sunshine duration of 1,355.3 h, and mean annual precipitation of 1,350 mm. The vegetation is dominated by subtropical evergreen-deciduous broadleaved mixed forests, with characteristic F. hayatae communities primarily occurring in deciduous broadleaved forests at 1,500–1,900 m elevation (Li et al., 2016a; Li et al., 2016b).

Experimental design

In July 2022, three vertical transects were established along an elevational gradient within the Micangshan Nature Reserve in Sichuan. The vertical spacing between adjacent elevation bands was 100–130 m. The elevation bands were: low elevation band: 1,670–1,700 m; mid elevation band: 1,770–1,800 m; high elevation band: 1,890–1,940 m. The low-elevation site (1,670–1,700 m) was established to ensure coverage of the upper elevational limit of Fagus hayatae within the study area while comprehensively assessing regeneration constraints across the species’ entire elevational gradient. Within each elevation band, three main plots (totaling nine plots; Table 1; Fig. 1), dominated by F. hayatae, were established, with a minimum distance of ≥1,000 m between adjacent main plots. Within each main plot, five 20 m × 20 m subplots were arranged in an X-shaped distribution pattern, with a spacing of >50 m between subplots. All F. hayatae individuals taller than two m were comprehensively surveyed within each plot; their diameter at breast height (DBH), tree height, and crown width were measured. Nested within each subplot, five 5 m × 5 m quadrats were established (maintaining the X-shaped distribution pattern) to monitor seedlings, saplings, small trees, and shrubs. Furthermore, nested within each of these 5 m × 5 m quadrats, five 1 m × 1 m herbaceous quadrats were placed, also maintaining the X-shaped distribution pattern.

To minimize precipitation effects on soil moisture, sampling was conducted ≥72 h post-rainfall. Within each nested quadrat, we removed the litter layer and collected 200 g soil samples (0–15 cm depth) following an X-shaped sampling pattern using a stainless-steel auger. We then homogenized soils from five quadrats to create composite samples were immediately stored in insulated containers and transported to the laboratory. Post-processing involved: Debris removal through manual sorting sieving (<1 mm mesh). Division into two aliquots: air-dried samples for physicochemical analyses (soil moisture, pH); Fresh samples preserved at 4 °C for enzymatic assays (e.g., urease activity).

The environmental factors investigated in this study were categorized into four groups: geographic factors: slope (S) and elevation (A); stand factors: canopy density (CD), litter thickness (LT), shrub cover (SC), herbaceous cover (HC), number of adult trees (ANP), diameter at breast height (DBH); Climatic factors: light intensity (LI), air temperature (AT), soil temperature (ST), air humidity (AH), and soil moisture (SM); soil factors: pH, available phosphorus (AP), cation exchange capacity (CEC), total nitrogen (TN), soil organic content (SOM), soil water content (SWC), available potassium (AK), urease activity (UA), and catalase activity (CAT).

Indicator measurement methods

Diameter at breast height (DBH) was measured at 1.20 m above ground level using a diameter tape. Geospatial parameters including elevation, longitude, latitude, and slope were recorded using a TX35-S300 GPS receiver (China) with satellite navigation timing (Liang & Wei, 2021). Litter layer thickness was measured following the vertical insertion method of a metal ruler to the soil surface (Liang & Wei, 2020). Canopy density was estimated through hemispherical photography analysis using the Canopy Capture application.

Vegetation coverage metrics were determined through systematic grid sampling: Shrub cover (SC) and herbaceous cover (HC) were quantified using a 10 × 10 grid system, calculated as the percentage of grid cells occupied by vegetation. Light intensity was measured using a Hipoint HR-350 sensor. Soil temperature (ST) and soil moisture (SM) were recorded between 10:00 and 16:00 local time using a TES-1365 thermohygrometer. Ambient air temperature (AT) and air humidity (AH) were measured with a Sunrise digital thermohygrometer.

Soil analyses followed standardized protocols: pH measurement using calibrated pH meter. Moisture content determination through constant-temperature drying (105 °C). Total nitrogen (TN) quantification via Kjeldahl digestion. Soil organic matter (SOM) analysis using dichromate oxidation. Cation exchange capacity (CEC) determination by sodium acetate method. Available phosphorus (AP) extraction via Olsen’s sodium bicarbonate method (Yan et al., 2015). Available potassium (AK) measurement using ammonium acetate extraction. Enzyme activities (urease and catalase) analyzed with microplate kits (Suzhou Keming Biotechnology).

Data analysis

Since natural regeneration data (seedling density, SED; seedling height, SEH; sapling density, SAD; sapling height, SAH; small tree density, STD; and small tree height, STH) could not be standardized using conventional methods, nonparametric Kruskal–Wallis ANOVA was employed to analyze treatment effects, with raw data ranked to examine interactions among multiple factors (Zhao, Sun & Gao, 2023). These regeneration metrics were designated as response variables, while environmental factors were divided into four categories to represent the explanatory variable: geographic factors: slope (S), elevation (A); stand factors: canopy density (CD), litter thickness (LT), shrub cover (SC), herbaceous cover (HC), number of adult trees (ANP), and diameter at breast height (DBH); climatic factors: light intensity (LI), air temperature (AT), soil temperature (ST), air humidity (AH), and soil moisture (SM).

Soil factors: pH, available phosphorus (AP), cation exchange capacity (CEC), total nitrogen (TN), soil organic content (SOM), soil water content (SWC), available potassium (AK), urease activity (UA), and catalase activity (CAT). The relationships between environmental factors and F. hayatae regeneration were analyzed using CANOCO 5.0. Preliminary detrended correspondence analysis (DCA) of Fagus regeneration data revealed gradient lengths <3.0, indicating linear responses of this species to environmental variation, thereby justifying linear multivariate methods (Li et al., 2018; Zhao, Sun & Gao, 2023; Liang & Wei, 2020). Based on this, species-environment redundancy analysis (RDA) was implemented to identify dominant drivers impacting Fagus regeneration. The ordination focused on environment-regeneration correlations, with eigenvalue significance evaluated via Monte Carlo permutation tests (99% confidence) (Chen & Cao, 2014; Chen, 2003). Regeneration data were log-transformed to mitigate extreme value effects (Gazer, 2011), while environmental variables were square-root transformed prior to analysis to ensure variance homogeneity. Transformed datasets were analyzed using ANOVA, with Pearson correlation coefficients calculated to quantify associations between environmental factors and regeneration. Statistical analyses were performed in SPSS 22.0 (Chicago, IL, USA).

Structural equation modeling (SEM) is a robust multivariate technique for examining hypothesized causal relationships among latent variables through analysis of direct and indirect associations (Zhao et al., 2019). This method simulates multivariate interactions using two or more structural equations (Gazer, 2011) and visualizes complex variable relationships via intuitive network diagrams. SEM enables quantification of partial contributions from correlated variables, distinction between direct and indirect effects, identification of multiple influence pathways, and estimation of pathway strength comparisons (Lv et al., 2023; Eisenhauer et al., 2015).

In this study, we implemented SEM using Amos 26 (IBM SPSS, USA) to assess the effects of diverse environmental factors on six regeneration parameters: seedling density (SED), seedling height (SEH), sapling density (SAD), sapling height (SAH), small tree density (STD), and small tree height (STH). Path coefficients between factors were systematically analyzed. The initial model underwent iterative refinement until achieving validation criteria: chi-square to degrees of freedom ratio (χ²/DF < 3), goodness-of-fit index (GFI > 0.90), comparative fit index (CFI > 0.90), root mean square error of approximation (RMSEA < 0.08), and non-significant P-values (P > 0.05) (Zhu et al., 2022; Trivedi et al., 2016). Final SEM diagrams were generated using Visio 2021 (Microsoft Corp., USA).

Distribution of Fagus hayatae at different elevations

We assessed the regeneration patterns of F. hayatae across ontogenetic stages (seedlings: mean height 0.2 m; saplings: 1.14 m; small trees: 2.96 m) along an elevation ranges. The census recorded 685 seedlings, 125 saplings, and 72 small trees, showing distinct elevational zonation.

Low-elevation zone (≤1,200 m): 132 seedlings, 34 saplings, 42 small trees. Mid-elevation zone (1,201–1,800 m): 480 seedlings, 62 saplings, 10 small trees. High-elevation zone (≥1,801 m): 73 seedlings, 29 saplings, 20 small trees.

Regeneration correlation with environmental factors

The Pearson correlation analysis revealed significant associations between environmental factors and F. hayatae developmental stages (Fig. 2). Seedling density demonstrated strong negative correlations with litter thickness (−0.498, P < 0.01) and soil pH (−0.556, P < 0.01). Sapling density showed a significant negative correlation with soil moisture content (−0.376, P < 0.05). Small tree height exhibited positive correlations with available potassium (0.300, P < 0.05) and litter thickness (0.411, P < 0.01). Notably, soil water content was inversely associated with pH (r−0.369, P < 0.01).

Environmental factors for RDA analysis and SEM

Redundancy analysis (RDA) was conducted to quantify the relative contributions of environmental factors to regeneration dynamics and evaluate variable associations within multidimensional data structures. This analytical approach enabled identification of optimal vegetation regeneration predictors through rigorous statistical frameworks. In the ordination diagram, the cosine values between environmental vectors indicate pairwise correlations, with this angular metric effectively projecting multivariate relationships onto a reduced-dimensional space. The geometric projection graphically represents the covariance structure between environmental variables and regeneration parameters. Cosine values approximating +1 reflect strong positive alignment with ordination axes, values approaching −1 indicate pronounced negative associations, and those near 0 demonstrate statistically insignificant relationships.

The RDA revealed that habitat factors collectively accounted for 68.1% of the variance across the first two ordination axes for F. hayatae density parameters (saplings, small trees, and seedlings; Fig. 3A). Regarding height parameters, the first two axes explained 47.2% of the cumulative variance (Fig. 3B). While inter-factor correlations showed concordance with Fig. 2 findings, the diffuse variable associations in RDA constrained precise identification of dominant drivers and their proportional impacts on density and height variations. To address this, we conducted Monte Carlo permutation tests to quantify environmental factor significance, with results integrated into Fig. 3’s tabular summary. The contribution ranking of habitat factors to density variation emerged as: pH > cation exchange capacity > soil temperature. Key determinants of height development followed the hierarchy: litter depth > air humidity > soil water content.

Based on correlation analysis results, structural equation models (SEMs) were constructed by selecting environmental factors with high correlation coefficients and those demonstrating contribution rates exceeding 10% through Monte Carlo permutation tests. Environmental factors showing non-significant effects on regeneration were systematically eliminated according to established model fit criteria. Through iterative refinements via multiple modifications, seven environmental factors exhibiting the strongest associations with regeneration were ultimately retained. Subsequently, three stratified SEMs (low-, mid-, and high-elevation levels) were developed using elevation gradient as a covariate.

Low-elevation SEM (Fig. 4): litter thickness demonstrated the strongest negative direct effect on seedling density (−0.80), followed by available potassium’s influence on sapling density (−0.46) and diameter at breast height (DBH) on small tree height (−0.63). While pH showed no direct effects on seedling density, it exhibited indirect positive effects through covarying factors (total effect: +0.32). These results establish litter thickness, available potassium, and DBH as pivotal drivers of F. hayatae regeneration in low-elevation habitats.

Mid-elevation SEM (Fig. 5): DBH displayed positive direct effects on sapling density (+0.46), small tree height (+0.57), and small tree density (+0.79). Contrastingly, negative direct effects emerged for pH on seedling density (−0.50), air temperature on small tree density (−0.41), and soil moisture on sapling density (−0.49). Available potassium manifested complex mediation pathways: while lacking direct effects on small tree density, it exerted substantial indirect negative influence through DBH mediation (total effect: −0.64), alongside indirect positive effects on seedling abundance (SAD) and small tree height (STH).

High-elevation SEM (Fig. 5): Available potassium emerged as the strongest positive predictor for sapling density (+0.70), followed by air temperature (+0.58) and litter thickness (+0.53) impacts on STH. A notable indirect relationship was observed where pH influenced small tree height exclusively through indirect pathways (total effect: +0.26), underscoring the complex interaction networks governing high-elevation regeneration dynamics.

Factors affecting regeneration at low elevations

Excessively thick litter layers may inhibit seedling establishment through mechanisms such as light interception (Wang & Zhang, 2008), prevention of seed-soil contact (Willis et al., 2021; Koorem, Price & Moora, 2011), and release of autotoxic substances (Willis et al., 2021; Pardos et al., 2007). In this study, the litter layer exhibited the strongest direct negative effect on seedling density (−0.80) (Fig. 4), which aligns with previous findings on the mechanical barrier effect of litter accumulation (Eckstein & Donath, 2005; Yu et al., 2017). The negative effect of available potassium on sapling density (−0.46) (Fig. 4) may stem from a mismatch between soil nutrient supply and regenerating individuals’ demands. Specifically, when potassium released through litter decomposition fails to meet sapling growth requirements, soil available potassium shows a negative correlation with regeneration (Li et al., 2021; Zhang et al., 2011). In the low-elevation SEM (Fig. 4), although pH exhibited no direct effect on seedling density, it showed an indirect positive effect via the litter layer (0.42). Shen & Zhao (2015) demonstrated that soil microorganisms not only enhance nutrient uptake efficiency but also influence plants through nutrient mobilization and transfer. Therefore, we hypothesize that this phenomenon may relate to soil microbial activity, as elevated pH could stimulate proliferation of specific microbial taxa, accelerating litter decomposition. This process would reduce litter thickness and mitigate mechanical shading on seedlings (Eckstein & Donath, 2005; Wang & Zhang, 2008), thereby indirectly increasing seedling density. Similarly, the indirect negative effect of available potassium on seedling density via the litter layer (−0.45) (Fig. 4) may arise from antagonistic mechanisms: on the one hand, high available potassium may enhance litter input by promoting mature tree growth (Zhang et al., 2011), intensifying its physical obstruction on seedlings (Koorem, Price & Moora, 2011); on the other hand, potassium enrichment might inhibit microbial-driven litter decomposition (Li et al., 2021), exacerbating the litter layer’s adverse effects. The SEM at low elevations revealed that air temperature had no direct effect on sapling density but demonstrated an indirect negative effect via tree diameter at breast height (−0.42), highlighting a cascading interaction between climatic factors and canopy structure. A plausible pathway is that relatively higher air temperatures enhance mature tree growth, leading to increased trunk diameter and canopy closure (Liu et al., 2020), which subsequently reduces understory light availability (Balandier et al., 2007). Given that saplings require significantly more light resources than seedlings (Wang et al., 2017; Zhao, Sun & Gao, 2023), canopy shading may suppress sapling survival and establishment by limiting photosynthetic product accumulation, ultimately manifesting as an indirect negative effect of temperature elevation on sapling density. This finding parallels observations reported by (Kovács, Tinya & Ódor, 2017).

Factors affecting regeneration at mid elevations

As shown in the mid-elevation SEM: the negative effects at mid-elevations primarily manifested as pH constraints on seedling density (−0.50) and soil moisture limitations on sapling density (−0.49). This aligns with (Dyderski et al., 2018), who reported reduced seedling survival under higher pH conditions, while we hypothesize that the inhibitory effect of soil moisture on saplings may relate to root hypoxia. Studies indicate that Fagaceae species require high light availability (Alfaro Reyna, Martínez-Vilalta & Retana, 2019). In our study, the positive effects of diameter at breast height (DBH) on sapling density (0.46), small tree density (0.79), and small tree height (0.57) likely originate from canopy openness regulating light conditions (Liu et al., 2020). Available potassium showed no direct effect on sapling density but exerted an indirect negative effect (−0.53) through two pathways: a negative effect via DBH (−0.66) and a positive effect via soil moisture (0.46). Similarly, available potassium had no direct effect on small tree density but generated an indirect negative effect (−0.64) through the same pathways: DBH (−0.66) and soil moisture (0.46). Li et al. (2005) demonstrated that mother trees with larger DBH typically acquire more soil water, light, and other resources to support seedling growth, thereby maintaining dominance in forests. As available potassium negatively affects mature trees, this indirectly reduces their support for saplings. Conversely, its positive effect on soil moisture may lead to hypoxia-induced root rot under elevated moisture levels. We hypothesize that available potassium suppresses the growth of saplings and small trees through these interactive pathways, ultimately decreasing their densities.

Factors affecting regeneration at high elevations

The strong positive effect of available potassium on sapling density (0.70) at high elevations (Fig. 6) corroborates the classical theory of nitrogen, phosphorus, and potassium as limiting elements for forest regeneration (Harpole et al., 2011; Wang et al., 2022). The positive effect of the litter layer on small tree height (0.53) (Fig. 6) may stem from its water and nutrient retention functions (Wang et al., 2022). The facilitative effect of air temperature (0.58) (Fig. 6) underscores the critical role of thermal conditions in high-elevation regeneration (Zadworny et al., 2021). The indirect positive influence of pH on small tree height (0.26) (Fig. 6) further emphasizes the pivotal role of microbially mediated nutrient cycling in later growth stages (Shen & Zhao, 2015).

Conservation recommendations for Fagus hayatae

This study elucidates the complex interactions between altitudinal environmental factors/drivers and natural regeneration dynamics of F. hayatae, revealing stage-specific and elevation-specific regulatory mechanisms. Our structural equation modeling (SEM) results highlight how key factors exert differential effects on regeneration outcomes across developmental stages and altitudinal gradients, providing critical insights for conservation strategies. To address key limiting factors for F. hayatae regeneration across altitudinal ranges, differentiated management strategies are required: low elevation (1,670–1,700 m): manually thin excessively thick litter layers to alleviate light interception and physical barriers, coupled with soil pH adjustment to enhance microbial decomposition. This accelerates litter degradation and mitigates inhibitory effects of available potassium accumulation on saplings. Additionally, reduce canopy closure to minimize light competition pressure on saplings. Mid-elevation (1,770–1,800 m): regulate canopy light transmittance to balance soil moisture and prevent root hypoxia. Adjust soil pH to weakly acidic ranges to improve seedling survival rates, while suppressing cascading negative effects of available potassium on saplings and small trees mediated through mature tree competition and soil moisture imbalance. High elevation (1,890–1,940 m): retain natural litter layers to strengthen water and nutrient retention capacities. Implement microclimate interventions to mitigate low-temperature limitations, thereby enhancing thermal resource utilization for small tree growth.