Rare Species Are Significant in Harsh Environments and Unstable Communities: Based on the Changes of Species Richness and Community Stability in Different Sub-Assemblages

Abstract

To understand the contribution of different sub-assemblages (sub-communities) in the shrub and herb layers to the distribution patterns of community species richness and the stability of the Pinus massoniana Lamb. community, this study was carried out by using 160 shrub quadrats (5 m × 5 m) and 200 herb quadrats (1 m × 1 m). These quadrats were selected from 40 plots on six islands. In this study, common and rare species were classified according to the frequency, and “new communities” (sub-communities) were formed by adding or removing species. Then the changes of species richness and community stability in the “new communities” were analyzed. A redundancy analysis was also used to explore the factors affecting the size of the species richness in the understory of the Pinus massoniana community. The results showed the following: (1) The distribution patterns of both shrub and herb layer species frequencies in this area were plainly to the right, indicating a large proportion of non-common species (common species accounting for 37.87% in the shrub layer and 16.67% in the herb layer). (2) The higher the frequency of species, the greater their contribution to the pattern of species richness. Common species had a high frequency and were the most important contributors to the patterns of species richness in plant communities (64 common species and 41 most common species in the shrub layer and 10 common species in the herb layer each accounted for 95.72, 88.9, and 90.52%, respectively, of the species richness distribution pattern). However, rare species also made significant contributions to the species richness in regions with hard conditions (the (most) rare species in the herb layer explained more than 70% of the species richness distribution pattern, and the remaining species after removing the (most) common species explained more than 90%). (3) In relatively stable communities, rare species had relatively little influence on the stability of the community, which was mostly governed by the dominant species (common species (10 species) were more stable than rare species (38 species), Exc.-rare species (22 species) were more stable than except-common species (50 species), and Exc.-rarest species (35 species) were more stable than Exc.-most common species (55 species) in the herb layer). In less stable communities, the stability of the community gradually increased with the increase in species richness, which may be associated with the growth habit of the increased species (the stability of the herb layer was higher than that of the shrub layer, as shown by the Euclidean distance). The community stability was determined by not only the dominant species in the community but also the rare species that were important contributors to the stability of the communities. (4) The species richness of the shrub layer was considerably influenced (p-value < 0.05) by the soil pH, soil organic matter, and wind speed, whereas the species richness of the herb layer was significantly influenced (p-value < 0.05) by the soil pH. The greater the pH and wind speed, the greater the species richness in the island community. On islands, the soil stability was maintained in large part by the soil organic matter. The lack of soil organic matter can affect soil nutrients, destroy island habitats, and reduce species richness, all of which are harmful to the community stabilization.

1. Introduction

Exploring the distribution patterns of species richness is conducive to biodiversity conservation [1]. The construction process and species composition of the community are important aspects for forming the patterns of species diversity of the community [2,3]. Among them, common species and rare species are important components of species composition, and their relative contributions to the patterns of species diversity have become a point of contention among scholars [4,5]. The research on the relative contributions of common species and rare species to the patterns of species diversity has been carried out mainly using the correlation analysis method, that is, forming a “new community” (subcommunity) and then comparing the degree of change in the coefficient of correlation between the “new community” and the original community for the species richness [3]. So it can indirectly measure the relative contributions of rare species or common species to the overall species richness of the community. It is generally believed that the patterns of species richness are mainly determined by a large number of rare species with narrow ranges of distribution and low species frequency (abundance) [6], while many studies suggest that common species determine the patterns of species richness [7,8,9]. These two conclusions are quite different, and thus, it still needs to be further verified which species are the main determinants of species richness patterns. In addition, at present, most studies on the patterns of species abundance mainly focus on inland areas. The geographical location of islands is special, quite different from the environment of inland areas. (1) What conclusions will be drawn from the contributions of common and rare species to the distribution patterns of species richness in the community in the special environment of islands? The relationship between community stability and species diversity is also controversial. A large number of studies have shown that the species diversity of the plant community promotes the stability of the community to a certain extent [10,11], but many scholars believe that the stability of the community is mainly determined by the number of dominant species in the community [12,13]. Therefore, when estimating the stability of the community, only the results of those communities with representative species are often considered. (2) Are the common species the determinants of community stability? What is the contribution of rare species to community stability?

Pinus massoniana Lamb. has strong adaptability and tolerance to drought and barren land. It is an excellent pioneer tree species for the restoration of degraded lands [14]. Therefore, it is often planted as plantation forests in subtropical areas. It is widely distributed in the Sandu Gulf region, especially on the islands of the Sandu Gulf. It occupies the largest local forest area and is the most representative forest type in the region with irreplaceable ecological service value [15]. The Pinus massoniana forest on the islands of the Sandu Gulf area was formed by aerial seeding or artificial afforestation over the last century. Restricted by local policies, it is a semi-natural community with relatively little human impact [15]. There are many disadvantages of artificial forests, such as simple community structure and lack of understory vegetation. In contrast, the understory of natural or semi-natural forests is rich and diverse. With the succession and renewal of the community and the superiority and inferiority of plants, the dominant species suitable for survival in the community gradually manifest. They together constitute a relatively stable community structure. Therefore, the study of understory plants in semi-natural forests can provide an important reference basis for the construction related to plantation forests.

Hence, to solve scientific problems 1 and 2, this study takes the shrub and herb layer species of the Pinus massoniana community as the research objects separately. By adding common or rare species one by one, two sequences of common–rare and rare–common were formed to analyze the change in the correlation coefficients of species richness between the “new community” and the original community. Then, the contributions of the most common and rare species to the community richness were assessed to answer scientific question (1). The dominant species of the community were represented by common species, and the M. Godron contribution law developed by Zheng [16] was adopted to compare and analyze the contributions of different species to the stability of the community with the “new community” composed of common species, rare species, non-rare species, and unusual species to answer scientific question (2). These were conducted in order to understand the differences in the contributions of different species in the shrub layer and herb layer of Pinus massoniana communities in the islands of Sandu Gulf, Ningde, to the distribution pattern and stability of community species richness and to explore the main influencing factors affecting the species distribution using redundancy analysis.

2. Materials and Methods

2.1. Research Area Overview

Sandu Gulf is located to the southeast of Ningde City within Fujian Province. It is the midpoint of China’s “Golden Coastline” (18,400 km), approximately 30 km away from Ningde. It is a world-class natural fine deep-water port. There are 126 islands and 17 residential islands, among which Sandu Island is the largest island, with an area of approximately 27.74 km². Sandu Island is the seat of the Sandu Town government. The research area is hilly and has a typical subtropical marine monsoon climate, and primarily red and yellow soils predominate [15]. The secondary Pinus massoniana coniferous forest and Pinus massoniana coniferous and broad-leaved mixed forest are the most widely distributed forests on the island. The main group species are Pinus massoniana, Heptapleurum heptaphyllum (L.) Y. F. Deng, Adinandra millettii (Hook. et Arn.) Benth. et Hook. f. ex Hance, etc. The main auxiliary species are Symplocos sumuntia Buch.-Ham. ex D. Don, Toxicodendron succedaneum (L.) O. Kuntze, Litsea rotundifolia var. oblongifolia (Nees) Allen, etc. The average plant height in the shrub layer was about 0.59 m, the number of plants per square meter was about 7.54, and the percentage of ground cover was about 92.99%; the average plant height in the herb layer was about 0.48 m, the number of plants per square meter was about 25.24, and the percentage of ground cover was about 60.68%.

2.2. Community Survey

Based on a field survey conducted from June to August 2022, the Relevé method [17] was used to set the sampling plots across six inhabited islands with large areas and high forest coverage, namely, Sandu Island, Qingshan Island, Changyao Island, Jigongshan Island, Baipao Island, and Doumao Island. A total of forty 20 m × 20 m plots were established (set 20, 6, 5, 4, 3, and 2 plots in six islands) within Pinus massoniana communities (Figure 1). One shrub quadrat with an area of 5 m × 5 m was set up in each of the four corners of the forest sampling plot, and there was a total of 160 shrub quadrats. One herb quadrate with an area of 1 m × 1 m was set at four corners and the center, totaling 200 herb samples. The species identity (based on Flora of China (http://www.iplant.cn/foc, accessed on 1 December 2022)), plant height, coverage, quantity, and other information on each individual shrub including bamboo, woody vines, and young trees with a height of <3 m and a chest diameter of <3 cm were recorded, and the same information was recorded for each individual herb including herbaceous vines and ferns. A GPS and compass were used to record the longitude and latitude, elevation, slope aspect, slope gradient, and slope position for the sampling plot. Satellite imagery and longitude and latitude measurements were used to calculate the distance from the sample plot to the coastline, the distance to the nearest island, and the distance to the mainland. These last two terms were used to generate an “island isolation factor” for each island. The slope was divided into upper, middle, and lower slopes. The slope aspect was considered to fall into one of four categories: sunny slope (157.5°~247.5°), half-sunny slope (112.5°~157.5°, 247.5°~292.5°), shady slope (0°~67.5°, 337.5°~360°), and half-shady slope (67.5°~112.5°, 292.5°~337.5°) [18]. The crown density was estimated in Photoshop, following methods laid out in Qi et al. [19]. We obtained the average annual temperature, precipitation, and wind speed of each sample plot from 2018 to 2020. The temperatures and precipitation were obtained from the National Qinghai-Tibet Plateau Scientific Data Center (https://data.tpdc.ac.cn, accessed on 24 December 2022). The wind speed data were obtained from the National Earth System Science Data Center, National Science & Technology Infrastructure of China (http://www.geodata.cn, accessed on 24 December 2022), at a resolution of one kilometer. The basic information of the sample plots is shown in Appendix A. A five-point sampling method was used to collect soil at a depth of 25 cm from each sample plot (each shrub plot). All soils were ground and sieved. Afterward, the soils were measured three times to determine nine indicators: the pH value, total potassium, available potassium, total phosphorus, available phosphorus, total nitrogen, alkali-hydrolyzed nitrogen, soil organic matter, and total amount of water-soluble salt [20]. The soil information is shown in Appendix B.

2.3. Analysis Method of the Distribution Patterns of Species Richness

Diversity is expressed by species richness, and the rarity and commonness of species are defined by the frequency. The species with a frequency ≤ 2 are rare species, whereas the species with a frequency ≥ 10 are common species [21]. The species are ranked according to the sequence of frequency from low to high or from high to low. Common or rare species are added one by one to form the two sequences of common–rare and rare–common. The relative contribution of these species to the patterns of species richness was determined by analyzing the degree of change in the diversity of the “new community” and the original community (the entire community consisting of all species in the shrub layer or herb layer) [3].

To further analyze the contribution of the sub-assemblages (“new communities”) composed of the rarest or the most common species to the patterns of the species richness of the whole community, the sub-assemblages composed of the rarest species with a frequency of 1 and the most common species (frequency ≥ 15) accounting for 25% of the total number of the species of the community were formed, and the correlation between the sub-assemblages, the whole community, and the sub-assemblages excluding the rarest and the most common species was analyzed [3,22].

2.4. The Method of Stability Analysis

The M. Godron contribution law [18] was used to analyze the stability of the community. In this study, the “new communities” composed of different species selected from shrub and herb layers were arranged in a descending order based on the relative frequency and then gradually accumulated. The cumulative relative frequency obtained by multiplying the results of the accumulation by 100% was taken as the ordinate (y) data. Different species were selected to form a “new community”. The cumulative result of the reciprocal of the number of these species was multiplied by 100%. Then the cumulative reciprocal percentage of species was obtained as the abscissa (x) data. Thus, a fuzzy model for the scattered points with a smooth curve was established. The intersection point of the straight line passed through the coordinates (0, 100) and (100, 0), where the smooth curve was at the current community stability coordinate point (x, y); (20, 80) was the stable coordinate point of the community. The stability of the community was judged by the Euclidean distance between the calculated point (x, y) and the stable point of the community (20, 80) [23]. The calculation formula is as follows:

$$x=\frac{n}{S}\times 100\%$$

y (relative frequency) = (frequency of species n in the sample/total frequency of all species) × 100%

The smooth curve simulation equation: y = ax² + bx + c

In the formula, n is the number of the species in descending order of relative frequency (n = 1, 2, 3, …n); S is the total number of species.

Crosspoint coordinates: ($x=\frac{-\left(b+1\right)\pm \sqrt{{\left(b+1\right)}^2-4a\left(c-100\right)}}{2a}$, y = 100 − x)

2.5. Environmental Interpretation

Depending on the species richness data of rare, common, Exc.-common (except for the common), and all species of shrub and herb layers in 40 sample plots, we used the Canoco 5.0 software to conduct a detrending correspondence analysis (DCA) on the relative richness of these species. Next, the forward selection method of redundancy analysis (RDA) and Monte Carlo tests (499 cycles) were used to screen the environmental factors with significant impacts (p < 0.05) and study the impacts of environmental factors on the species richness [24].

3. Results and Analysis

A total of 169 plant species in 160 shrub quadrats from 40 sampling plots were recorded. The three species of Ilex pubescens Hook. et Arn., Litsea rotundifolia var. oblongifolia, and Smilax china L., which were the most dominant species in the shrub layer, were distributed in all 40 sampling plots. A total of 60 plant species were recorded in 200 herbaceous samples, and Dicranopteris pedata (Houttuyn) Nakaike was the most dominant species in the herb layer with a absolute frequency of 36. The distribution diagram of species frequency and the species number shows that the patterns of the frequency distribution of the species in the study area were observed on the right side, indicating that the proportion of uncommon species in the community was relatively large (Figure 2). There were 58 rare species (frequency ≤ 2) and 38 most rare species (frequency = 1), accounting for 34.32% and 22.49%, respectively, of the total number of species in the shrub layer, and 64 common species (frequency ≥ 10) and 41 most common species (frequency ≥ 15), accounting for 37.87% and 24.26%, respectively. There were 38 rare species (frequency ≤ 2) and 25 most rare species (frequency = 1), accounting for 63.33% and 41.67%, respectively, in the herb layer, and 10 common species (frequency ≥ 10) and 5 most common species (frequency ≥ 15), accounting for 16.67% and 8.33%, respectively. In addition to the three most dominant species in the shrub layer, the common species that occurred were Gardenia jasminoides Ellis, Symplocos sumuntia, Mussaenda pubescens W. T. Aiton, Heptapleurum heptaphyllum, Toxicodendron succedaneum, Adinandra millettii, etc. The common species of the herb layer included Gahnia tristis Nees, Adiantum flabellulatum L., Lindsaea orbiculata (Lam.) Mett. ex Kuhn, Lophatherum gracile Brongn., etc. A list of species in the shrub and herb layers is provided in Appendix C and Appendix D.

3.1. Contribution of the Common and Rare Species to Species Richness of the Community

According to the species cumulation–correlation coefficient plots for common and rare species, there was a general upward trend in the correlation between the species richness of the “new community” formed and the overall community as more common and rare species were introduced (Figure 3). The degree of variation in the correlation coefficients caused by different species was different. A total of 64 common species in the shrub layer explained 95.72% of the distribution pattern of species richness of the whole community, while 58 rare species explained only 44.2%. The 10 common species in the herb layer explained 90.52% of the distribution pattern of species richness of the whole community, while 38 rare species explained only 83.11%. The growth rate of the common–rare sequence was significantly higher than that of the rare–common sequence.

In the shrub and herb layers, four “new communities” composed of (1) the most common species (41 shrubs and 5 herbs) with the frequency ≥15, (2) the rarest species (38 shrubs and 25 herbs) with the frequency of 1, (3) the total number of species except for the most common species (Exc.-most common, 128 shrubs and 55 herbs), and (4) the total number of species except for the rarest species (Exc.-rarest, 131 shrubs and 35 herbs), and the whole community (all, 169 shrubs and 60 herbs) were analyzed for the correlation of richness patterns, and the results are shown in

Table 1. Correlation of species richness patterns among different types of species in the shrub (lower left) and herb (upper right) layers.

Item	Most Common	Rarest	All	Exc.-Most Common	Exc.-Rarest
Most common		0.411 **	0.479 **	0.214	0.447 **
Rarest	0.337 *		0.731 **	0.683 **	0.561 **
All	0.889 **	0.523 **			0.975 **
Exc.-most common	0.671 **	0.586 **	0.936 **		0.942 **
Exc.-rarest	0.899 **	0.442 **	0.996 **	0.922 **

* p < 0.05; ** p < 0.01; Exc.: except for; The bolded parts are correlation coefficients for the herb layer.

. In the shrub layer, the most common species explained 88.9% (p-value < 0.01) of the species richness distribution patterns of the shrub layer, whereas the rarest species explained 52.3% (p < 0.01). After removing the rarest species, the remaining species explained 99.6% (p < 0.01) of the shrub layer, while after removing the most common species, the remaining species explained 93.6% (p < 0.01). In the herb layer, the most common species explained 47.9% (p < 0.01) of the herb layer, whereas the rarest species explained 73.1% (p < 0.01). After removing the rarest species, the remaining species explained 97.5% (p < 0.01) of the herb layer, while after removing the most common species, the remaining species explained 96% (p < 0.01).

3.2. Contributions of the Common and Rare Species to the Community Stability

M. Godron stability analysis was carried out on the whole community (all, 169 shrubs and 60 herbs) and six “new communities” composed of (1) the species except for the most common species (Exc.-most common, 128 shrubs and 55 herbs), (2) species except for the rarest species (Exc.-rarest, 131 shrubs and 35 herbs), (3) species except for the common species (Exc.-common, 105 shrubs and 50 herbs), (4) species except for the rare species (Exc.-rare, 111 shrubs and 22 herbs), (5) the common species (64 shrubs and 10 herbs), and (6) the rare species (58 shrubs and 38 herbs) (

Table 2. Results of the M. Godron stability analysis.

Layer	Item	Simulated Curves	R2	p	Crosspoint Coordinates	Euclidean Distance	Cumulative Species	Cumulative Frequency
Shrub layer	All	y = −0.0144x2 + 2.2455x + 13.061	0.984	0.000	(31.07, 68.93)	15.66	169	1634
	Exc.-rarest	y = −0.0122x2 + 2.0847x + 8.9627	0.993	0.000	(34.12, 65.88)	19.97	131	1596
	Exc.-most common	y = −0.0135x2 + 2.2199x + 7.111	0.990	0.000	(33.57, 66.43)	19.19	128	606
	Exc.-rare	y = −0.0106x2 + 1.9535x + 7.2182	0.996	0.000	(36.09, 63.91)	22.75	111	1556
	Exc.-common	y = −0.0118x2 + 2.0577x + 7.0064	0.992	0.000	(35.19, 64.81)	21.48	105	333
	Common	y = −0.0069x2 + 1.6342x + 3.6914	0.998	0.000	(40.95, 59.05)	29.63	64	1301
	Rare	y = −0.0056x2 + 1.5046x + 1.7447	0.996	0.000	(43.45, 56.55)	33.16	58	78
Herb layer	All	y = −0.0114x2 + 1.761x + 30.797	0.929	0.000	(28.39, 71.61)	11.87	60	295
	Exc.-rarest	y = −0.0116x2 + 1.8979x + 20.21	0.970	0.000	(31.51, 68.49)	16.28	35	270
	Exc.-most common	y = −0.0106x2 + 1.8035x + 18.973	0.967	0.000	(33.02, 66.98)	18.41	55	163
	Exc.-rare	y = −0.0104x2 + 1.8818x + 12.492	0.991	0.000	(34.71, 65.29)	20.80	22	224
	Exc.-common	y = −0.0082x2 + 1.6187x + 15.681	0.988	0.000	(36.34, 63.66)	23.11	50	109
	Common	y = −0.0074x2 + 1.6722x + 4.3996	0.998	0.000	(40.30, 59.70)	28.71	10	186
	Rare	y = −0.0055x2 + 1.4935x + 1.964	0.995	0.000	(43.49, 56.51)	33.22	38	51

R²: R-squared; p: p-value.

). The stability was highest for the whole community consisting of all species in the shrub and herb layers and lowest for the “new community” composed of rare species.

3.3. Correlation between Species Richness Distribution and Environmental Factors

Species richness distribution and environmental data (with nine soil factors in schedule 2, five topographic factors (elevation, slope aspect, slope, slope position, and distance from coastline), two island isolation factors (distance to nearest island and distance to mainland), three climatic factors (temperature, precipitation, and wind speed), and crown density in schedule 1) of rare species (R), common species (C), Exc.-common species (EC), and all species (A) in shrub and herb layers of 40 sample plots were analyzed using a redundancy analysis. According to DCA sorting, we can obtain that the maximum value of lengths of gradient < 3 (0.5 in the shrub layer and 0.8 in the herb layer). Therefore, the RDA analysis was conducted (Figure 4). The analysis results screened out three environmental factors that significantly affected (p < 0.05) the distribution of species richness in the shrub layer: the soil pH (p = 0.006), soil organic matter (SOM, p = 0.024), and wind speed (WS, p = 0.046). One environmental factor that significantly affected (p < 0.05) the herb layer was the soil pH (p = 0.002). In the shrub layer, the soil pH and wind speed were positively correlated with the distribution of richness of all four types of species. And the soil organic matter was negatively correlated with the species richness of rare and uncommon species but positively correlated with the common species. In the herb layer, the soil pH was positively correlated with the distribution of the species richness of all four types of species. The three environmental factors, soil pH, soil organic matter, and wind speed, together explained 30.4% of the species richness distribution in the shrub layer, indicating that another 69.6% unknown factors influenced the species richness distribution in the shrub layer. The soil pH explained 22.7% of the herb layer, indicating that 77.3% unknown factors influenced the species richness distribution in the herb layer.

4. Discussion

With a significant disparity in the species richness between the two layers, the shrub layer of the research location recorded 169 species of plants, whereas the herb layer only recorded 60 species. Moreover, the proportion of common species in the shrub layer was much larger than that in the herb layer. The main reason for this is that the growth height and canopy characteristics of the dominant species in the tree layer, Pinus massoniana, make it easier for light to reach the understory, encouraging the establishment of heliophilous tree species and enhancing the habitat and species richness of the shrub layer. The herb layer of the study site is dominated by Dicranopteris pedata, which has an allelopathic effect on weed density, seed germination, and seedling growth [25]. The formation of a dense layer of Dicranopteris pedata in the understory also traps a large amount of litter fall from the tree canopy, which has an impact on how quickly it decomposes and how nutrients cycle through the soil [26]. At the same time, the high density of Dicranopteris pedata makes it difficult for positive herbaceous plants to survive in the herb layer. As a result, the herb layer has fewer species and a more hostile environment than the shrub layer. Figure 3 shows that compared with the rare–common species series, the association of richness between the “new community” of common–rare species and the original community increased significantly faster. A total of 64 common species and 41 most common species in the shrub layer each accounted for 95.72 and 88.9%, respectively, of the species richness distribution pattern of the entire shrub layer, and 10 common species in the herb layer accounted for 90.52%. The contribution to species richness was not significantly correlated with species relative proportions since the species proportions of common species (37.87% in the shrub layer and 16.67% in the herb layer) were much smaller than those of Exc.-common species (62.13% in the shrub layer and 83.33% in the herb layer). However, the high frequency and wide distribution of common species make them more susceptible to environmental interference that produce maximum variability [8]. Each rise or fall in common species affected the species richness in many sample sites, whether in the shrub layer or in the herb layer, causing significant modifications to the richness correlation index. The common–rare species sequence in the shrub layer almost stopped changing after adding species ranked around 80 in frequency, and the common–rare species sequence in the herb layer almost stopped changing after adding species ranked around 10 in frequency, at which time the addition of rare species had few changes to the correlation index. This indicates that the relative frequency is one of the dominant factors affecting the pattern of species richness. The contribution of a species to the pattern of richness increases with their relative frequency. It is shown that the reduction in common species is more likely to lead to a reduction in the diversity of species underlying the community, which in turn may lead to a reduction in the genetic diversity of species that diverge under microenvironmental conditions [27]. Therefore, common species are indeed the most important contributors to the species richness pattern of plant communities and should be a priority species for conservation [28]. Some scholars believe that the overall distribution patterns of species richness in the community do not reflect rare species, and rare species can only reflect some random effects, which would interfere with the results rather than provide valuable information. Therefore, rare species are often left out in a certain way for the biological assessment of benthos [29,30]. However, as can be seen from the rare–common species sequence in Figure 3, the rare species (uncommon species) joined the “new community”, although their growth rate was lower than that of the common species, which would also lead to a sharp increase in the correlation of species richness between the “new community” and the original community, especially in the herb layer. The “new community” of the (most) rare species in the shrub layer explained more than 50% of the species richness distribution pattern in the shrub layer, and the “new community” of the remaining species after removing the (most) common species explained more than 80%. The “new community” of the (most) rare species in the herb layer explained more than 70% of the species richness distribution pattern in the herb layer, and the “new community” of the remaining species after removing the (most) common species explained more than 90%. This suggests that the richness of rare species also contributes significantly to the overall community richness [22,31]. The emergence of rare species may make up for some missing functions of the community and make the spatial distribution of plant functional traits of the community more uniform [32]. According to Mouillot et al. [33], in more vulnerable ecological functions, rare species maintain most of the more significant functional characteristics of ecosystems. In areas with poor plant survival conditions and harsh environments [3,21,22], where species richness is significantly low, the increase in rare species can more effectively increase the species richness of the area and improve the diversity index of the region. Thus, in this study, rare species contributed significantly more to the richness correlation in the herb layer with the relatively harsh environment than in the shrub layer with the relatively superior environment. It indicates that rare species are more ecologically significant for environmentally harsh areas, and we cannot simply discard rare species or ignore the contributions of rare species to community diversity patterns.

As shown in Table 2, common species (10 species) were more stable than rare species (38 species), Exc.-rare species (22 species) were more stable than except-common species (50 species), and Exc.-rarest species (35 species) were more stable than Exc.-most common species (55 species) in the herb layer. The main reason may be that these species consist mainly of dominant species with a higher frequency. Many scholars believe that community stability is mainly determined by the proportion of dominant species in the community [13,34]. Therefore, when calculating the community stability, low-frequency species and occasional species in the community are often excluded, and only the stability results of the “community” composed of the main species are calculated [23]. However, some scholars argue that the exclusion of rare species will not reflect whether the community is under disturbance stress, which in turn creates uncertainty in the evaluation of community stability [22]. The results showed that among the four “new communities” in the shrub layer, namely, except the rarest, except the most common, except the rare, and except the common species, the “new communities” composed of species except for the most common and except for the common were more stable (smaller Euclidean distance) than those except for the rarest and except for the rare, respectively. And the Euclidean distance between the intersection coordinates of the “new community” composed of common species and the intersection coordinates of the whole community was 13.97, which was very different. Meanwhile, the frequencies of all except for the most common species and all except for the common species were much greater than those except for the rarest species and except for the rare species, but the stability was smaller. The frequencies of the common species were much greater than those of the rare species, and the stability was larger. This indicates that the difference in frequency is not the dominant factor in determining the community stability and that the frequency is an important indicator to distinguish common and rare species. Thus, this suggests that the community stability is not solely determined by the dominant species but that rare species (uncommon species) are also important contributors to maintaining community stability. Numerous studies have concluded that the species diversity in plant communities contributes to community stability to a certain extent and characterizes community stability by the magnitude of species diversity [10,11]. In the present study, a similar conclusion was reached. The overall trend showed that the Euclidean distance between the desired point and the stability point decreased gradually as the accumulation of species (community species richness) in the shrub layer increased, and the community tended to be more stable, indicating a positive correlation between the species richness and the stability of the community. However, many studies have shown that species diversity can only reflect some aspects of community stability and that the characteristics of the community itself, environmental factors, and community structure all affect the community stability [12,35,36]. The stability of the herb layer is higher than that of the shrub layer, as shown by the Euclidean distance. The shrub layer has a relatively superior environment, where different light-tolerant plants all occupy a sufficient living environment. The better the growth of multiple plants, the more complex the community becomes, which gradually leads to a decline in the status of dominant species. This has resulted in the shrub layer that exhibits the phenomenon of coexistence of multiple dominant species, and the community continues to evolve, with fierce competition among species and relatively low stability. However, with the continued succession of the community, the tree layer will gradually be occupied by shade-tolerant plants. At this time, the increase in the species richness of the shrub layer, especially the increase in shade-tolerant plants, can accelerate the community succession process, which in turn makes the community more stable. While the herb layer is mostly dominated by Dicranopteris pedata as a single species, D. pedata basically monopolizes the herbaceous habitat exclusively, and the herb layer is relatively less competitive and more stable. At this time, the stability of the community is mainly determined by the dominant species such as Dicranopteris pedata, and the influence of rare species on the stability of the community is relatively small.

The results of the RDA analysis showed that the abundance of plants in the shrub and herb layers of the study site was positively correlated with the soil pH. The main reason is that a low pH will lead to a weakened uptake of massive elements by plants and will be detrimental to plant growth [37]. The abundance of shrub layer plants was positively correlated with the wind speed. The fruit types of the species were counted and found to be mainly wind-borne and predominantly microfruits (fruit diameter ≤ 1 cm) of the shrub layer species in the study site [15], and the fruits of these species can be effectively dispersed by wind vectors [38]. Since the study site is an island, it is more exposed to wind, and the plant fruits are spread by wind vectors, which reflects the adaptive mechanisms of plants to their environments. However, excessive wind speed may affect the normal growth of plants, and the plants in the study site community mostly prefer acidic soil [15]. Therefore, the conclusion should be defined as follows: the greater the pH value, the greater the species richness of shrub layer and herb layer plants in the sample plots within a certain range; the higher the wind speed, the greater the species richness of shrub layer plants in the sample plots. The soil organic matter was positively correlated with the species richness of common species in the shrub layer. The main reason is that islands are often subjected to typhoon waves, which lead to severe soil salinization. And organic matter can effectively regulate the content of nitrogen and effective phosphorus elements in the soil, while reducing the content of alkaline nitrogen and the possibility of soil salinization [39], thus facilitating plant growth. However, soil organic matter was negatively correlated with the species richness of rare and uncommon species in the shrub layer. The main reason for this is that islands are typically fragile ecosystems that exhibit vulnerability to damage in the context of their own unique environment and complex disturbances [40]. In areas with impaired environments (e.g., changes in the soil nitrogen, phosphorus, and pH due to soil organic matter deficiency [39]), species decline dramatically, leading to vacancies in the original ecological niches, thus allowing new species to move in to maintain the existing ecological balance. However, rare and uncommon species are more susceptible to disturbances caused by environmental changes and competitive exclusion than common species due to their small number of individuals and restricted distribution range, leading to a higher risk of extinction [41]. Rare and uncommon species have higher rates of loss and migration due to their slower responses to environmental or climate change compared with common species [42]. Therefore, if the region’s environment continues to be impacted and damaged, rare species may also continue to cycle through loss and migration, which still leads to a decline in community richness and is detrimental to the long-term stability of the community.

5. Conclusions and Recommendations

A study related to species richness and stability of shrub layer plants in Pinus massoniana communities revealed that:

(1)

The frequency magnitude of species is an important influence in determining the degree of species contribution to community richness patterns. Common species are frequent and widely distributed, and the decrease in common species can greatly reduce the species diversity of the area. Consequently, common species are the highest-priority species that should be protected in the community. However, rare species with lower frequencies are more important for areas with harsh environments, and the increase in rare species can significantly contribute to the species richness index in areas with lower abundance. Therefore, rare species are ecologically significant for environmentally harsh areas, and we cannot arbitrarily ignore the contributions of rare species to community richness patterns.

(2)

Community stability can be influenced by many factors. In a relatively stable community, the stability of the community is mainly determined by the dominant species, and rare species have a relatively small effect on the stability. In less stable, continuously successional communities, the stability of the community increases gradually with increasing species richness, which may be related to the growth habits of the added species but not significantly correlated with the frequency of the added species. That is, both common species with high frequency and rare species with low frequency may be important contributors to community stability. Therefore, when calculating the community stability, one should never only calculate the stability results represented by the dominant species of the community.

(3)

Within a certain range, the greater the soil pH and wind speed, the greater the species richness in the island community will be. The soil organic matter plays an important role in maintaining soil stability on islands. When organic matter is insufficient, it may lead to the alteration of the soil nutrients, destruction of island habitats, and reduction in species richness, which are detrimental to the stable development of communities. Islands are typical fragile ecosystems, and rare species have a very important ecological status in the harsh environment of islands, especially in terms of their contributions to the richness and stability of communities on the islands. Therefore, in the future research process, rare species should not be simply discarded but should be comprehensively analyzed in combination with other information to delve deeper into the important value of rare species on the islands.