Spatial Distribution and Intraspecific and Interspecific Associations of Dominant Tree Species in a Deciduous Broad-Leaved Forest in Shennongjia, China

Abstract

Studying spatial distribution patterns and intraspecific and interspecific associations of tree species is crucial for understanding the maintenance of biodiversity and offering insights into community dynamics and stability. The Shennongjia National Park, located in the transition zone between the (sub)tropics and the temperate climate, holds great significance for understanding how species interact with each other and coexist within forest communities. We used data from a fully mapped 25 ha montane deciduous broad-leaved forest dynamic plot at Shennongjia (SNJ) National Park, central China, to conduct a community-level evaluation of spatial distribution patterns and intraspecific and interspecific associations. We analyzed the spatial distribution patterns of 20 dominant species with univariate and bivariate g(r) functions, as well as intraspecific and interspecific associations across different life-history stages. We assessed the relative contributions of underlying processes in community assembly with three models: complete spatial randomness (CSR), heterogeneous Poisson (HP), and antecedent condition (AC). The results showed that all 20 tree species exhibited aggregated distribution patterns within a 100 m scale. After excluding the influence of environmental heterogeneity, the degree of aggregation decreased, and with the increasing spatial scale from 0 to 100 m, the distribution gradually shifted from aggregated to random or uniform appearance. Positive associations were common in different life-history stages. Negative associations were common across different species, while most of the intraspecific and interspecific associations turned out to be irrelevant when environmental heterogeneity was excluded. We concluded that habitat heterogeneity and dispersal limitation may primarily determine the spatial distribution of species in subtropical montane deciduous broad-leaved forests. This indicates that species distribution may align with environmental patterns, and interspecific correlations may exist. However, the exact responses of these species to environmental changes remain uncertain. Upcoming management approaches ought to concentrate on ongoing observation, which is crucial for mitigating how climate change might affect species distribution and community interactions, thus guaranteeing enduring stability and the conservation of biodiversity.

1. Introduction

Understanding local species interactions within communities is crucial for maintaining species diversity, which is one of the most important ways to mitigate biodiversity loss and a major driver of global ecosystem change [1,2]. The spatial distribution pattern refers to the arrangement of individuals within a geographic area, whether clustered, evenly spaced, or randomly dispersed. These findings uncover underlying ecological aspects like the intensity of clustering, the scale of independent effects, and populations’ spatial layout and elucidate how environmental factors, along with local competition both within and between species, impact plant distribution in specific regions [3]. Such models play a crucial role in discovering the processes that foster species survival and community structure, especially in diverse ecosystems like warm, temperate, mountainous, deciduous broad-leaved woodlands. They embody plants’ ecological responses to environmental changes and competitive pressures, serving as a vital tool for detecting community assembly rules [4,5], which contribute to the development of effective strategies for biodiversity conservation and forest management [1,2].

Habitat heterogeneity and dispersal limitation are two key hypotheses explaining the aggregated distribution of tree species. Habitat heterogeneity, rooted in niche-based theory, posits that deterministic factors such as environmental conditions, local species interactions, and resource availability drive species aggregation patterns [6,7]. Previous studies have shown that the effects of habitat heterogeneity on species distribution accumulate across different life stages, leading to variations in spatial patterns depending on the growth stage and habitat type [8,9,10,11]. The Janzen–Connell theory reinforces this idea, proposing that increased death rates in seeds or seedlings close to original trees due to predation or particular pests open avenues for new species to colonize, influencing their distribution and promotion levels [12,13]. Differing from niche-based theory, dispersal restriction is seen as unbiased, suggesting that the patterns of species distribution and community behavior mainly stem from random occurrences like birth, death, movement, and migration restrictions [14]. This perspective emphasizes the role of random processes in shaping ecological communities rather than deterministic biotic or abiotic factors. Hubbell et al. (1979) [15] noted, for instance, an almost exponential reduction in seedling density as the distance from the mother tree increases, underscoring the critical role of seed dispersal processes in shaping the patterns of species clustering. While habitat heterogeneity and dispersal limitation provide frameworks for understanding the aggregated distribution of tree species, research on the effects of varying scales of habitat heterogeneity on species distribution and the potential influence of local climate variations on dispersal mechanisms remains insufficient. Further exploration of how species interactions change with fluctuations in environmental conditions can deepen our understanding of community assembly processes and inform effective biodiversity conservation strategies.

Analyzing spatial point patterns is a crucial ecological method for clarifying the spatial spread and connections in plant communities, offering a vital understanding of fundamental ecological dynamics [16]. This method categorizes spatial distributions into three primary patterns: clustered (aggregate or clumped), regular (uniform or segregated), and random. Likewise, spatial correlations among various point types can be classified into categories like attraction (positive correlation), the absence of interaction (no correlation), or repulsion (negative correlation) [17,18]. An essential advancement in this method involves replacing Ripley’s K function using conventional circles with univariate pairwise and pair correlation functions for circles with radius r, opting for circles with a defined width instead. This alteration successfully diminishes the impact of minor aggregate influences, leading to improved precision and dependability in analyzing spatial patterns [19,20]. By maximizing the utilization of spatial coordinates of individual plants, this method provides a detailed and nuanced understanding of ecological dynamics, making it an invaluable tool for ecological research and conservation efforts. This study utilizes a range of point methods like the homogeneous Poisson process, followed by the heterogeneous Poisson process, the homogeneous Thomas process, and the heterogeneous Thomas process again, to analyze point patterns and spatial associations at diverse scales. Such procedures facilitate a numerical evaluation of pattern traits and provide robust explanations for the mechanisms underlying the processes of species coexistence [4,21,22].

Shennongjia National Park lies in a region that acts as an intermediary zone between warm temperate and cold temperate climates and is celebrated worldwide for its crucial role in biodiversity preservation. The dominant vegetation in this region is the deciduous broad-leaved forest, primarily composed of Fagaceae species, which are characteristic of Northern Hemisphere forests. Given its high biodiversity and sensitivity to climate change, this transitional forest is of critical importance for studying species coexistence and developing strategies to better cope with future climate change [23]. While holding ecological importance, this region highlights the urgent requirement for extensive ecological studies, focusing on tree populations and their community dynamics in large plots in particular. To address this gap, our study aims to investigate the spatial distribution patterns of 20 dominant tree species and their associations within the forest. In this study, we explored the spatial patterns of tree species under two models—CSR (Completely Spatially Random) and HP (Habitat Patchiness)—to test the applicability of the Janzen–Connell hypothesis, which posits that seedling survival decreases with the proximity to parent trees due to density-dependent mortality and herbivory. Specifically, we address the following research questions: (a) Do the mechanisms influencing species spatial distribution vary across different spatial scales? (b) How do intraspecific and interspecific associations vary across different life-history stages? (c) Are the distribution patterns of dominant species consistent across the canopy, sub-canopy, and shrub layers? (d) What ecological processes drive the formation of the observed distribution patterns of dominant species in this forest plot? By addressing these questions, this study explores the role of negative density dependence in shaping the spatial distribution of dominant species in subtropical mountainous forests.

2. Materials and Methods

2.1. Study Site

The study site is located in Shennnongjia National Park (31°24′27.2657″ N, 110°23′57.6958″ E) in central China. The altitude range is 1824–2158 m. This area is characterized by a mid-subtropical mountain deciduous broad-leaved forest (Quercus aliena var. acutiserrata—Fagus engleriana) (Figure 1). For the benefit of international readers, a map of China will be included, followed by a regional map indicating the location of Shennongjia National Park, and finally, a detailed plot map. According to the meteorological station (1700 m a.s.l.), the mean annual temperature was 10.6 °C, and precipitation averaged 1330 mm, mainly in June-October. The relative humidity averages 75%. Due to precipitation, the humidity at the sample site remained high throughout the year, with the highest levels typically occurring in July and the lowest levels typically occurring in April (>70%).

2.2. Data Collection

A 25 ha (500 m × 500 m) forest dynamic plot (Shennongjia, China, SNJ) was established in August 2022, following the protocol of ForestGEO (The Forest Global Earth Observatory). The plot measures 500 m by 500 m, oriented in a true north–south direction. Initially, the area was subdivided into 2500 plots of 10 m × 10 m using differential GPS, with PVC pipes installed as markers. The 25 ha plot was systemically divided into 625 subplots of 20 × 20 m. At each vertex of the 20 m × 20 m plots, cement stakes were embedded for permanent fixation. Subsequently, each 20 m × 20 m plot was further divided into 16 smaller squares of 5 m × 5 m using PVC pipes and packaging lines. Within each quadrat, all trees with a diameter at breast height (DBH) ≥ 1 cm were tagged, recorded, measured, identified by species, and mapped. The elevation range of the plot is 1824–2158 m, with an average elevation of 1986.88 m. The plot has a slope of 4.54–46.90° over an area of 20 m × 20 m. The direction of the slope ranges from 4.57 to 229.56°, and the degree of convexity ranges from −5.7 to 8.31. The first censuses showed that there were 149 species (including varieties) and 61,054 individuals (including 97,664 individuals with trunks and branches) in this plot, belonging to 44 families and 79 genera, of which 28 were evergreen and 121 were deciduous species [24]. The dominant canopy species are Q. aliena var. acuteserrata, F. engleriana, and Betula albo-sinensis. The dominant sub-canopy species are Acer pentaphyllum, Lindera trifoliata, and Acer tetrastigma, and the shrubs are dominated by Viburnum sibiricum and Phyllanthus thunbergii.

We first calculated the importance value for each tree species and selected the important value ranked in the top 20 for the latter analyses. The importance value is defined as follows: [(relative abundance + relative basal area + relative frequency)/3] (

Table 1. Basic information of the top 20 dominant species in importance value.

Species Name	Family Name	Abundance	Basal Area (m2/ha)	Maximum DBH (cm)	Importance Value	Life Form
Quercus aliena var. acutiserrata	Fagaceae	2141	136.12	105.92	0.0749	D
Acer oliverianum	Sapindaceae	7556	109.66	51.80	0.0742	D
Fagus engleriana	Fagaceae	2345	95.90	83.50	0.0663	D
Betula albosinensis	Betulaceae	1670	50.56	102.23	0.0571	D
Lindera obtusiloba	Lauraceae	2552	50.50	59.82	0.0455	D
Acer stachyophyllum subsp. betulifolium	Sapindaceae	3284	31.80	21.22	0.0310	D
Viburnum dilatatum	Adoxaceae	3089	22.93	14.45	0.0277	D
Sorbus alnifolia	Rosaceae	1951	19.01	42.36	0.0248	D
Toxicodendron vernicifluum	Anacardiaceae	924	17.27	53.60	0.0247	D
Litsea ichangensis	Lauraceae	2075	15.08	48.11	0.0223	D
Acer flabellatum	Sapindaceae	1700	13.45	52.56	0.0211	D
Corylus ferox var. thibetica	Betulaceae	1376	12.97	49.31	0.0190	D
Acer pictum subsp. mono	Sapindaceae	1679	12.93	47.80	0.0190	D
Styrax hemsleyanus	Styracaceae	1208	12.22	47.10	0.0183	D
Sorbus hemsleyi	Rosaceae	805	11.89	92.99	0.0176	D
Meliosma veitchiorum	Sabiaceae	801	11.50	97.50	0.0170	D
Pinus armandii	Pinaceae	972	10.66	78.47	0.0166	E
Acer maximowiczii	Sapindaceae	1234	10.48	32.12	0.0164	D
Litsea veitchiana	Lauraceae	1443	10.49	20.11	0.0163	D
Cornus controversa	Cornaceae	695	9.66	49.40	0.0158	D

(D: deciduous species; E: evergreen species. C: canopy layer; U: sub-tree layer; S: shrub layer. The same is shown below).

). The relative density calculation did not include the number of branches and only used the number of individual plants. The unit area of the quadrat was 20 m × 20 m for relative frequency. Relative dominance was calculated as the sum of the thoracic height basal area of independent plants and branches [25]. The forest was divided into three vertical layers: the canopy layer (≥15 m), the understory layer (≥5 m and <15 m), and the shrub layer (<5 m) [26]. For each selected species, we classified different life-type stages based on the height and growth type of the species. Canopy layer: 1 cm ≤ DBH < 5 cm (saplings), 5 cm ≤ DBH < 20 cm (juveniles), DBH ≥ 20 cm (adults); understory layer: 1 cm ≤ DBH < 5 cm (saplings), 5 cm ≤ DBH < 10 cm (juveniles), DBH ≥ 10 cm (adults); and shrub layer: 1 cm ≤ DBH < 2 cm (saplings), 2 cm ≤ DBH < 3 cm (juveniles), DBH ≥ 3 cm (adults). Then, using the diameter class method, we analyzed the spatial distribution pattern of dominant species within the community. The spatial analysis was less reliable in cases where the number of individuals was low. To meet the large sample size required for accurate point pattern analysis, life stages with fewer than 200 individuals were excluded. Finally, 6 dominant species with maximal importance values were taken into consideration in intraspecific and interspecific association analyses. The number of individuals at different life-history stages of the analyzed species is shown in

Table 2. Dominant species with more than 200 individuals in each life stage.

Species Name	Layer	Abundance	Individuals of Small Trees	Individuals of Medium Trees	Individuals of Large Trees
Quercus aliena var. acutiserrata	Canopy	2141	222	718	1201
Acer oliverianum	Understory	7556	4613	1673	1270
Fagus engleriana	Canopy	2345	824	899	622
Lindera obtusiloba	Understory	2552	368	617	1567
Viburnum dilatatum	Shrub	3089	1874	384	231
Acer flabellatum	Understory	1700	1135	337	228

.

2.3. Data Analysis

2.3.1. Spatial Distribution Pattern

Point pattern analysis offers a valuable understanding of the spatial distribution of communities by analyzing spatial patterns at various sizes [27]. Here, we applied a K function to analyze the distribution patterns for the 20 dominant species at different scales. The K function is derived from the g(r) function but is more sensitive than the K function in determining how much the points at a given scale depart from the predicted values [19]. The univariate pairwise correlation g(r) function is calculated as follows:

$$g\left(r\right)={\displaystyle \frac{1}{2\pi }}·{\displaystyle \frac{dK\left(r\right)}{dr}}$$

where r is the spatial scale (m), the K(r) function is the ratio of the expected number of points compared to the density of sample points in a circle with any point in the study area as its center, and r is the radius [28,29]. If g(r) > 1, it indicates a clustered distribution. If g(r) < 1, it indicates a regular distribution. If g(r) = 1, it indicates a completely random distribution [19]. We chose the complete spatial randomness (CSR) and heterogeneous Poisson (HP) models as the null models [30]. The CSR model is often used as the null hypothesis to examine the effect of habitat heterogeneity on spatial distribution, which assumes that the spatial points of the species are independent of each other and not affected by any abiotic or biotic processes. It also assumes that the species have the same probability of occurring at each point in the study area [31]. The heterogeneity Poisson model is a null hypothesis model that excludes the effect of spatial heterogeneity, which can accurately represent the population’s real spatial distribution characteristics.

2.3.2. Intraspecific and Interspecific Associations

Bivariate pairwise correlation functions (PCFs) were applied to determine interspecific relationships among the dominant species and intraspecific associations of the six selected dominant species at different life-history stages. The PCF can be calculated as follows:

$$g_{12}\left(r\right)={\displaystyle \frac{1}{2\pi }}·{\displaystyle \frac{{dK}_{12}\left(r\right)}{dr}}$$

$$K_{12}\left(r\right)=({\displaystyle \frac{A}{n_1{n}_2}}){\displaystyle \sum _{i=1}^{n_1}{\displaystyle \sum _{j=1}^{n_2}\frac{1}{W_{ij}}}}{I}_r(t_{ij}) i\ne j$$

Here, i and j represent two distinct populations, while n₁ and n₂ denote the total number of surviving individuals in populations i and j, respectively. When g₁₂(r) > 1, it indicates a positive association between the two populations; g₁₂(r) = 1 suggests no significant association; and g₁₂(r) < 1 reflects a negative association.

The antecedent condition (AC) and the CSR were selected as null models; the AC null model assumes that the positions of those with smaller DBHs are randomized, while those with bigger DBHs remain unchanged [16]. Due to the fact that the life-history stages of tree species are sequential but not realized simultaneously, we first used the AC null model to fix the positions of individuals with larger DBHs and then analyzed the association between individuals with smaller and larger DBHs.

To assess the accuracy of the empirical function, we performed a 95% simulation envelope with maximum and lowest values and ran 199 Monte Carlo simulations of the null model for all spatial pattern analyses [32]. These analyses were performed using the “spatstat” package in R 4.2.2 [33].

3. Results

3.1. Spatial Distribution Pattern

Based on the CSR null model, the results of the univariate g(r) function showed that the 20 dominant tree species had an aggregated spatial pattern across the test area (r < 100 m). As the scale increased (from 0 to 100 m), the degree of aggregation for most species (16 in 20) decreased (Figure 2). However, there were four species, Viburnum dilatatum, Toxicodendron vernicifluum, Styrax hemsleyanus, and Meliosma veitchiorum, that showed distribution patterns other than aggregated distributions (Figure 3a). The actual discrete distribution of each tree species in the SNJ plot was generally consistent with the results of the spatial distribution pattern of tree species under CSR (Figure 2). Therefore, the spatial distribution pattern of dominant tree species is mostly found in the aggregate distribution when other contributing factors are not taken into account. In order to further obtain a rough estimate of the large-scale impact of habitat heterogeneity on local trees, we examined the distribution of all 20 species and compared it with the HP null model.

The HP null model showed that once the influence of environmental heterogeneity was removed, the distribution pattern of different tree species underwent changes (Figure 3b). Compared to the CSR model, the degree of dominant species aggregation was lower in the HP null models. In total, 80% (Sixteen) of tree species showed new patterns of spatial distribution after controlling environmental heterogeneity (Figure 3b). Specifically, most dominating species (except for Acer oliverianum) showed similar individual distribution patterns, where trees were aggregated at small scales and then distributed equally or randomly as the scale increased. At scales greater than 35 m, the distribution pattern changed to a random distribution after removing environmental heterogeneity, although Lindera obtusiloba, Sorbus alnifolia, and Litsea veitchiana all showed aggregated distribution across the scale. With the exception of these, all sixteen species showed three distribution patterns in the HP null model, which differed from the CSR model.

3.2. Intraspecific and Interspecific Associations

3.2.1. Intraspecific Patterns from Saplings to Adults

We used the bivariate pairwise correlation function (g₁₂ (r) function) to analyze the spatial correlation of six dominant species in different life-history stages (Figure 4) based on the null models of CSR and AC.

Under CSR, Acer flabellatum showed positive associations across different life-history stages (Figure 4a). There were also positive associations at all scales between the small and medium trees of Quercus aliena, Acer oliverianum, and Fagus engleriana, and between the medium and large trees of Q. aliena and L. obtusiloba. Furthermore, a positive correlation was observed among the three life stages of V. dilatatum on a smaller scale (<50 m). However, this correlation disappeared on a larger scale. After removing the effect of environmental heterogeneity, the results of the AC null model showed no associations between small and medium or small and large trees in the tree layers of Q. aliena and A. oliverianum at any scale (Figure 4b). In detail, the life-history stages of medium and large trees of Q. aliena and A. flabellatum showed positive associations at the small scale, which shifted to non-significant associations as the scale increased. A shrub species, V. dilatatum, only showed positive associations across three stages on scales less than 15 m, and it turned out to be insignificant or negative with the scale increased. On a larger scale, small trees and large trees of V. dilatatum were mainly negatively correlated.

3.2.2. Interspecific Association

We employed the bivariate pair correlation function (g₁₂(r) function) again to analyze the spatial correlation among the six dominant tree species (Figure 5). As shown in Figure 5, Q. aliena and F. engleriana, Q. aliena and V. dilatatum, F. engleriana, and L. obtusiloba showed attraction patterns at all scales under the CSR null model, while Q. aliena and A. oliverianum, L. obtusiloba, and A. flabellatum showed repulsion patterns at all scales. Meanwhile, Q. aliena and L. obtusiloba; A. oliverianum and V. dilatatum; L. obtusiloba and V. dilatatum; V. dilatatum and A. flabellatum were positively correlated at small scales, followed by no or negative correlations as the scales increased. The trees of Q. aliena were positively associated with small trees of A. flabellatum at scales of less than 80 m, although there was no significant correlation between them at some scales. The spatial correlation between the tree species A. flabellatum and F. engleriana was similar to that of these two species. On scales of less than 5 m, there was no significant correlation between A. oliverianum and A. flabellatum, while on scales exceeding 5 m, there was a negative correlation. Because all p-values of the goodness-fit test were less than 0.05, our results at the community level were statistically over the range of distances we studied (0–100 m).

However, the analysis results under the AC showed that the correlation among tree species changed on a large scale, mainly showing a lack of associations after excluding the influence of environmental heterogeneity (Figure 5). Compared to the CSR null model, the associations between different species were different on most scales, which was reflected in the irrelevance found between species on most scales. Specifically, at a scale of more than 50%, the spatial correlation was non-associated except for A. oliverianum and V. dilatatum. In contrast to the predominant lack of correlation, A. oliverianum and V. dilatatum showed a positive correlation at the 0–33, 35–36, 38–39, and 43–45 m scales, and F. engleriana and A. flabellatum showed a negative correlation at the 0–46 and 49–53 m scales. These results indicate that environmental heterogeneity has a substantial impact on the relationships among dominant species.

4. Discussion

4.1. Spatial Distributions of Dominant Species

In this study, we explored the spatial patterns of tree species under two models—CSR (completely spatially random) and HP (Habitat Patchiness)—to test the applicability of the Janzen–Connell hypothesis, which posits that seedling survival decreases with their proximity to parent trees due to density-dependent mortality and herbivory [12,13]. The CSR model revealed that all species showed clustered patterns over a broad spatial range, possibly because of restricted seed dispersal and habitat preferences. However, it should be considered that habitat heterogeneity in the HP model led to a narrowed aggregate distribution spectrum and the transition of spatial layouts to a more uniform or random pattern, which then became the dominant pattern. This shift highlights the role of habitat patches in shaping species distribution patterns [11,34]. This research aligns with earlier works indicating that species typically group at smaller scales and show either random or consistent distribution over larger scales [11]. The Janzen–Connell theory offers a credible explanation for the noted trends. Within the smaller scope (0–20 m), the 20 dominant species in our research demonstrated clustered dispersal, mainly attributed to the constraints of dispersal [35,36]. For instance, Q. aliena var. acuteserrata (a nut species) has large, heavy seeds that are dispersed mainly by gravity, resulting in high seedling density around parent trees. In contrast, A. pentaphyllum (a samara species) can disperse its seeds over longer distances due to its winged morphology, allowing seedlings to establish in suitable environments away from the parent tree. It is important to note that despite the larger number of seeds around the parent tree, seed predation affects those closer to the seed tree more heavily, which increases the survival rate of seeds located further away. Additionally, animal dispersal plays a fundamental role in the distribution and survival of seedlings as they are moved away from the seed tree. With the expansion in scale, aggregation levels diminish, which is in line with the Janzen–Connell theory, suggesting an exponential reduction in seedling density as the distance from the originating tree grows [15].

In addition to dispersal limitations, habitat heterogeneity also plays a crucial role in shaping species distribution patterns at larger scales [11,37]. The results from the HP model revealed that small parts of habitats have a notable effect on the distribution of species. For example, Q. aliena var. acuteserrata prefers well-drained, moist, neutral to slightly acidic soil, while F. engleriana thrives in fertile, acidic soils with cool and humid climates. A. oliverianum is adapted to sunny, drought-tolerant environments, often found at forest edges or in sunny gaps. Pinus armandii can tolerate a variety of soil types but is sensitive to high temperatures. Betula albosinensis, S. alnifolia, and T. vernicifluum mainly aggregate on sunny slopes, showing that clustering density increases with the terrain (slope position, slope direction, and slope gradient). These habitat preferences lead to aggregated distributions in locally suitable environments, further supporting the role of habitat heterogeneity in shaping species distribution patterns [38,39].

The aggregated distribution at small scales may also be related to the high niche overlap among the dominant species, especially among small and medium-sized individuals. This aggregation can enhance population-level effects, increase interspecific competition, and ensure long-term population persistence [40,41]. For example, the canopy layer species showed higher aggregation at larger scales, likely because species with higher population densities tend to cluster more strongly (Figure 2). In summary, the spatial distribution patterns of tree species in natural plant communities are influenced by both habitat heterogeneity and dispersal limitations. Future research should further explore the mechanisms underlying these patterns and their implications for community dynamics and conservation.

4.2. Associations Across Species and Life-History Stages

In this study, the six dominant species generated 15 groups of species pairing patterns, which were dominated by positive and negative correlations under the CSR model and by non-correlation under the AC model. These findings are inconsistent with those in the karst secondary forest in central Guizhou, where the interrelationships among the four dominant species mainly showed negative or no correlation. This discrepancy may be related to differences in life forms, climate types, and topography among the dominant species. These results can be interpreted through the lens of the Species Herd Protection hypothesis [37,42], which suggests that species interactions are influenced by both resource competition and mutual benefits.

Positive correlations between species indicate a high similarity in resource use and niche overlap. The stronger the positive correlation, the greater the complementarity between populations within the community, leading to more efficient resource utilization and enhanced community stability. Under the CSR model, eight species pairing patterns in this study showed positive correlations across most scales. This result may be attributed to the high niche overlap among these species. For example, Q. aliena var. acuteserrata exhibited positive correlations with several other species, suggesting that they share similar resource requirements but may benefit from mutual facilitation. This finding aligns with the Species Herd Protection hypothesis.

Negative correlations between species indicate niche space isolation and competition. In this study, six species pairs mainly exhibited negative correlations under the CSR model. For example, Q. aliena var. acuteserrata had a negative correlation with A. oliverianum and A. flabellatum. These species are heliophilous plants, suggesting strong competition for light and other resources. Spatial separation may be a strategy to reduce competitive consumption. Notably, A. flabellatum showed strong negative correlations with four dominant species, except for V. dilatatum, which was primarily concentrated in the high-altitude northern area of the plot. Resource competition is likely the main reason for this exclusionary effect. Given the limited space and resources, plants compete for light, water, nutrients, and space, resulting in mutual exclusion. This finding supports the idea that competition drives spatial segregation among species, as predicted by the Species Herd Protection hypothesis. Non-correlations between species indicate a lack of interactions, which may result from spatial isolation or environmental heterogeneity. Under the AC model, non-correlation accounted for a larger scale range, suggesting that after accounting for environmental heterogeneity, interspecific associations were independent of each other across broader scales. This finding suggests that the spatial separation of species on a larger scale can hinder interactions between two species. Additionally, aggregation within species can lead to large-scale spatial separation between species. This pattern may explain the non-correlations observed in our study.

5. Conclusions

This study focused on the spatial distribution patterns and intraspecific and interspecific associations of dominant tree species in the deciduous broad-leaved forest of the mid-subtropical mountain in Shennongjia, China. The results show that 20 dominant tree species are clustered on a small scale, and with the increase in scale, the degree of aggregation gradually decreases and changes to random or uniform distribution. The spatial distribution of dominant tree species is mainly affected by environmental heterogeneity and diffusion restrictions. The intraspecific association was mainly a positive correlation, indicating that the population was well connected. Interspecific associations are mainly influenced by the biological characteristics and environmental heterogeneity of each species. These findings confirm that habitat heterogeneity and dispersal limitations are the two primary drivers of species spatial distributions and potentially shape species coexistence in the subtropical deciduous broad-leaved forest. The spatial distribution pattern is the first step to understanding the plant community structure, revealing the mechanism of species coexistence, and showcasing the influence that the intensity of different factors has on species coexistence, requiring further investigation and research.