The Spatial Pattern of the Tertiary Relict Plant Tetracentron sinense Oliver and Its Influencing Factors

Abstract

Tertiary relict plants are of great scientific value in the study of flora evolution, angiosperm systems, and ancient origins. Paying attention to their spatial patterns can better reflect the change dynamics of the species to implement targeted protection countermeasures. In this study, we investigated the spatial patterns of Tetracentron sinense Oliver, a tertiary relict plant, and further studied the intra- and interspecific and environmental factors impacting the patterns. The results reveal that most of the individuals of T. sinense were distributed in the 1700–1800 m altitudinal belt, and they were highly aggregated at a small scale. The young trees showed a positive interaction with adult trees. The dominant species showed a positive interaction with T. sinense; the interaction gradually became non-significant or negative as the scale increased. The key abiotic factors affecting the distribution of T. sinense were altitude, litter depth, zinc, and calcium. These results highlight the intra- and interspecific interactions and environmental factors influencing the spatial pattern of T. sinense. Our results provide new insights into tertiary relict species’ spatial patterns and nearline factors. Moreover, these findings have relevant implications for conserving and managing tertiary relict plants in a constantly fragmented habitat.

1. Introduction

A spatial pattern is the position of an individual in a spatial range and its state of dispersion [1]. Understanding the survival status and environmental requirements of a species needs a comprehensive understanding of its spatial pattern [2,3]. The spatial pattern is ultimately generated by the interaction of its individuals with their surroundings [2]. Spatial patterns are usually categorized into three types: random, regular, and aggregated distributions [4]. Different distribution patterns reflect the ability and status of a population to utilize environmental resources, as well as the status and viability of the population within the community [4]. Since the 1920s, the spatial patterns of plants have become a hot topic in ecological research [5,6,7,8,9,10,11,12,13,14]. To overcome the influence of the sampling area in the traditional spatial pattern analysis method [13] and provide comprehensive information for spatial scales [13], point pattern analysis has become a common method for studying the spatial patterns of species in recent years [7,8,9,10,11]. However, many studies have only focused on the spatial patterns of populations at large scales [6,7,11,13,14]. The spatial pattern of a population is closely related to the spatial scale [15]. Studying a spatial pattern at a small scale can further reveal the survival status of the population, which is of great significance for the study of plant conservation, especially tertiary relict plant conservation [16].

Previous studies have focused on the influence of intra- and interspecific relationships on community development, but little study has been conducted on the effects of these factors on spatial patterns [9,11,17]. By studying spatial patterns and their influence on biotic and abiotic factors, we can gain a deeper understanding of the formation process and influence mechanisms [17].

It is widely recognized that the Arcto-Tertiary flora originated in the high northern latitudes of the Cretaceous and Paleocene. In the Oligocene, it was forced to migrate southward to the mid-latitudes due to global cooling [18,19]. Among these regions, East Asia retains the largest number of representatives of the early Arcto-Tertiary flora, especially in the Mesozoic mixed forests of Japan and China. A large amount of fossil material proves that Tetracentron sinense Oliver is one of the tertiary relict plants currently restricted to East Asia [19]. It is a deciduous tree in the Trochodendracea family and is mainly distributed in the temperate and subtropical regions of East Asia. It plays an important role in understanding the evolution of paleophyte flora, the development of systematics, and the origin of angiosperms [20]. As a result of over-exploitation by humans and global climate change, the habitat of T. sinense has been destroyed, and its natural populations are getting smaller and smaller, showing patchy distribution. It is now listed as an endangered species in Appendix III of the Convention on International Trade of Endangered Species (CITES) [19,21,22]. Therefore, scholars at home and abroad need to clarify its endangered status and protect its germplasm resources. To conserve the germplasm resources of T. sinense, much research has been carried out on its survival status [23,24,25], seed production and dispersal [26], seedling and young tree establishment [27], and genetic diversity [19,21], and the mechanism of limiting regeneration has also been studied. To date, little is known about the spatial pattern and influencing factors of the distribution of T. sinense populations. Xu (2016) found that the effective dispersal distance of T. sinense seeds does not exceed 8 m, and younger individuals usually grow around parent trees, resulting in a patchy relic distribution pattern [28]. Tian et al. (2018) [24] posited that there was a strong niche overlap between T. sinense and its dominant trees, indicating relatively strong intra- and interspecific competition. Does the spatial pattern of T. sinense present aggregation at small scales similar to its seed distribution pattern? Do intra- and interspecific correlations and environmental factors have an impact on its spatial pattern and result in poor regeneration of T. sinense? Answering these questions can help reveal the factors limiting the natural regeneration of T. sinense, which is of great significance for the effective conservation and management of this species.

In this paper, we studied the spatial pattern and influence factors of T. sinense in Leigong Mountain. The aims were (1) to reveal the spatial patterns of T. sinense at different spatial scales, and (2) to discuss the impacts of intra- and interspecific interactions and environmental factors on the distribution of T. sinense. We also discussed the effective strategies for the conservation and management of T. sinense.

2. Materials and Methods

2.1. Study Area and Species

This study was conducted in Leigong Mountain in Guizhou Province, China. The region is characterized by a typical subtropical humid climate. The average annual temperature is approximately 9.2–16.3 °C, the average annual humidity is approximately 85%–91%, and the annual precipitation is approximately 1300–1600 mm. The vertical spectrum of vegetation belts on the Leigong Mountain is defined as evergreen broadleaved forest (approximately 1350 m a.s.l.), evergreen and deciduous broadleaved mixed forest (1350–2100 m a.s.l.), and shrubs (above 2100 m a.s.l.). T. sinense is mainly distributed in evergreen and deciduous broadleaved mixed forests with Fagus longipetiolata Seemen, Prunus tomentosa (Thunb.) Wall., Pterostyrax psilophyllus Diels ex Perkins, Styrax japonicus Siebold and Zucc., Acer sinense Pax, and Chimonobambusa angustifolia C.D. Chu and C.S. Chao [25,29,30,31].

T. Sinense is a tertiary relict plant and has been documented as an endangered species in China [32]. The natural populations of T. sinense are only sporadically distributed in damp valleys in southwestern and central China at an altitude of approximately 1100–3500 m [33]. T. sinense is a monoecious, hermaphroditic, and parthenogenetic plant. It is capable of cross-pollination via wind and insects. It flowers in June–July. Since its flowering period usually coincides with the rainy season, it ensures reproduction via self-fertilization [21]. T. sinense can produce a large number of fertile seeds after pollination, and its seeds are mainly wind-borne [26].

2.2. Data and Sample Collection

A comprehensive survey was performed between early April 2018 and late June 2020. It showed that T. sinense is patchily distributed in the gullies of Leigong Mountain, and the patches are divided by ridges [21,25,34] (Supplementary Figure S1). In this study, four patches of T. sinense with a relatively large number of individuals, complete age structure, and similar abiotic and biotic conditions were selected to study spatial distribution patterns and their influencing factors. Plots were set up in these four patches, and the plots encompassed all T. sinense individuals in each patch: Plot 1 (P1): 50 m × 140 m, Plot 2 (P2): 128 m × 150 m, Plot 3 (P3): 50 m × 148 m, and Plot 4 (P4): 60 m × 120 m, with a total area of 40,800 m² (Figure 1 and Supplementary Table S1). The plots were established following guidelines from the Center for Tropical Forest Science [35], and the altitude, slope gradient, slope aspect, litter depth, and humidity were recorded for each [36,37,38]. Each of the four plots was further subdivided into 10 m × 10 m subplots. All trees with a diameter at breast height (DBH) of ≥1 cm in each subplot were recorded, tagged, and measured [37,38]. The southwest corner of the plot was used as the origin, and the height of the trees, the height under the branches, and the crown width were determined with an electronic total station, and the coordinates of the trees were recorded [36,38]. According to the important value [(relative density + relative frequency + frequency of prominence) × 100] [14], we selected four dominant species (A. sinense, P. tomentosa, S. japonicus, and P. psilophyllus) with an importance value over 55% (Supplementary Table S2).

We used the diameter class method instead of age class and divided individuals of T. sinense into 10 age classes [25,36]: I (seedling), H < 0.33 m; II (sapling), H > 0.33 m, DBH < 2.5 cm; III, 2.5 ≤ DBH < 7.5 cm; IV, 7.5 ≤ DBH < 12.5 cm; V, 12.5 ≤ DBH < 17.5 cm; VI, 17.5 ≤ DBH < 22.5 cm; VII, 22.5 ≤ DBH < 27.5 cm; VIII, 27.5 ≤ DBH < 32.5 cm; IX, 32.5 ≤ DBH < 37.5 cm; and X, DBH ≥ 37.5 cm. For the analysis of intraspecific interactions, T. sinense trees in the four plots were classified into three life-history stages: young trees (I–III), adult trees (IV–VI), and old trees (VII–X).

Soil samples were collected in each subplot using the five-point method [24,37]. The upper layer of the litter and gravel was removed, and the soil samples were collected at a depth of approximately 0–20 cm. Approximately 1 kg per subplot was sampled, and a total of 234 loam soil samples were collected. The soil samples were transported back to the laboratory and then dried naturally [39,40,41]. After drying, the soil samples were ground and analyzed to determine their mineral elements. The total contents of nitrogen (N) and phosphorus (P) were determined with a UV spectrophotometer [40,41]. Other mineral elements, such as potassium (K), sodium (Na), magnesium (Mg), zinc (Zn), iron (Fe), manganese (Mn), copper (Cu), and calcium (Ca), were determined with a flame photometer and atomic spectrophotometer [40,41]. After correlation analysis of the contents and principal components, the seven mineral elements with high correlations and the highest contribution rates (N, P, K, Ca, Mg, Zn, and Na) were selected [40,41].

2.3. Data Analysis

2.3.1. Point Pattern Analysis

The spatial patterns and spatial interactions (intra-and interspecific interactions) of T. sinense were analyzed using the O-ring function for point pattern analysis, which was performed with Programita [42,43,44,45,46,47,48]. The approach was supplemented with Monte-Carlo random simulation. The univariate function (O(r)) was used to analyze the spatial pattern of T. sinense in the four plots, and the bivariate function (O12(r)) was used to analyze the spatial interactions between different age classes and between T. sinense and dominant species in the same plot. The screening of the effective null hypothesis model was the key to accurately analyzing the distribution of populations and the relationships between them. Complete spatial randomness (CSR) was chosen to perform the analysis [43]. If the O(r) value is within the upper and lower wrap traces of the confidence intervals, the distribution is random or the two are directly independent of each other. If the O(r) is above the upper and lower wrap traces of the confidence interval, the distribution is aggregated or there is a significant positive interaction between the two [48]. If the O(r) is below the upper and lower wrap traces of the confidence interval, the distribution is uniform or there is a significant negative interaction between the two species [30]. The O-ring function for univariate statistics is a spatial interaction statistic by hypothesizing the pattern analysis of a single variable to a bivariate having two identical patterns. The bivariate O-ring statistic is calculated as follows [38,43,44]:

$${\widehat{O}}_{12}^w\left(r\right)=g_{12}\left(r\right){\lambda }_2=\frac{\frac{1}{n_1}{\Sigma }_{i=1}^{n_1}Point{s}_2\left[R_{1,\dot{i}}^w\left(r\right)\right]}{\frac{1}{n_1}{\sum }_{i=1}^{n_1}Area\left[R_{1,i}^w\left(r\right)\right]}$$

(1)

where n₁ is the number of individuals in pattern 1, $R_{1,i}^w\left(r\right)$ is the circle with i as the center, r is the radius, and w is the width in pattern 1. $Point{s}_2\left[x\right]$ is the number of points of pattern 2 in region X. $Area\left[x\right]$ is the area size.

$$Point{s}_2^{\left[R_{I,i\left(r\right)}\right]}={\sum }_{allx}{\sum }_{ally}S\left(x,y\right)P_2(x,y)I_{r }(x_i,y_i,x,y)$$

(2)

where (x_i, y_i) are the coordinates of pattern 1, S(x, y) is a variable, P₂(x, y) is the number of points in pattern 2 per grid, I_r is the amount of change that occurs when changing the circle with the point (x_i, y_i) as the center, and r is the radius of the circle in pattern 1.

$$I_r\left(x_i,y_i,x,y\right)=\left\{\begin{array}{c}1 if\\ 0\end{array}\right.\sqrt{{(x-x_i)}^2+{(y-y_i)}^2}\le 2$$

(3)

$$Area[R_{I,i\left(r\right)}]=Z^2\sum _{allx}\sum _{ally}S(x,y)I_r(x_i,y_i,x,y)$$

(4)

where Z² is the size of a grid. Similarly, the univariate O-ring statistic is computed by setting pattern 1 equal to pattern 2.

For all analyses, the Monte Carlo simulation was repeated 199 times to yield 99% confidence for each process with the corresponding null model. The spatial scales of P1, P2, P3, and P4 were 0–25 m, 0–64 m, 0–25 m, and 0–30 m, respectively, with a step size of 0.5 m.

2.3.2. Redundancy Analysis (RDA)

Canoco 5.0 was used in the RDA to analyze the key factors influencing the spatial pattern of T. sinense. Detrended correspondence analysis (DCA) revealed that the longest axis for the data was less than 3 (=2.4, 1.7, 1.5, 1.5). Before applying the RDA, we selected the environmental variables that were relatively high correlation in the species data using the Monte Carlo technique and Adonis test [49].

Species richness considers the number of species. The heat maps were obtained using Origin, and the other statistical analyses were performed using SPSS V29, Excel V2019, etc.

3. Results

3.1. Spatial Pattern

The number of T. sinense individuals was 155 in the four plots, including 33 for P1, 51 for P2, 34 for P3, and 37 for P4. As is seen in the table, the largest number of individuals were of middle age, and the smallest number of individuals were of young and older ages (Supplementary Table S3).

We divided the altitudes into four altitudinal belts to study the vertical distribution of T. sinense in the four plots: H1 (1700–1800 m), H2 (1800–1900 m), H3 (1900–2000 m), and H4 (2000–2100 m) (Figure 2). The total individuals’ richness showed a bimodal pattern with altitude, with maximum richness at 1700–1800 m and minimum richness at over 2000 m (Figure 2A). The clustered heat maps obtained from the hierarchical classification showed that the low-altitudinal belts (H1 and H2) clustered, and the high-altitudinal belts (H3 and H4) clustered (Figure 2B).

The overall spatial distributions of T. sinense in the four plots were similar, showing significant aggregation at scales of less than 5 m. P1 showed significant aggregation at scales of less than 14 m. As the scale increased, the distribution pattern shifted to a random distribution. In P2, the individuals were aggregated at a scale of 0–34 m, randomly distributed at 34–38 m, and then evenly regularly distributed at scales over 38 m. In P3, the individuals were aggregated at scales of up to 5 m and then randomly distributed with an increasing spatial scale. On the whole, the individuals in P4 were aggregated at a scale of 0–8 m and then randomly distributed at scales greater than 8 m (Figure 3).

3.2. Spatial Interaction

We analyzed the intraspecific interaction between three life histories of T. sinense in four plots. The intraspecific interactions were different among the three life histories of T. sinense. In P1, young and adult trees were positively correlated in the 0–1 m and 12–13 m scale ranges, respectively, while there was no correlation at other scales. There was no obvious correlation between young and old trees within the study scales. The adult and old trees were positively correlated at scales of 0–1 m and 12–13 m, and no correlation was observed at the remaining scales. In P2, young and adult trees were significantly positively correlated at 0–11 m, and little correlation was observed at other scales. Young and old trees showed a certain positive correlation in the ranges of 2.5–7.5 m and 15–16 m; at other scales, they had no interaction. Adult and old trees showed a positive correlation at 5–6 m and did not correlate at increased scales. There was no significant correlation between the other types of trees in P3 except between young and adult trees in the range of 1–2 m. Similarly, in P4 and P3, young trees and adult trees had a significant positive correlation at scales of less than 1.5 m, and no correlation appeared among the other types of trees (Figure 4).

We further tested the interspecific interactions between T. sinense and its dominant species in four plots (Figure 5). In P1, there was no correlation between T. sinense and P. psilophyllus and S. japonicus at the study scales; T. sinense with P. tomentosa showed a negative interaction at a scale of 14–21 m and no correlation at other scales; and T. sinense and A. sinense showed a negative interaction at scales of 2.5–4 m and 7–11 m, and no correlation at other scales (Figure 5, P1). In P2, the interaction between T. sinense and its dominant species changed greatly. P. psilophyllus with T. sinense showed a negative interaction at scales of less than 14 m and a strong positive interaction in the ranges of 24.5–25.5 m and 31–48 m. S. japonicus with T. sinense had a negative interaction at a scale of 1–6 m and had a positive interaction at scales of 11–24 m, 34–36 m, and 39–42 m; T. sinense with P. tomentosa had a positive interaction at scales of 10–14 m and 39–44 m and had no interaction at other scales; and A. sinense showed a negative interaction with T. sinense at scales of less than 5 m and a positive interaction with T. sinense at scales of 10.5–15 m and 39–42 m (Figure 5, P2). In P3, T. sinense with P. psilophyllus and S. japonicus had no interaction; T. sinense with P. tomentosa had a negative interaction at a scale of 4–7.5 m and had no interaction at other scales; and A. sinense showed no interaction with T. sinense at any scale, except for a negative interaction at a scale of 17.5–21.5 m (Figure 5, P3). In P4, T. sinense with P. psilophyllus had a negative interaction at scales of more than 25.5 m and had no interactions at other scales; T. sinense with S. japonicus showed a positive interaction at a scale of 9–11 m and had no interactions at other scales; T. sinense with P. tomentosa showed a negative interaction at scales of less than 9 m; and A. sinense showed a positive interaction with T. sinense at a scale of 2.5–4.5 m and had no interaction at other scales (Figure 5, P4).

3.3. Correlation Analysis of Key Factors

We conducted Pearson correlation analyses between environmental factors and T. sinense richness in four plots (Supplementary Figure S2). In P1, the humidity and altitude showed a notable correlation with the richness of T. sinense. A higher humidity or altitude contributed less to T. sinense. In P2, the litter depth, altitude, and Ca had great influences on the richness of T. sinense but did not reach a significant level. In P3, N and P had a significant effect on the richness of T. sinense, which decreased with an increase in the two. In P4, altitude, Mg, and Zn had a significant effect on the richness of T. sinense, which decreased with an increase in the three.

The Monte Carlo technique selected key factors in four plots that affected the distribution of T. sinense. In P1, the key environmental factors were altitude, humidity, Na, shade density, litter depth, and K. In P2, the key environmental factors were litter depth, altitude, humidity, shade density, N, Na, and Mg. In P3, the key environmental factors were shade density, altitude, litter depth, humidity, K, Ca, and Zn. In P4, the key environmental factors were altitude, humidity, shader density, Mg, Na, Zn, P, and K (

Table 1. Forward selection results.

Plot	Statistic	Explains%	Contribution%	Pseudo-F	P
P1	Altitude/m	29.2	29.2	4.1	0.022
	Humidity/%	22.1	22.1	4.1	0.028
	Na/(mg/Kg)−1	13.0	13.0	2.9	0.068
	Litter depth/cm	7.1	7.1	1.8	0.216
	K/(g/Kg)−1	4.9	4.9	1.1	0.366
	Shade density/°	3.1	3.1	0.7	0.470
P2	Litter depth/cm	47.6	47.6	6.4	0.064
	N/(g/Kg)−1	11.5	11.5	1.7	0.234
	Na/(mg/Kg)−1	21.0	21.0	5.3	0.078
	Shade density/°	11.7	11.7	5.7	0.028
	Altitude/m	3.8	3.8	2.6	0.118
	Humidity	2.6	2.6	2.8	0.134
	Mg/(mg/Kg)−1	1.5	1.5	3.7	0.228
P3	Ca/(mg/Kg)−1	42.5	42.5	5.2	0.008
	Shade density/°	18.5	18.5	2.8	0.064
	Altitude/m	12.1	12.1	2.2	0.092
	K/(g/Kg)−1	8.2	8.2	1.8	0.22
	Zn/(mg/Kg)−1	4.5	4.5	0.9	0.43
	Litter depth/cm	4.1	4.1	0.8	0.574
	Humidity/%	5.2	5.2	1	0.478
P4	Altitude/m	30.3	30.3	3.5	0.052
	Humidity/%	13.3	13.3	1.6	0.192
	P/(g/Kg)−1	11.9	11.9	1.6	0.210
	Mg/(mg/Kg)−1	11.2	11.2	1.7	0.214
	K/(g/Kg)−1	13.4	13.4	2.7	0.146
	Na/(mg/Kg)−1	8.8	8.8	2.4	0.132
	Zn/(mg/Kg)−1	6.9	6.9	3.3	0.106
	Shade density/°	3.1	3.1	2.9	0.306

).

According to the RDA, the first two axes could explain the relationships between the species and environment in P1, P2, P3, and P4 (Figure 6;

Table 2. RDA ordination summary of plots.

Plot	Statistic	Axis 1	Axis 2	Axis 3	Axis 4
P1	Eigenvalues	0.4066	0.2775	0.1091	0.1626
	Explained variation (cumulative)	40.66	68.41	79.32	95.58
	Pseudo-canonical correlation	0.8519	0.9418	0.9272	0
P2	Eigenvalues	0.9164	0.0564	0.0182	0.0048
	Explained variation (cumulative)	91.64	97.28	99.1	99.58
	Pseudo-canonical correlation	0.9999	0.9961	0.9779	0.8098
P3	Eigenvalues	0.5461	0.1793	0.1288	0.0799
	Explained variation (cumulative)	54.61	72.54	85.42	93.41
	Pseudo-canonical correlation	0.9735	0.9729	0.9704	0.9996
P4	Eigenvalues	0.6821	0.1967	0.0836	0.0212
	Explained variation (cumulative)	68.21	87.88	96.24	98.36
	Pseudo-canonical correlation	0.9986	0.9923	0.9811	0.9783

). In P1, the distribution of T. sinense was positively correlated with Na and significantly negatively correlated with altitude, humidity, A. sinense, and S. japonicus. In P2, the distribution of T. sinense was significantly positively correlated with shade density, Na, S. japonicus, and P. psilophyllus. It was negatively correlated with litter depth. In P3, the distribution of T. sinense was negatively correlated with shade density, altitude, A. sinense, and P. tomentosa and positively correlated with Ca. In P4, the distribution of T. sinense was positively correlated with K and A. sinense and negatively correlated with altitude, Zn, and P. tomentosa (Figure 6). The results reveal that altitude, humidity, shade density, litter depth, Na, Zn, and K were the key abiotic factors affecting the distribution of T. sinense in the four plots. A. sinense, P. tomentosa, S. japonicus, and P. psilophyllus also affected the distribution of T. sinense in the four plots.

4. Discussion

4.1. Spatial Pattern

In this study, the age structure of T. sinense in the four plots was generally similar, with missing seedlings, more adult trees, and fewer old trees, which is consistent with the results of Zhang et al. (2020) [25]. Spatial patterns are an important means of studying the different structures of communities and the interactions within and between populations and between species and the environment [45,46,47,48,49,50]. If the study scale is small, the spatial pattern of the population is mainly affected by its seed dispersal limitation, intraspecific interactions, interspecific interactions, etc. As the study scale increases, the distribution pattern is mainly determined by external environmental (e.g., altitude, soil, etc.) factors [51]. In this study, we observed that T. sinense tended to initially increase and then decrease with altitude. In the vertical distribution, T. sinense individuals distributed at the altitudinal belts of 1700–1800 m and were least distributed at higher altitudinal belts (over 2000 m). But at smaller scales, the spatial pattern of T. sinense was highly aggregated. A regular distribution and random distribution were alternately observed with increasing scale. The highly aggregated distribution at a small scale (<5 m) might be attributed to the close effective dispersal distance of T. sinense seeds, which is consistent with the results reported by Xu (2016); Wang (2017); and Kong et al. (2021) [28,50,52]. As the scale increased, the distribution of T. sinense became regular or random, which might be the result of gradually decreasing intraspecific interactions. The effect of environmental factors on the spatial distribution of the trees became more significant [52,53].

4.2. Spatial Interactions

Recent studies on species coexistence suggest that conspecific negative density dependence is an important mechanism for regulating plant populations [54,55]. The spatial interaction between different growth stages of a species can characterize the interrelationships of individuals within a population over a certain period, which can be used to reflect the stability of the population and the interrelationships within the population and to further predict the population dynamics [55,56,57,58]. The competitive adaptive ability of different diameter classes of T. sinense may lead to its different spatial patterns. By examining the intraspecific interactions of T. sinense, we found that adult and old trees were relatively independent at a small scale and then tended to have a negative interaction at increasing scales, while young and adult trees had a strong positive interaction at a certain spatial scale. These results are the same as those for Pinus koraiensis Siebold and Zucc. and Tilia amurensis Rupr. [57]. The significant positive interaction between young and adult trees indicated that individuals with a larger DBH supported the survival of individuals with a smaller DBH by changing the niche within the plot. Two individuals with adjacent diameter sizes showing interdependence could act synergistically when utilizing habitat resources and competing with other species within the community. As the study scales increased, the relationship between adult and old trees tended to be negative, which might be attributed to the number of adult trees being far greater than that of old trees [25], resulting in fierce competition for space, essential nutrients, and other external conditions, which is consistent with the results reported by Yang et al. (2021) [5]. The spatial interaction between the adjacent diameter sizes of T. sinense changed from a positive to a negative interaction, predicting stronger competition between individuals, which was not conducive to habitat adaptation. Moreover, this changing trend reflected the limited survival and development abilities of T. sinense, indicating that the natural population of this species on Leigong Mountain is in decline, which is consistent with the results reported by Zhang et al. (2020) [25].

The spatial interactions between different species in a community reflect the relationship between the adaptation of external sub-elements to the environment, with a positive interaction indicating that the species can utilize the same resources and that ecological niches overlap, while a negative interaction suggests that there is a competitive relationship between different species [57]. An analysis of the spatial distribution relationships between companion species and T. sinense populations in the four plots revealed that there were differences in the interactions between T. sinense and its dominant species at different scales. At the study scale, dominant plants in the four plots showed some degree of competition with T. sinense. The distribution of T. sinense was inhibited by its dominant species, which might be a major reason for its aggregated distribution. As the study scale increased, the interaction gradually changed into an unrelated or positive interaction, showing that the competitive relationship between species improved. Young trees might not have been competitive enough and were gradually eliminated, and then adult trees constantly adapted to the environment in the community. As the number of old trees greatly decreased, the competition with the dominant species was reduced, and the interference in the distribution of T. sinense decreased. Generally, if community succession is at an early stage and the niches between species are similar, there will often be negative interactions between species, and there will be a competitive relationship between species at this time [40,58]. As succession continues, the relationship becomes positive [59], and the niches between species differentiate as coexistence is achieved. When coexistence occurs, the community is at the stage of late succession [60,61,62].

4.3. Correlation Analysis of Key Factors

We observed that the spatial pattern and richness were strongly associated with biotic and abiotic factors. The correlation environmental factors affecting the richness of T. sinense were altitude, humidity, litter depth, N, P, Ca, Zn, and Mg. The altitude, shade density, litter depth, Na, Ca, Zn, K, A. sinense, P. tomentosa, S. japonicus, and P. psilophyllus also affected its distribution. The results of this study show that most of the companion species were distributed in areas with medium altitude gradients and high contents of various mineral elements, so the companion species will likely compete for the same habitat resources with T. sinense, thus affecting the distribution pattern of T. sinense. T. sinense is selective for habitat factors such as altitude and soil mineral elements, and its main concentration is in the middle- and low-altitude areas. This result is consistent with the distribution of T. sinense in the fields: P1 had the highest altitude and the lowest number of T. sinense, and no seedlings were found. P2 was at a medium-low altitude and had the highest number of T. sinense and was the only sample site where seedlings were found. Meanwhile, the results of this study also conformed to the law of decreasing plant diversity with increasing altitude, which is consistent with the results for species such as Taxus mairei [62]. The distribution of T. sinense is greatly affected by altitude, and the higher the altitude, the less it is distributed, while its dominant species are also distributed at higher altitudes, which may be because T. sinense does not have a strong adaptive ability compared with other dominant species. The increase in altitude is accompanied by a decrease in temperature and environmental conditions, which puts the population at a disadvantage when competing with the dominant species. In summary, the altitude, litter depth, Ca, Zn, and the four dominant species were the key factors affecting the spatial pattern of T. sinense. The mid- and low-altitude areas with a relatively low litter depth and more Ca and Zn were suitable for T. sinense. Similar findings were observed for other relict plants.

5. Conclusions

In this study, the spatial pattern and factors limiting the regeneration of T. sinense were studied. T. sinense clusters were distributed in the 1700–1800 m altitudinal belt. At a small scale, individuals of T. sinense were highly clustered, and there were interactions between different diameter classes, which might reduce its genetic diversity. Intense competition within populations led to their declines. Therefore, proper thinning of seedlings and transplanting of T. sinense will help reduce population declines caused by competition. Moreover, artificial breeding technology can be used to breed well-growing T. sinense seedlings and expand their population. On the other hand, T. sinense and its dominant species had a strong competitive relationship at a small scale, with T. sinense being in a disadvantaged position. To change this situation without changing the stability of the community, appropriately cutting down the dominant species would help the growth and survival of T. sinense. Our study shows that T. sinense is suitable for distribution in mid-altitude areas with a low litter depth and more Ca and Zn. Given these findings, when returning artificial seedlings to the wild, the location should be selected based on quantitative analyses.

Overall, our results highlight the importance of intra- and interspecific interactions and environmental factors for the spatial pattern of T. sinense. This study shows that the distribution of T. sinense was highly aggregated at a small scale, which was fully confirmed in the field survey. However, the kinship between the clustered individuals is not clear and needs to be further studied. If the relationships of these individuals can be determined, it will be possible to discuss the endangerment mechanism of the tree in conjunction with the genetic structure of the population. In short, this study disentangles the spatial pattern of T. sinense and its influencing factors. Furthermore, it has relevant implications for conserving and managing rare and relict trees in a constantly fragmented habitat.