Stand Structure and Successional Pathway in an Artificial Hybrid Pine (Pinus × rigitaeda) Plantation from a Temperate Monsoon Region

Abstract

Artificial hybrid pine (Pinus × rigitaeda) plantations, widely established in Northeast Asia for reforestation and timber production, have reached maturity, necessitating an evaluation of their ecological sustainability and successional dynamics. Although numerous studies have examined succession in pure Pinus rigida or Pinus densiflora stands, the long-term structural transition and regeneration potential of hybrid P. × rigitaeda plantations remain poorly understood. This study quantitatively assessed the successional stage and potential transition pathways of P. × rigitaeda stands using an integrated analytical framework combining vegetation classification (TWINSPAN), ordination (NMDS), successional index, survival analysis (Weibull model), and growth–environment modeling (GAM). Multi-layer vegetation data were analyzed to evaluate compositional changes, structural attributes, and nonlinear environmental responses. The results revealed that the dominance of P. × rigitaeda declined markedly while native deciduous species increased in lower strata. The Weibull survival model (k = 1.3) indicated accelerating mortality with stand aging, and the successional index showed the highest value (0.4) for Castanea crenata, followed by other Quercus species, confirming an ongoing shift toward hardwood dominance. GAM analysis confirmed that growth stability was influenced by stand age and precipitation. These findings demonstrate that P. × rigitaeda plantations are not merely artificial production forests but function as self-organizing systems facilitating natural forest recovery. In this respect, the hybrid pine plantation can be interpreted as a spontaneous ecological experiment, highlighting the restoration value of artificial hybrids as transitional stages bridging artificial afforestation and natural forest succession in temperate monsoon regions.

1. Introduction

Artificial conifer plantations played a pivotal role in reforestation and land rehabilitation across East Asia during the latter half of the twentieth century. In Korea, one of the most distinctive outcomes of this large-scale restoration effort was the development of the hybrid pine Pinus × rigitaeda Hyun & Ahn, artificially bred in the early 1960s by Dr. S.N. Hyun and Dr. M.S. Ahn [1]. This hybrid combined Pinus rigida Mill., characterized by strong drought and fire resistance, with Pinus taeda L., known for rapid growth and a straight stem form, to produce a conifer capable of thriving on nutrient-poor sandy soils and degraded mountain landscapes. As a result, P. × rigitaeda became a productive and ecologically adaptable species well suited to Korean climatic and edaphic conditions, contributing substantially to national reforestation programs.

During the 1970s and 1980s, P. × rigitaeda became a cornerstone species in Korea’s large-scale reforestation and erosion-control programs. The species exhibited high survival and fast growth even on dry and degraded slopes, contributing decisively to the success of national afforestation projects [2]. These plantations played essential roles in initial vegetation cover, soil stabilization, and early-stage ecosystem recovery. Over time, however, natural regeneration and gradual invasion of native broadleaf species have been observed, suggesting that P. × rigitaeda stands may function as transitional communities within long-term successional trajectories toward mixed or deciduous forests. Despite its ecological and historical importance, academic and administrative attention to P. × rigitaeda has waned in recent decades as restoration policy increasingly prioritizes native species. This shift reflects growing recognition that single-species or hybrid plantations may face limitations in long-term ecological stability and biodiversity potential [3]. Nonetheless, P. × rigitaeda plantations continue to provide crucial ecological functions—maintaining soil cover, stabilizing slopes, and facilitating successional processes—making them valuable experimental models for afforestation and physiological ecology in marginal environments [4].

From a global perspective, numerous studies have explored structural and successional dynamics in aging conifer plantations. Aging stands commonly undergo self-thinning, canopy densification, and declines in radial growth, processes that alter light regimes and initiate understory diversification [5,6]. These transformations often culminate in post-plantation succession—an ecologically spontaneous shift from single-species conifer dominance to more diverse, broadleaf-dominated communities [7,8]. The extent and speed of these transitions depend on canopy closure and resource competition, which modulate light availability and regeneration niches in the understory [9]. While such processes are well documented in tropical and temperate plantations, hybrid pine systems remain underrepresented in global succession studies. In East Asia, research on P. × rigitaeda has focused primarily on silvicultural performance and site adaptability rather than long-term ecological succession [10,11,12,13]. Therefore, empirical field-based studies integrating dendrochronological and vegetation data are required to elucidate the aging mechanisms and successional potential of these hybrid plantations.

The P. × rigitaeda stand investigated in this study is located in the Dosan Forest of Mt. Goryeong, central Korea, covering approximately 3.2 ha and consisting of about 300 mature individuals. National forest inventory data classify this stand as roughly 45 years old, yet dendrochronological analysis revealed an actual establishment age of approximately 67 years—indicating that some early plantations were established prior to official record-keeping. This discrepancy emphasizes the necessity of accurate, field-based age verification for understanding structural dynamics in aging artificial forests. The stand exhibits generally vigorous growth but displays increasing structural heterogeneity, with expanding canopy gaps and signs of partial natural regeneration. These features suggest that gap succession is actively occurring, potentially driving a shift in species composition and forest structure.

Understanding how such mature P. × rigitaeda stands evolve under limited human intervention provides critical insight into the ecological pathways of hybrid conifer succession in temperate monsoon climates. The specific objectives of this study were to: (1) determine the actual stand age through dendrochronological analysis and evaluate discrepancies with national forest inventory data; (2) analyze age-related growth decline and canopy closure dynamics; and (3) examine how canopy closure mediates understory diversity and compositional shifts. We hypothesize that the stand’s true aging stage drives both growth deceleration and canopy densification, which in turn accelerate the replacement of conifer dominance by shade-tolerant broadleaf species. By linking verified stand age, canopy structure, and understory composition, this study seeks to clarify the self-organizing successional processes in hybrid pine plantations and to contribute to a broader understanding of post-plantation forest transitions in East Asia.

2. Materials and Methods

2.1. Study Site

The study was conducted in the Dosan Forest, located on the southeastern slope near Majang Lake, Gisan-ri, Gwangtan-myeon, Paju City, Gyeonggi Province, Republic of Korea (37°48′ N, 126°56′ E) (Figure 1). The study area map was produced using QGIS (version 3.40) [14] with the VWorld map as the base layer. The site covers approximately 3.2 ha and comprises about 300 individuals of P. × rigitaeda Hyun & Ahn. According to the National Forest Spatial Information Service [15], the stand is categorized as a mid-aged (approximately 45 years) plantation; however, dendrochronological analysis revealed an actual age of approximately 67 years. This discrepancy underscores the need for field-based age verification to improve ecological interpretation of stand aging processes.

The forest lies on a southeast-facing slope (15–25° inclination) with brown forest soils classified as brown moist and somewhat wet forest soils [16]. These soils exhibit moderate drainage and high water-holding capacity, favorable for mixed conifer–broadleaf forest succession [17]. The site is designated as a “landscape and recreation forest zone” [15].

2.2. Data

2.2.1. Field Sampling

Field surveys were conducted in 2025 in a P. × rigitaeda plantation within the Dosan Forest. Fifteen 10 × 10 m quadrats were systematically established to capture spatial variation in topography and stand structure. The plots were distributed across slopes of 15–25°, with aspects ranging from southeast to south and elevations of 180–230 m above sea level.

Within each of the fifteen 10 × 10 m quadrats, vegetation was surveyed by vertical layer—canopy (T1), subcanopy (T2), shrub (S), and herb (H)—following the Braun–Blanquet cover–abundance scale [18,19]. To characterize stand structure, representative individuals from the canopy and subcanopy layers were selected, and their diameter at breast height (DBH) and total height were measured. A summary of the overall floristic composition recorded across all plots is provided in Supplementary Materials (Table S1), while layer-specific vegetation cover and dominant species composition are presented in Supplementary Materials (Table S2). Herb-layer species were recorded within 1 × 1 m subquadrats located at the center of each plot to improve the precision of abundance estimation. All vascular plant species were identified in the field, and their nomenclature followed the National Standard Plant List of Korea [20].

2.2.2. Tree-Ring

To verify stand age and assess radial growth, increment cores were collected from 16 dominant trees using a Pressler borer (CO450, Haglöf, Långsele, Sweden) at 1.2 m above ground level. Core samples were air-dried, mounted, sanded, and measured to the nearest 0.001 mm using a Mitutoyo digital caliper. Tree-ring data were subsequently used for age determination and growth-trend analyses.

2.2.3. Environmental and Climatic Data

Annual precipitation data (1962–2025) were obtained from the Korea Meteorological Administration [21] for the Paju weather station (code 119) (Supplementary Materials: Table S3). The long-term precipitation record was used to examine the relationship between annual rainfall and radial growth derived from tree-ring data. Soil physical properties and site classification were extracted from the National Forest Soil Database [15].

2.3. Methods

All statistical analyses were performed using R 4.5.1 [22]. Vegetation classification, ordination, successional assessment, growth–climate modeling, and survival analysis were performed using established ecological methods as summarized below.

2.3.1. Two-Way Indicator Species Analysis (TWINSPAN)

Vegetation communities were classified using the TWINSPAN [23] in R [24], a divisive hierarchical algorithm that iteratively partitions plots based on correspondence analysis (CA) scores and identifies indicator species for each division [25]. This method is widely used for detecting compositional discontinuities and defining ecologically homogeneous groups in community data [26,27]. The species-by-plot abundance matrix was constructed from Braun–Blanquet cover values converted to percentage midpoints. Group classification was determined by recursive binary splits using eigenvector scores from CA as follows:

$$G_1=\left\{ j |{ s}_j \le t \right\},G_2=\left\{ j |{ s}_j >t \right\}$$

(1)

where $s_j$ represents the CA score of plot j and t is the partition threshold. Group membership ($z_{jg}$) was defined as 1 if plot $j$ belongs to group $G_g$, and 0 otherwise, allowing explicit binary partitioning of plots. Indicator species are identified based on their differential abundance and frequency between the two subgroups. The recursive division proceeds until the eigenvalue or the minimum group size criterion is reached, resulting in a classification tree that reflects the hierarchical organization of vegetation assemblages. For each terminal group, the mean Braun–Blanquet abundance of each species was computed as:

$${\overline{A}}_{ig}={\displaystyle \frac{1}{n_g}}\sum _{j\in G_g}{A}_{ij}$$

(2)

where $A_{ij}$ is the Braun–Blanquet cover value of species i in plot j, $n_g$ is the number of plots belonging to group g, and $G_g$ denotes the set of plots classified into group g [18].

2.3.2. Non-Metric Multidimensional Scaling (NMDS)

To identify variation in species composition among plots, non-metric multidimensional scaling (NMDS) was performed using Bray–Curtis dissimilarity [28,29,30]. NMDS is a rank-based ordination method that minimizes Kruskal’s stress between distance matrices in reduced-dimensional space [26,27,31,32]. The Bray–Curtis dissimilarity between sites i and j was calculated as:

$$D_{ij}=1-{\displaystyle \frac{2C_{ij}}{S_i+S_j}}$$

(3)

where $C_{ij}$ is the sum of the shared abundances between sites, and $S_i$ and $S_j$ are the total abundances at each site. Ordination was then obtained by minimizing Kruskal’s stress function:

$$S=\sqrt{{\displaystyle \frac{\sum _{i<j} (D_{ij}-{\widehat{D}}_{ij}{)}^2}{\sum _{i<j} D_{ij}^2}}}$$

(4)

where $D_{ij}$ and ${\widehat{D}}_{ij}$ denote the observed and ordinated distances, respectively. This iterative procedure continues until the stress value stabilizes, indicating convergence of the ordination configuration. Stress values below 0.2 were considered acceptable, indicating a stable ordination. To visualize ecological gradients, environmental variables (e.g., slope, aspect, and elevation) were fitted as vectors onto the ordination space using the envfit() function in the vegan package [33]. Through this process, the relationships between species composition and site conditions were quantitatively examined.

Although NMDS is based on an ordinal and non-additive framework, in which the independence of axes is heuristically rather than formally justified [31,34,35], this method was adopted for its robustness in handling nonlinear and non-normal ecological data.

In this study, NMDS was applied as a practical ordination tool that emphasizes empirical interpretability while acknowledging its theoretical limitations in community composition analysis.

2.3.3. Successional Index (SI)

The relative importance of species in successional trends was evaluated using the importance value (IV) index [36], calculated as:

$$I{V}_{k,q}={\displaystyle \frac{R_{k,q}+F_{k,q}+C_{k,q}}{3}}$$

(5)

where $R_{k,q}$, $F_{k,q}$, and $C_{k,q}$ denote relative density, frequency, and cover of species k in quadrat q. The SI was initially computed as the mean IV across all quadrats, serving as a proxy for potential dominance in future successional stages [37].

To incorporate the vertical structure of vegetation, the SI was further extended to a layer-weighted successional index, integrating the importance values ${IV}_{il}$ across canopy layers ($l\in \left\{T1, T2, S, H\right\}$). For each species i, a weighted mean was calculated as:

$${SI}_i= \sum _l{w}_l{ IV}_{il}$$

(6)

where $w_l$ represents the successional weight assigned to layer l (with $w_{T1}$ > $w_{T2}$ > $w_S$ > $w_H$). This formulation preserves the interpretability of the traditional IV while accounting for species distribution across layers, providing a more realistic measure of successional position [38].

2.3.4. Coefficient of Variation (CV)

To explore nonlinear relationships between tree-ring growth and environmental factors, a Generalized Additive Model (GAM) was employed. Prior to GAM fitting, the raw ring-width series were examined for internal consistency by computing the coefficient of variation (CV) across individual trees [39]. The CV was calculated for each tree as:

$$CV={\displaystyle \frac{\sigma }{\mu }}$$

(7)

where $\sigma$ and $\mu$ denote the standard deviation and mean of the annual ring-width series, respectively. Outliers (TreeIDs with $CV>2$ or $CV<0.2$) were excluded to ensure stable mean growth estimates. Subsequently, age-dependent variations in relative radial growth were assessed using ANOVA and smoothed LOESS curves to identify characteristic growth trajectories with age [40]. These preliminary analyses served as diagnostics for both nonlinearity and data reliability prior to formal GAM fitting.

2.3.5. Generalized Additive Model (GAM)

A GAM was used to examine nonlinear relationships among ring-width growth, stand age, and precipitation [41,42]. The model was specified as:

$$y_i={\beta }_0+f_1\left({\mathrm{Age}}_i\right)+f_2\left({\mathrm{PRCP}}_{i-lag}\right)+{ϵ}_i$$

(8)

where $y_i$ is the annual ring width of observation i, ${\mathrm{Age}}_i$ is the tree age, ${ϵ}_i$ is the error term and ${\mathrm{PRCP}}_{i-lag}$ denotes total precipitation in the current or preceding year (lag = 0–2). The functions $f_1$ and $f_2$ are penalized regression splines estimated via REML. This approach allows for smooth nonlinear responses while controlling model complexity.

Separate GAMs were fitted for each precipitation lag (0–2 years) to quantify delayed climatic effects [43,44]. Model diagnostics (generalized cross-validation score, effective degrees of freedom, and adjusted R²) were evaluated to assess model fit. Stress values of $f_1\left(Age\right)$ and $f_2\left(PRCP\right)$ were inspected to ensure adequate smoothness and prevent overfitting. Visualization of fitted splines with 95% confidence intervals was used to interpret the nonlinear relationships between ring growth and predictors [43,45,46].

The combination of preliminary CV and LOESS analyses with the GAM framework provides a robust assessment of growth–climate relationships [40]. The CV and relative growth rate analyses ensured data stability and identified representative age-related trends, while GAM captured the complex nonlinear interactions between tree age and precipitation.

2.3.6. Weibull Survival Function

To describe age-related mortality and survival patterns of P. × rigitaeda, a Weibull survival model [47,48] was applied. The survival probability function was expressed as:

$$S\left(t\right)=e^{-(t/\lambda {)}^k}$$

(9)

where $S\left(t\right)$ represents survival probability at time t, $\lambda$ is the scale parameter, and $k$ is the shape parameter describing the aging rate. The parameters were estimated using nonlinear least squares (NLS) in R. The Weibull function provides a flexible representation of stand dynamics and has been widely used in forestry to model growth decline and mortality over time [47,49,50,51].

3. Results

3.1. Overall View of Data

A total of 15 quadrats (10 × 10 m each) were surveyed within the P. × rigitaeda plantation at Dosan Forest during 2025. Across all plots, a total of 69 vascular plant species were identified, representing 41 families and 56 genera (Supplementary Materials: Table S1). The species composition exhibited a distinct stratification pattern across vegetation layers. The tree layer comprised 10 species from 6 genera and 5 families, while the sub-tree, shrub, and herb layers contained 22, 33, and 50 species, respectively (Supplementary Materials: Table S2).

Among the 69 species, eight species appeared in more than 12 of the 15 quadrats, indicating a core assemblage that defines the structure of this hybrid pine stand. These were Pinus × rigitaeda, Castanea crenata, Styrax japonicus, Fraxinus rhynchophylla, Styrax obassia, Acer pseudosieboldianum, Viburnum dilatatum, and Disporum smilacinum. Of these, P. × rigitaeda, C. crenata, and S. japonicus occurred in all plots, suggesting that they play a fundamental role in both canopy and subcanopy dynamics.

Species overlap across vertical layers also revealed significant structural integration. C. crenata, S. japonicus, and S. obassia were shared among the canopy, subcanopy, and shrub layers, implying an ongoing process of vertical regeneration and species turnover. The consistent presence of A. pseudosieboldianum and D. smilacinum in both shrub and herb layers further suggests a successional trajectory toward mixed deciduous composition, consistent with reports from similar P. rigida plantations in central Korea [3,52].

Overall, the data indicates that the P. × rigitaeda stand in Dosan Forest maintains a relatively high level of species diversity and structural heterogeneity, which is atypical for even-aged artificial stands of hybrid pine. The dominance of deciduous broad-leaved species in the lower layers suggests that the forest is currently in a mid-successional stage, transitioning from a conifer-dominated canopy toward a mixed deciduous community. These results provide a basis for subsequent analyses of stand dynamics, growth patterns, and successional modeling (Section 3.2–Section 3.4).

3.2. Structural Attributes of the Stand

The P. × rigitaeda stand exhibited a moderate level of structural differentiation, reflecting the gradual transition from a homogeneous plantation to a mixed forest structure. The mean diameter at breast height (DBH) of dominant P. × rigitaeda individuals was 30.8 ± 10.4 cm, with an average height of 22.1 ± 1.4 m. The mean stem density was about 100 trees ha⁻¹, and the calculated canopy closure averaged 0.81, indicating relatively high crown overlap and limited understory light penetration.

In contrast, subcanopy and accompanying deciduous species, mainly C. crenata, S. japonicus, and F. rhynchophylla, displayed smaller mean DBH values (9.5 ± 7.2 cm) and lower heights (8.8 ± 5.5 m). These species collectively accounted for approximately 42% of total stem count, suggesting substantial recruitment beneath the hybrid pine canopy.

The size distribution of P. × rigitaeda followed a right-skewed unimodal pattern, typical of even-aged stands [5,49,53], while deciduous species exhibited a near-lognormal distribution, implying active regeneration and differential growth rates among subordinate species. The high frequency of small-diameter C. crenata and S. japonicus individuals supports the interpretation that natural succession toward a mixed broad-leaved community is underway.

3.3. Quadrat Analysis

3.3.1. TWINSPAN

Vegetation data were classified using the Two-Way Indicator Species Analysis (TWINSPAN) method [23] to delineate ecologically meaningful vegetation groups based on species composition and abundance. This classification approach, widely adopted in vegetation ecology [26,27], produced seven terminal community types—*00, *010, *0110, *0111, *100, *101, and *11—each representing a distinct assemblage of dominant and diagnostic species (Figure 2).

Across all groups, P. × rigitaeda was identified as the most dominant and widely distributed species, indicating its strong ecological adaptability and competitive advantage within the study area. Group *00 (PlotID: P10) consisted primarily of P. × rigitaeda, Celtis jessoensis, and Acer pictum var. mono. Group *010 (P01, P02) was characterized by P. × rigitaeda, Acer palmatum, and C. crenata, whereas Group *0110 (P09, P11, P12, P14) exhibited co-dominance of P. × rigitaeda, C. crenata, and A. pseudosieboldianum. Group *0111 (P15) was dominated by A. pseudosieboldianum, P. × rigitaeda, and S. japonicus. The *100 (P03, P05, P07) and *101 (P04, P06, P08) groups displayed similar compositional structures, both dominated by P. × rigitaeda and C. crenata, representing intermediate community stages of mixed forest development. Finally, Group *11 (P13) was distinct, characterized by P. × rigitaeda, D. smilacinum, and Quercus × urticifolia, indicating an understorey-differentiated assemblage associated with site-specific microenvironmental or successional conditions.

Overall, the TWINSPAN classification effectively differentiated vegetation assemblages and revealed compositional gradients dominated by P. × rigitaeda, demonstrating the utility of the TWINSPAN approach for identifying hierarchical community structures in forest ecosystems [23,26,27]

3.3.2. Integrated NMDS Ordination of Vegetation Layers

An integrated non-metric multidimensional scaling (NMDS) ordination combining the tree (T1), subcanopy (T2), shrub (S), and herb (H) layers revealed that the sampling plots were widely scattered across the two-dimensional ordination space (Figure 3). No distinct floristic assemblages or clustering patterns were observed, indicating a high degree of compositional heterogeneity among plots and the absence of clearly delineated community groups. The stress value of the ordination (0.185) suggested a satisfactory representation of multidimensional species relationships [29].

Environmental vectors fitted by the envfit procedure showed that slope, elevation, and aspect were primarily aligned along the NMDS1 and NMDS2 axes. Among these, slope exhibited a significant correlation with the ordination (R² = 0.559, p = 0.014), whereas elevation and aspect were not significant (p > 0.05). Despite the significant slope effect, the relatively short vector lengths indicated that these topographic variables collectively explained only a small proportion of the overall community variation. This finding suggests that local vegetation heterogeneity within the P. × rigitaeda stand is influenced more by stochastic or micro-environmental processes than by the measured topographic gradients. Such a pattern aligns with the interpretation that fine-scale environmental variation often overrides coarse topographic effects in structurally complex forest stands [30].

3.3.3. Layer-Specific NMDS Ordination and Environmental Gradient Patterns

Non-metric multidimensional scaling (NMDS) analyses [29] conducted separately for each vegetation layer revealed distinct spatial patterns in community composition in relation to environmental gradients (Figure 4). Stress values were below 0.2 across all strata (0.077–0.155), indicating satisfactory ordination fits [31,54]. The ordinations were based on Bray–Curtis dissimilarity [28], a robust metric for assessing compositional variation among plots [33].

In the canopy (T1) layer, ordination points were relatively compact, but correlations with topographic factors were weak and statistically non-significant (slope: R² = 0.251, p = 0.183; elevation: R² = 0.187, p = 0.202; aspect: R² = 0.169, p = 0.325). Similarly, in the subcanopy (T2) layer, correlations with elevation (R² = 0.106, p = 0.519), slope (R² = 0.100, p = 0.552), and aspect (R² = 0.027, p = 0.861) were non-significant.

By contrast, the shrub (S) layer exhibited a significant relationship with slope (R² = 0.433, p = 0.028), whereas elevation (R² = 0.183, p = 0.303) and aspect (R² = 0.051, p = 0.734) were not significant. This suggests that slope-driven gradients, such as drainage and soil moisture, exert a measurable influence on species composition in the lower strata. In the herb (H) layer, all correlations were weak and non-significant (slope: R² = 0.098, p = 0.524; elevation: R² = 0.108, p = 0.496; aspect: R² = 0.069, p = 0.669).

Overall, quadrat groupings (Groups 1–5) displayed partial overlaps among strata, but greater separations were observed in the S and H layers along slope and elevation gradients. These results indicate that while broad-scale topographic gradients have limited effects on the upper layers, they play a more pronounced role in structuring shrub-layer composition within P. × rigitaeda stands.

3.3.4. Successional Index (IV-Weighted)

The IV-weighted successional index, derived from species importance values (IV), was applied to identify potential successor species for each vegetation layer (Figure 5). The calculation followed the classic method of Curtis and McIntosh [36] and Mueller-Dombois and Ellenberg [19], in which relative frequency, density, and dominance were integrated to estimate species importance values.

In the T1 layer, C. crenata exhibited the highest index, followed by Q. variabilis, Q. acutissima, Q. mongolica, and A. pseudosieboldianum. The T2 layer was similarly dominated by C. crenata, along with S. japonicus, F. rhynchophylla, and S. obassia. In the S layer, A. pseudosieboldianum and S. obassia showed the highest successional scores, whereas in the H layer, D. smilacinum and A. pseudosieboldianum were the most prominent species.

Among canopy-forming taxa, C. crenata consistently ranked highest across both the tree and subcanopy layers, designating it as the most probable successor to P. × rigitaeda. Several Quercus species—Q. variabilis, Q. acutissima, Q. mongolica, and Q. aliena—also exhibited high successional indices, indicating an ongoing ecological transition toward a Castanea–Quercus-dominated deciduous forest. Such a pattern is consistent with previous studies on post-pine decline succession in Korean temperate forests [55,56], suggesting that the studied P. × rigitaeda stand is progressing toward a mixed hardwood community dominated by shade-tolerant, late-successional species.

3.4. Tree-Ring Analysis

3.4.1. Growth Variability Analysis

The analysis of ring-width variability (Figure 6) revealed that the coefficient of variation (CV) of relative growth rate among P. × rigitaeda individuals ranged from 0.4 to 0.9, indicating stable and uniform growth dynamics across the stand. Most trees exhibited moderate variation (CV ≈ 0.5), while a few individuals (Tree 37, Tree 41, and Tree 51) showed slightly higher variability (CV ≈ 0.8–0.9). No individual exceeded the upper reference threshold (CV = 2.0) for extreme anomalies, and none displayed excessively low variability (CV < 0.2).

The use of the coefficient of variation (CV) as an indicator of growth stability has been widely adopted in dendrochronological and forest ecological studies to quantify interannual growth variability and detect anomalous growth behavior [43,57,58]. In this study, the observed CV range suggests moderate inter-individual variation, consistent with expectations for mature P. × rigitaeda stands under relatively homogeneous site conditions.

These findings confirm that P. × rigitaeda trees maintained moderate interannual growth variability without evidence of severe physiological stress or abrupt growth suppression. Such stability is characteristic of mature stands growing under relatively uniform microenvironmental and competition-related conditions, reflecting consistent resource availability and resilience at the individual-tree level.

3.4.2. Age-Dependent Growth Dynamics and Climatic Effects

Relative growth rate (RGR) analysis demonstrated a clear age-dependent decline in radial growth (Figure 7). High early-stage growth (RGR > 0.04 yr⁻¹) persisted until approximately age 10, followed by a marked decrease after 20–25 years. Beyond age 30, RGR stabilized below 0.02 yr⁻¹, and only minor fluctuations were observed in older individuals (>50 years). The mean inflection point occurred around age 23, indicating the onset of physiological slowdown. This age-related reduction in productivity has been widely reported across conifers [5,59].

Generalized Additive Model (GAM) fitting (Figure 8) revealed a strong nonlinear relationship between age and annual ring increment. The smooth terms (Age) showed a significant negative slope beyond 20 years, whereas s(PRCP_L0)—representing total annual precipitation—remained statistically insignificant and nearly flat. This finding supports that P. × rigitaeda growth is primarily governed by ontogenetic aging rather than interannual rainfall variability. Similar approaches using GAMs have been applied in tree-ring and growth studies to identify nonlinear climate–growth relationships [42,60].

3.4.3. Survival Analysis of Artificial Hybrid Pine Using the Weibull Function

To quantify the age-related survival dynamics of P. × rigitaeda, the Weibull survival function was fitted to stand-level mortality data. The estimated parameters were shape (k) = 1.3 and scale (λ) = 68.4, indicating a progressively increasing mortality rate with stand age (Figure 9). The survival probability remained above 0.8 until approximately 50 years, followed by a rapid decline beyond 60 years, signifying the onset of late-successional senescence [61,62,63,64,65,66,67].

This accelerating hazard (k > 1) is characteristic of self-thinning and physiological aging in maturing conifer stands [6,68,69]. Comparable Weibull-based mortality patterns have been observed in P. sylvestris and P. taeda plantations, where age-dependent mortality is driven by stand density, competition, and hydraulic limitation [53,70].

The Weibull model yielded a median residual lifespan of approximately 15–20 years, suggesting that plantations established in the 1960s are approaching a demographic transition phase. These results quantitatively demonstrate that P. × rigitaeda follows a predictable sigmoid-type survival curve governed by ontogenetic constraints and site-specific stress, confirming the utility of the Weibull model for predicting stand longevity [51].

4. Discussion

This study investigated the successional dynamics and stand structure of an artificial hybrid pine (P. × rigitaeda) plantation, which represents a characteristic conifer system of temperate monsoon Asia. The observed mechanisms—canopy aging, gap dynamics, and hybrid-driven facilitation of deciduous recruitment—reflect ecological processes widely recognized in conifer plantations across temperate regions worldwide (e.g., P. taeda, P. rigida, P. sylvestris). These findings provide new insight into how hybrid conifer plantations can serve as transitional systems bridging artificial afforestation and natural forest recovery under comparable climatic regimes.

4.1. Structural and Compositional Transitions Toward Mixed Forests

4.1.1. Vegetation Classification (TWINSPAN)

The TWINSPAN classification delineated seven distinct vegetation groups within the P. × rigitaeda plantations, representing different compositional stages along the post-plantation successional gradient (Figure 2). Although P. × rigitaeda remained dominant across all groups, the increasing co-dominance of deciduous broadleaved species such as C. crenata, A. pseudosieboldianum, and S. japonicus indicates an active floristic reorganization within the stands. This shift suggests that differences in shade tolerance and regeneration ability are key ecological drivers shaping the community structure.

From an ecological standpoint, groups located on the left branches of the TWINSPAN dendrogram (e.g., *00, *010) correspond to early- to mid-successional assemblages characterized by strong P. × rigitaeda dominance and limited understory diversification. In contrast, right-branch groups (*0110, *0111, *11) exhibit higher broadleaf representation, particularly C. crenata and A. pseudosieboldianum, signifying transitional phases toward mixed deciduous forest composition. These latter groups show greater species richness and vertical stratification, implying that endogenous canopy gap formation is facilitating the recruitment of shade-tolerant hardwoods beneath the aging pine canopy.

Such a hierarchical floristic progression aligns well with previous findings in P. rigida and P. densiflora plantations across central Korea, where canopy dominance gradually shifts toward Quercus and Castanea species over several decades [71,72,73,74]. The presence of mesic understory herbs such as Disporum smilacinum in the 11 group further supports the inference that microclimatic stabilization and litter accumulation are progressing toward conditions typical of mature deciduous forests.

Therefore, the TWINSPAN results not only delineate compositional boundaries but also reveal the underlying ecological processes governing community turnover. The coexistence of pioneer (P. × rigitaeda) and late-successional (C. crenata, A. pseudosieboldianum) species indicates that the stands are entering a phase of self-thinning and structural diversification driven primarily by endogenous canopy dynamics rather than exogenous disturbance. From a restoration perspective, this floristic mosaic demonstrates the potential for aging hybrid pine plantations to evolve into semi-natural mixed forests through autogenic succession, emphasizing the ecological importance of retaining mature hybrid pines as transitional canopy facilitators rather than removing them prematurely.

Overall, these results corroborate previous studies that reported spontaneous successional trajectories in Pinus plantations of Korea. P. rigida stands, for instance, were shown to gradually transform into mixed broadleaved forests dominated by Q. variabilis and C. crenata with stand aging [71]. Similarly, P. densiflora forests have been observed to undergo comparable compositional shifts through shade-tolerant broadleaf recruitment and canopy stratification [72,73]. Park et al. [74] further interpreted these trends as a self-driven successional process resulting from canopy aging and internal gap dynamics. The present study thus provides quantitative evidence consistent with these successional models, reinforcing the notion that P. × rigitaeda plantations in Korea can naturally transition toward ecologically stable mixed deciduous forests.

4.1.2. Structural Heterogeneity (NMDS)

The NMDS ordination of vegetation layers (stress = 0.185; Figure 3) revealed no clear clustering among plots, indicating substantial structural heterogeneity within the P. × rigitaeda stands. The broad scatter of sample points suggests that species composition varies primarily along fine-scale environmental gradients rather than strong topographic controls. Among tested variables, only slope showed a significant correlation (R² = 0.559, p = 0.014; Figure 3), whereas elevation and aspect were not significant, implying limited topographic influence on overall community organization.

This pattern reflects an internally driven structural differentiation process associated with canopy openings and localized variation in understory light conditions. Similar results have been reported in aging Pinus rigida and P. densiflora plantations in Korea, where within-stand heterogeneity and canopy gap dynamics play crucial roles in regulating understory composition [71]. Such weak coupling between topography and species assemblages aligns with the ecological interpretation that fine-scale microsite conditions dominate community turnover in later successional stages [9,30].

Therefore, Figure 3 captures a key feature of the P. × rigitaeda forest mosaic—high internal heterogeneity under minimal environmental constraint—signifying an autogenic transition toward a more structurally complex and ecologically resilient mixed-forest system.

4.1.3. Vertical Ecological Differentiation Across Vegetation Layers (NMDS)

Layer-specific NMDS ordinations (stress ranging from 0.077 to 0.155; Figure 4) revealed moderate but distinct vertical variation in community responses to topographic gradients within P. × rigitaeda stands.

In the canopy (T1) and subcanopy (T2) layers, the weak and statistically non-significant correlations with slope, elevation, and aspect indicate that upper strata are relatively homogeneous in composition and weakly coupled to microtopographic variation. This likely reflects the uniform light and drainage regimes experienced by dominant canopy trees and the buffering effect of overstory cover that reduces environmental heterogeneity in the subcanopy.

In contrast, the shrub (S) layer displayed a significant correlation with slope (R² = 0.433, p = 0.028), suggesting that microtopography and associated soil-moisture gradients strongly influence species turnover and compositional heterogeneity near the forest floor. The herb (H) layer, showing weak and non-significant relationships with all environmental variables, reflects stochastic assembly processes and high microsite dependence typical of forest understories.

As illustrated in Figure 4, these vertical differences represent a form of partial ecological differentiation, characteristic of forest stands transitioning from even-aged monocultures toward structurally and functionally complex systems. Such patterns denote the early stage of vertical niche partitioning, where lower strata begin to respond more sensitively to fine-scale resource gradients [9,75].

Comparable patterns have been observed in aging P. densiflora and P. rigida plantations in Korea, where slope- and elevation-driven microgradients regulate understory regeneration and species turnover across layers [71]. Likewise, studies in subtropical and temperate forests have demonstrated that microtopographic heterogeneity enhances ecosystem resilience by promoting niche diversification and buffering microclimatic fluctuations [76,77,78].

Collectively, these findings suggest that P. × rigitaeda stands are evolving toward a moderately stratified and environmentally responsive system, though the strength of vertical ecological differentiation remains limited to the lower strata. The significant slope effect in the shrub layer underscores the ecological importance of maintaining microtopographic heterogeneity to support understory regeneration, functional diversity, and long-term successional stability in mixed deciduous forest development.

4.1.4. Successional Trajectories (SI Index)

The IV-weighted successional index (Figure 5) illustrates a clear trajectory from P. × rigitaeda dominance toward a hardwood assemblage led by C. crenata and Q. variabilis. Elevated index values for Q. acutissima, Q. mongolica, and A. pseudosieboldianum indicate an ongoing recruitment of shade-tolerant deciduous species in both canopy and subcanopy layers. However, unlike naturally regenerated P. rigida or P. densiflora forests, this transition emerges within artificial hybrid stands that originated from human-assisted crossing between P. rigida and P. taeda. Consequently, P. × rigitaeda exhibits intermediate physiological and ecological traits—rapid juvenile growth inherited from P. taeda and moderate cold tolerance from P. rigida—which create a unique successional context not observed in pure-species plantations.

This hybrid legacy leads to an atypical successional pathway: rapid canopy closure in early stages suppresses understory diversity, but as stand aging induces canopy gaps, regeneration accelerates under partially moderated microclimates. The successional indices demonstrate that this hybrid system is capable of autogenic reorganization, transitioning toward native deciduous dominance without external intervention [55,79]. Such behavior diverges from earlier reports on pure P. rigida or P. densiflora plantations, which required disturbance or management to initiate hardwood invasion [71,73].

Comparable adaptive trajectories have been documented in subtropical artificial conifer hybrids, where genetic admixture broadens ecological amplitude and promotes spontaneous succession [7,8]. In the case of P. × rigitaeda, this hybrid vigor appears to facilitate a self-sustaining shift toward mixed deciduous forest structures. Ecologically, C. crenata dominates newly opened gaps through high drought and light tolerance, whereas Quercus species occupy shaded microsites, forming a stable hardwood matrix beneath remnant hybrid pines.

From a management perspective, these results underscore the ecological potential of artificial hybrids as transitional facilitators in post-plantation succession. Rather than viewing P. × rigitaeda as a temporary silvicultural product, it should be recognized as a dynamic biological bridge that accelerates the conversion of artificial forests into resilient, semi-natural mixed deciduous systems.

4.2. Growth Decline, Physiological Aging, and Mortality Dynamics

4.2.1. Growth Trend (Tree-Ring)

Tree-ring analyses (Figure 5, Figure 6 and Figure 7) revealed a gradual yet consistent reduction in radial growth with stand age in the artificial hybrid pine, P. × rigitaeda. The mean annual ring width declined from 0.41 mm yr⁻¹ during the first 20 years to 0.20 mm yr⁻¹ after approximately 60 years, accompanied by increased inter-annual variability. This pattern reflects a typical age-dependent physiological decline. However, in natural pine species such as P. taeda and P. sylvestris, radial growth typically decreases more sharply—from about 0.9–1.0 mm yr⁻¹ in the first two decades to 0.25–0.30 mm yr⁻¹ by 60 years of age [5,80]. The hybrid origin of P. × rigitaeda thus modifies the conventional trajectory observed in natural pine species. Specifically, the hybrid’s heterotic advantage—combining the rapid juvenile growth of P. taeda with the stress tolerance of P. rigida—appears to delay the onset of senescence, sustaining moderate productivity even under advanced stand age.

Climatic correlation analyses (Figure 6) further showed that older trees exhibited reduced sensitivity to summer temperature and precipitation, suggesting enhanced internal buffering as canopy closure stabilized microclimatic conditions. The resulting growth stability under ageing distinguishes these artificial hybrid stands from their parental or natural analogues. Similar patterns have been noted globally in old conifer systems [81,82], yet P. × rigitaeda displays a more gradual decline curve, indicative of hybrid vigor rather than simple senescence.

As shown in Figure 7, older cohorts maintain narrow but steady increments, signifying a transition from active biomass accumulation to maintenance metabolism. Such sustained, low-amplitude growth mirrors findings from ageing P. densiflora and P. rigida plantations in Korea [71,83] and from Mediterranean conifer systems [84], where endogenous gap dynamics maintain minimal yet continuous radial expansion. In the artificial hybrid, however, this internal equilibrium arises not merely from structural feedbacks but from genetic complementarity, allowing persistence under both climatic stress and resource limitation.

Ecologically, these results position P. × rigitaeda as a transitional bridge species within post-plantation succession. Its prolonged canopy function and moderated growth decline provide a structural and physiological framework for the recruitment of shade-tolerant deciduous species beneath. From a management perspective, recognizing the growth slowdown in artificial hybrids as an intrinsic successional process—not as a silvicultural failure—is crucial. Retaining mature hybrid pines can thus enhance microclimatic stability, facilitate natural hardwood regeneration, and accelerate the ecological conversion of artificial forests into resilient mixed deciduous ecosystems.

4.2.2. Longevity and Mortality (Weibull)

The Weibull survival analysis (Section 3.4.3; Figure 9) revealed a distinct increase in mortality after approximately 60 years, signifying the transition of P. × rigitaeda stands into a late-successional, self-thinning phase. This pattern parallels the age-related productivity decline commonly reported in conifer plantations (P. rigida, P. taeda, P. densiflora) where photosynthetic efficiency, hydraulic conductivity, and nutrient uptake efficiency decrease with age [5,6,53].

Similar climatic limitations have been observed in P. densiflora stands of central Korea, where radial growth becomes increasingly sensitive to temperature and precipitation variability with age [83]. This suggests that P. × rigitaeda may experience comparable physiological constraints during its late-successional phase, particularly under intensifying summer droughts and heat stress.

Despite these constraints, P. × rigitaeda exhibits delayed senescence and extended canopy persistence, reflecting heterotic vigor derived from interspecific hybridization [85,86]. These traits enhance drought tolerance and maintain structural stability for nearly seven decades, exceeding the typical lifespan of its parental species.

The post-60-year mortality surge corresponds to canopy turnover and gap formation, promoting regeneration of shade-tolerant hardwoods such as C. crenata and Quercus spp. [84,87,88]. Ecologically, this phase functions as a self-regulated mechanism facilitating autogenic succession toward mixed deciduous forest systems [89].

From a management standpoint, partial retention of senescent hybrid pines during this transition can stabilize microclimate, preserve soil structure, and ensure ecological continuity. Consequently, P. × rigitaeda should be regarded not merely as a transient plantation species but as a transitional facilitator linking artificial afforestation with natural forest recovery under climate-driven environmental change.

4.2.3. Density–Closure Effects (GAM)

A generalized additive model (GAM) was used to evaluate the nonlinear effects of stand density and canopy closure on the variance of diameter at breast height (DBH) among P. × rigitaeda individuals [42]. The fitted model was expressed as:

$${\mathrm{DBH}}_{ij}={\beta }_0+f_1\left({\mathrm{Density}}_i\right)+f_2\left({\mathrm{Closure}}_i\right)+{ϵ}_{ij}$$

(10)

where $f_1$ and $f_2$ are smooth spline functions for stand density and canopy closure, respectively. Model performance was high (adjR² = 0.85), explaining 94.3% of total deviance (Figure 10).

The smooth functions revealed a hump-shaped response of DBH variance to stand density, peaking at intermediate density levels (≈40–45 in cover-sum index units) and declining at both lower and higher densities. This pattern indicates that moderate stand crowding promotes structural differentiation by balancing competitive stress and light availability. In contrast, a strong negative response was observed with increasing canopy closure, reflecting suppressed diameter differentiation under highly shaded conditions.

Such relationships are consistent with classical theories of competition and self-thinning [90,91] and align with long-term findings from mixed and hybrid pine plantations, where canopy openings enhance growth differentiation. These responses also reflect the heliophilous nature of P. × rigitaeda, which allocates more resources to stem thickening under reduced shading and lower competition.

Ecologically, these results suggest that moderate canopy openness (closure ≈ 0.6–0.7) and intermediate density are optimal for maintaining diameter variability and structural diversity in hybrid pine plantations. Excessive closure or crowding reduces DBH variance, signaling increased competition, growth stagnation, and reduced resilience. Conversely, maintaining mid-density levels can enhance structural diversification and facilitate early-stage mixed-forest succession beneath the hybrid pine canopy [70,92].

4.3. Ecological and Management Implications

As demonstrated by the Weibull survival analysis (Figure 10), the survival probability of P. × rigitaeda sharply declines beyond approximately 60 years, indicating a demographic shift toward senescence and hardwood replacement. This trend corresponds with the non-linear age–growth relationship observed in preceding analyses, suggesting that the hybrid pine has largely completed its functional role as a pioneer stabilizer in early- to mid-successional stages and now faces biological limitations to long-term persistence.

These demographic and physiological patterns are consistent with established observations of age-related productivity decline and self-thinning in coniferous plantations [6,53,70]. Similar successional transitions have been reported in aging P. densiflora stands of central Korea, where radial growth becomes increasingly sensitive to climatic variability [83], as well as in old-growth systems of southern Europe, where mortality pulses trigger canopy turnover and hardwood regeneration [84].

Collectively, these findings demonstrate that P. × rigitaeda stands in Korea are transitioning from monocultural, even-aged plantations to heterogeneous, multi-species forests dominated by Castanea and Quercus. This transformation reflects both biological aging and ecological succession, signaling that the hybrid species has fulfilled its role in early-stage land stabilization but possesses limited potential for continued dominance in mature forest ecosystems.

Therefore, forest management should adopt a succession-facilitation approach:

• Retain mature P. × rigitaeda individuals as temporary canopy cover to buffer microclimate and stabilize soil,
• Encourage natural regeneration of native hardwoods through selective gap creation, and
• Maintain understory integrity and herb-layer diversity to support seedling recruitment and nutrient cycling.

Implementing such adaptive management strategies will accelerate the autogenic transition from artificial hybrid pine plantations toward structurally complex, climatically resilient mixed forests characteristic of the temperate zone of East Asia

In, summary, this study provides integrative empirical evidence linking structural heterogeneity, successional direction, and physiological aging in P. × rigitaeda plantations. By combining NMDS ordination, successional indices, and dendrochronological modeling, it quantifies the mechanistic transition from artificial to natural forest structures—an underexplored process in hybrid pine research. These insights expand the ecological understanding of hybrid pine lifespan and inform adaptive management strategies for facilitating native forest recovery across East Asia.

5. Conclusions

This study provided quantitative evidence of ecological changes in P. × rigitaeda stands, confirming that growth decline and partial dieback begin to occur approximately 60–70 years after establishment. Both the canopy and subcanopy layers are now dominated by various Quercus species (e.g., C. crenata, Quercus variabilis, Q. acutissima, Q. mongolica), indicating that P. × rigitaeda forests have entered a natural succession phase toward deciduous broadleaved forests.

The results of the GAM, NMDS, and Weibull analyses revealed that stand age, physiological constraints, and topographic gradients—rather than precipitation—were the primary drivers of community variation. While the NMDS ordinations indicated only partial vertical differentiation, the significant influence of slope in the shrub layer highlights that microtopographic heterogeneity, rather than broad-scale elevation or aspect, primarily governs species turnover in the lower strata. This suggests that the physiological limitations of P. × rigitaeda and slope-driven microsite variability are key mechanisms driving forest succession [93]. In particular, the Weibull survival model demonstrated that P. × rigitaeda stands have reached the overmature stage, emphasizing the need for ecological transition management in these forests.

Although P. × rigitaeda contributes to slope stabilization and soil coverage in the short term, its long-term replacement by Quercus-dominated deciduous forests appears inevitable. Therefore, instead of artificial removal, management strategies should aim to facilitate natural regeneration through gap succession processes.

While natural hybrids represent spontaneous genetic integration shaped by ecological constraints, artificial hybrids such as P. × rigitaeda embody intentionally assembled trait combinations that allow us to examine how designed genotypes behave under natural successional and climatic conditions. Thus, artificial hybrids function not only as silvicultural tools but also as experimental systems to explore the interface between managed and natural forest dynamics.

Future studies should integrate tree-ring-based growth data with fine-scale environmental variables such as light availability, soil moisture, and microclimatic conditions. Furthermore, developing integrated multivariate models that incorporate wood density, physiological tolerance, and regeneration success of successor species will enhance predictive accuracy for forest transition and strengthen the scientific foundation for the ecological management of P. × rigitaeda stands.