Effects of Stand Structure of Artificial Shelter Forest on Understory Herb Diversity in Desert-Oasis Ecotone

Abstract

The relationship between the spatial structure of shelter forests and the diversity of understory herbaceous plants in desert–oasis ecotones is important for maintaining biodiversity indices and protecting the oasis ecosystem. In this paper, we explore the coupling relationship between tree layer structure (competition index, angle scale, neighborhood comparison, DBH, etc.) and understory herb diversity in the transition zone of shelter forest plots near oases and near deserts; in addition, we also aim to elucidate the dominant stand structure factors affecting herb biodiversity. The results indicated the following: A total of 13 herbaceous plant species were discovered in the transitional zone, with 11 species found near the oasis area and 4 species near the desert region. The Shannon, Simpson, and Pielou indices of understory herbaceous plants were significantly higher near the oasis area compared to the desert region. The Margalef index mean was higher in the oasis area compared to the desert region. Pearson and canonical correlation analyses revealed significant associations between specific stand structure indicators and diversity in the herbaceous layer. The results of the multiple linear regression analysis revealed that the competition index had a significant impact on the Shannon, Simpson, and Pielou diversity indices of the herbaceous layer in the understory of the shelterbelt forest near the oasis, with corresponding impact coefficients of 0.911, 0.936, and 0.831, respectively. The mingling degree was found to be the primary influencing factor for the Margalef index, with an impact coefficient of 0.825. However, in the understory of the shelterbelt forest near the desert, the neighborhood comparison ratio negatively affected the Shannon and Margalef indices, with impact coefficients of −0.634 and −0.736, respectively. Additionally, tree height negatively impacted the Simpson and Pielou indices, with impact coefficients of −0.645 and −0.677, respectively. In order to enhance the diversity of understory herbaceous species in the transitional zone and preserve the ecological system of the oasis, specific modifications to the forest structure and arrangement are essential. Pruning and thinning are necessary for shelterbelt forests located near desert regions, while shelterbelt forests near oases should use a suitable mix of tree species. These measures can help preserve or enhance the diversity of understory herbaceous plants.

1. Introduction

The transitional zone bordering the Taklimakan Desert and the 11th regiment of Alar City in southern Xinjiang belongs to the alluvial–pluvial fan oasis–desert fragile zone [1]. The fragile zone is characterized by obvious environmental degradation, land desertification, serious salinization, weak soil fertility, and a fragile ecological environment. As a special ecological landscape embedded in desertified land, the oasis is a sensitive area of the arid ecosystem, and it is vulnerable to wind and sand disasters. Optimization and desertification are two geographical processes of opposite development. In order to avoid desertification in desert oases and their peripheries, the local government protects the soil by increasing the surface vegetation coverage. While increasing the area of cultivated land, protective shelter forests are planted outside of farmland to form a “green barrier”, which effectively blocks the expansion of the Taklimakan into the oases to the North and greatly improves agricultural productivity. In addition to environmental improvement, planting shelter forests in sandy areas can also conserve biological resources, increase biodiversity, and increase plant biomass [2]. As an integral part of the arid ecosystem, understory vegetation plays a crucial role in fixing sand flow and soil nutrient enrichment [3]. The study of the biodiversity of understory herbaceous plants in the desert–oasis transition zone can reflect the adaptability of plants to the environment and the succession patterns in this region. Whittaker proposed three types of diversity measures (α, β, and γ diversity) in 1972, which have been used up to the present day, among which α diversity is applicable to a small-scale range [4], including species richness, species relative abundance, species diversity index, and species evenness index.

Understory herbaceous plants play a vital role in maintaining the normal functioning of the ecosystem. The distribution pattern and biodiversity of organisms are influenced by both environmental factors and the structure of the upper layer. Most studies analyze the effects of environmental factors, such as topography [5], soil nutrients [6], and microclimates [7,8]. However, the diversity of understory herbaceous plants is also significantly affected by the stand structure of tree layers at small scales [9,10]. Previous studies have suggested that non-spatial structural factors (such as tree height, diameter at breast height, and density), as well as spatial structural factors (such as angle scale, mingling degree, and competition index) [11], restrict the diversity of understory herbaceous plants. For example, Li et al. [12] found that the diversity of understory herbaceous plants in an aerially sown Pinus massoniana forest was mainly affected by the stand density index. Zhang et al. [13] suggested that the opening degree and mingling degree were the dominant stand structures affecting the overall level of herbaceous plant diversity in Platycladus orientalis forests. Liu et al. [14] conducted a study on the spatial structure of forests at three different stages of succession. They found that forest structure measures, such as angular scale, forest layer index, and openness in spatial structure parameters, had a significant impact on the diversity of shrubs and grasses within the forests.

The above studies show that understory vegetation is closely related to the structure of the upper forest stand. In this study, the transition zone between the Taklamakan Desert and the 11th regiment of Alar City was taken as the study area. We investigated the species and biodiversity of understory herbaceous vegetation, as well as the structural characteristics of shelterbelt forests located at the edge of deserts and near oases in the transition zone, explored the dominant stand structural factors influencing the diversity of understory herbaceous plants in these shelterbelt forests, and analyzed the relationship between the characteristics of understory herbaceous communities and changes in the spatial structure of the forest stand. The impacts of shelterbelt construction on understory herbaceous communities in this area will be explored in order to adjust the shelterbelt structure based on local conditions and promote the recovery of vegetation in the desert–oasis transition zone. This research is of great value for local ecological management.

2. Materials and Methods

2.1. Overview of the Study Area

The experimental area (Figure 1) is located in the transitional zone (80°30′–81°58′ E, 40°22′–40°57′ N) between the 11th Regiment of Alar City and the Taklimakan Desert in the Xinjiang Uygur Autonomous Region. It possesses a warm, temperate continental arid desert climate, with an average altitude of 1015 m, with a maximum altitude of 1028 m and a minimum altitude of 1000 m, an average annual temperature of 10.7 °C, a maximum temperature of 43 °C, a minimum temperature of −27 °C, and an average frost-free period of 220 days. The average annual amount of solar radiation is between 133.7 and 146.3 kcal/cm². The average annual daylight hours are 2556.3–2991.8 h, the average annual precipitation is 54.8 mm, the maximum annual precipitation is 82.5 mm, and the minimum is 40.1 mm; the maximum annual evaporation is 2558.9 mm, and the minimum is 1876.6 mm [15,16]. The dominant wind direction is northwest and southwest, and the annual average wind speed is 3.5~4.5 m/s. The northwest wind is dominant in the winter, and the southwest wind is dominant in the summer. The vegetation coverage in this area is low, and the surrounding farmland shelterbelts are mostly small grids and narrow forest belts; the main tree species in the area are Populus alba var. pyramidalis and Populus euphratica.

2.2. Sample Plot Setting

In this study, the plots were located in the desert–oasis transition zone of the eleventh regiment of Alar City. Plots with similar tree species, slope, slope direction, and soil texture were selected based on the results of previous forestry surveys. Thirteen standard shelter forest plots with Populus alba var. pyramidalis as the dominant tree species and Populus euphratica as the associated tree species were selected (including 7 shelter forest plots near the desert (Area A) and 6 shelter forest plots near the oasis (Area B)), and the average area of the protective forest is 11860 m² (593 m × 20 m). Three small quadrats with an area of 900 m² (90 m × 10 m) were established in each plot. The survey included recording the tree species composition, DBH, tree height, crown width, etc., and a grid method was used to record the relative coordinates of each tree (x, y). At the same time, according to the five-point sampling method, five 1 m × 1 m herb quadrats were established in each sample plot to record the species name and number of herbaceous species in each quadrat.

2.3. Research Methods

2.3.1. Stand Structure Calculation

Stand structure includes stand spatial structure and non-spatial structure. In this study, the stand spatial structure parameters measured were mingling degree (M), neighborhood comparison (U), competition index (CI), angle scale (W), and opening degree (K); the stand non-spatial structure parameters were tree height (H) and diameter at breast height (DBH).

The spatial structure of a forest stand is a crucial basis for evaluating the distribution and growth conditions of trees within the stand. It is defined as the positional relationship of all tree species within a forest stand, based on spatial structure units. Each spatial structure unit consists of an object tree and its neighboring trees. In this study, Voronoi polygons were employed to determine the competition units of trees [17,18]. The tree coordinates were imported into ArcGIS software to construct Voronoi diagrams and Delaunay triangulations (Figure 2). The trees within each Voronoi polygon are object trees, while the trees in the neighboring Voronoi polygons are considered competing trees. The number of competing trees (n) in an object tree is equal to the number of neighboring Voronoi polygons. The distance between the object wood and its nearest neighbor wood is represented by the length of the side of the triangle in the Delaunay triangular mesh construction. Additionally, the angle between the two sides originating from the object wood represents the angle of the nearest competing neighbor wood (α) [19,20]. This method ensures maximum correlation between competing and object trees and improves the precision of the study results. It avoids the shortcomings of selecting too few near-neighbor trees, which results in failing to take into account all the surrounding near-neighbor trees, or selecting too large a number of near-neighbor trees, which includes trees that are not actually near-neighbors in the consideration [21]. Additionally, it overcomes the drawback of the traditional fixed-radius circle that incorrectly classifies the object wood from the competing woods [22].

Calculation formula for spatial structure parameters of forest stands:

(1) Neighborhood comparison (U) is an index that represents the difference between neighboring trees based on the relative difference in diameter (or height, crown width) between the target tree and the competing tree. It is calculated by examining the relationship between the diameter (or height, crown width) of the target tree and the competing tree [23,24,25,26]. The formula is as follows:

$${\mathrm{U}}_{\mathrm{i}}=\left(\frac{1}{\mathrm{n}}\right)\underset{\mathrm{j}=1}{\stackrel{\mathrm{n}}{\sum {\mathrm{K}}_{\mathrm{ij}}}}$$

(1)

(2) Mingling degree (M) is a measure of the extent to which different tree species are mixed together in a forest stand. It quantifies the level of interspecies interactions and diversity within the stand. The mingling degree can be calculated by determining the percentage of competing trees that belong to species other than the target tree species [23].

$${\mathrm{M}}_{\mathrm{i}}=\left(\frac{1}{\mathrm{n}}\right)\underset{\mathrm{j}=1}{\stackrel{\mathrm{n}}{\sum {\mathrm{V}}_{\mathrm{j}}}}$$

(2)

$${\mathrm{V}}_{\mathrm{j}}=\left\{\begin{array}{cc}1,& \mathrm{T}\mathrm{h}\mathrm{e} \mathrm{c}\mathrm{o}\mathrm{m}\mathrm{p}\mathrm{e}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e} \mathrm{i}\mathrm{s} \mathrm{o}\mathrm{f} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{s}\mathrm{a}\mathrm{m}\mathrm{e} \mathrm{s}\mathrm{p}\mathrm{e}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{s} \mathrm{a}\mathrm{s} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{t}\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{e}\mathrm{t} \mathrm{t}\mathrm{r}\mathrm{e}\mathrm{e} \mathrm{i}\\ 0,& \mathrm{C}\mathrm{o}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{y}\end{array}\right.$$

(3) Angle scale (W) is an index that describes the uniformity of competing trees around a target tree. It is calculated by taking the target tree as the vertex and counting the number of neighboring competing trees whose angle $\left(\alpha \right)$ with the target tree is less than the standard angle $({\alpha }_0=\frac{360°}{n}\pm \frac{360°}{10n})$, divided by the total number of n closest neighboring trees to the target tree [25].

$${\mathrm{W}}_{\mathrm{i}}=\left(\frac{1}{\mathrm{n}}\right)\underset{\mathrm{i}=1}{\stackrel{\mathrm{n}}{\sum {\mathrm{Z}}_{\mathrm{ij}}}}$$

(3)

$${\mathrm{Z}}_{\mathrm{ij}}=\left\{\begin{array}{cc}1,& \mathrm{T}\mathrm{h}\mathrm{e} \mathrm{j}\mathrm{t}\mathrm{h} \mathrm{a}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e} \left(\alpha \right) \mathrm{i}\mathrm{s} \mathrm{l}\mathrm{e}\mathrm{s}\mathrm{s} \mathrm{t}\mathrm{h}\mathrm{a}\mathrm{n} \mathrm{t}\mathrm{h}\mathrm{e} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{d} \mathrm{a}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e} \left({\alpha }_0\right)\\ 0,& \mathrm{C}\mathrm{o}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{y}\end{array}\right.$$

(4) Competition index (CI) is an indicator that reflects the relationship between individual tree growth and spatial occupation. It utilizes Hegyi’s diameter–distance competition index [27,28], where D represents the distance between the target tree and the competing tree, and H_i and H_j represent the diameters of the target tree i and the competing tree j, respectively.

$${\mathrm{CI}}_{\mathrm{i}}=\underset{\mathrm{i}=1}{\stackrel{\mathrm{n}}{\sum (\frac{1}{{\mathrm{D}}_{\mathrm{ij}}}\times \frac{{\mathrm{H}}_{\mathrm{j}}}{{\mathrm{H}}_{\mathrm{i}}})}}$$

(4)

(5) Opening degree (K) is a relative measure of light intensity within a forest stand [29]. It is defined as the ratio between the distance and the height of the target tree and the competing tree. D_i represents the height of each individual plant, and H_ij represents the distance.

$${\mathrm{K}}_{\mathrm{i}}=\underset{\mathrm{i}=1}{\stackrel{\mathrm{n}}{\sum \left(\frac{{\mathrm{D}}_{\mathrm{i}}}{{\mathrm{H}}_{\mathrm{ij}}}\right)}}$$

(5)

Drawing on previous research findings [30,31,32,33], the mingling degree (M), neighborhood comparison (U), uniform angle scale (W), and opening degree (K) were divided into five grades, as shown in

Table 1. Stand spatial structure parameters.

Stand Spatial Structure Parameter	Stand Spatial Structure Grade
	I	II	III	IV	V
Neighborhood comparison (U)	0	(0, 0.33)	[0.33, 0.67]	(0.67, 1)	1
Opening degree (K)	(0, 0.2]	(0.2, 0.3]	(0.3,0.4]	(0.4,0.5]	(0.5, +∞]
Angle scale (W)	(0, 0.475)	[0.475, 0.517]	[0.517, +∞]	-	-
Mingling degree (M)	0	(0, 0.25]	(0.25, 0.5]	(0.5, 0.75]	(0.75, 1.00]

. The parameters U of each grade from small to large represent dominance, sub-dominance, mean, bad state, and absolute bad state. Each level of parameter K represents a serious shortage, insufficient, basically sufficient, sufficient, or very sufficient; the grades of parameter W represent uniform distribution, random distribution, and cluster distribution. Each grade of parameter M represents zero-degree mingling, weak-degree mingling, moderate-degree mingling, strong-degree mingling, and extremely strong-degree mingling. The lower the competition index (CI), the more advantageous the position of the target wood is in the competition [34].

2.3.2. Calculation of Understory Herbaceous Plant Diversity Index

In this study, the Simpson index, Shannon–Wiener index, Margalef richness index, and Pielou evenness index were selected to reflect the species diversity of understory herbaceous plants. The calculation formulae are as follows [35,36]:

Margalef richness index:

$$\mathrm{R}=\frac{(\mathrm{S}-1)}{\mathrm{In}\left(\mathrm{N}\right)}$$

(6)

Pielou evenness index:

$$\mathrm{E}=\frac{\mathrm{H}}{\mathrm{In}\left(\mathrm{S}\right)}$$

(7)

Simpson index:

$$\mathrm{D}=1-\underset{\mathrm{i}=1}{\stackrel{\mathrm{s}}{\sum {\left({\mathrm{P}}_{\mathrm{i}}\right)}^2}}$$

(8)

Shannon–Wiener index:

$$\mathrm{H}={\sum }_{\mathrm{i}=1}^{\mathrm{S}}{\mathrm{P}}_{\mathrm{i}}\mathrm{In}{\mathrm{P}}_{\mathrm{i}}$$

(9)

In the formulae, S is the total number of species;

N is the number of individuals of all species;

H is the species richness of various species in the habitat;

P_i is the proportion of species in the total community;

N_i is the number of individuals of the species.

2.3.3. Data Processing

The article utilized ArcGIS 10.8 to create Voronoi diagrams and Delaunay triangulation grids in order to calculate spatial structure parameters for forest stands (see details in Section 2.3.1). Pearson correlation analysis is employed to select forest structure factors that are related to various diversity indices (p < 0.05). The R language (yacca package) is used for canonical correlation analysis to explore the relationship between diversity indices and forest structure factors, which are two distinct sets of variables. By using the species diversity index as the dependent variable and the forest structure index as the independent variable, stepwise regression is conducted to establish a multiple linear regression model and determine the directly related parameters that affect the diversity index of understory herbaceous plants.

3. Results

3.1. Understory Herbaceous Species Diversity Characteristics

The study area is part of the Asian Desert Plant Subregion of the Pan-Arctic Flora, which originated from the dry-hot flora on the southern coast of the ancient Mediterranean. After long-term drought and species invasion, its species composition is complex and ancient [37]. According to the plot survey findings (

Table 2. Herbaceous plant classification and importance value.

Department	Genus	Plant Names	Importance Value
			Oasis (A)	Desert (B)
Leguminosae	Glycyrrhiza	Glycyrrhiza uralensis	0.084	-
	Alhagi	Alhagi sparsifolia	-	0.052
Gramineae	Phragmites	Phragmites australis	0.433	0.569
	Eleusine	Eleusine multiflora	0.210	-
Compositae	Taraxacum	Taraxacum monochlamydeum	0.141	-
	Cirsium	Cirsium setosum	0.121	-
	Sonchus	Sonchus wightianus	0.145	-
		Sonchus brachyotus	0.057	-
	Karelinia	Karelinia caspia	-	0.261
Polygonaceae	Polygonum	Polygonum patulum	0.057	-
	Rumex	Rumex acetosa	0.323	-
Asclepiadaceae	Cynanchum	Cynanchum cathayense	0.147	0.391
Chenopodiaceae	Chenopodium	Castanopsis fissa	0.061	-

), Area A contained 11 plant species in 6 families and 10 genera. Herbaceous plants were concentrated in Leguminosae, Gramineae, Compositae, and Polygonaceae, with Compositae containing the highest number of plant species. The dominant species in this area were Phragmites australis, Rumex acetosa, and Eleusine multiflora, with importance values of 0.433, 0.323, and 0.21, respectively. In contrast, Area B had fewer herbaceous species overall, with only four plants in four families and four genera. The absolute dominant species in this area was Phragmites australis, with an importance value of 0.569. Cynanchum cathayense followed with an importance value of 0.391. Interestingly, two desert plants, Alhagi sparsifolia and Karelinia caspia, appeared in the sample plots in Area B. The importance value of Karelinia caspia (0.261) was higher than that of Alhagi sparsifolia (0.052).

The diversity index of understory herbaceous plants in each plot is displayed in

Table 3. Basic characteristics of species diversity indices.

Index	Oasis (A)		Desert (B)		t Test
	Mean ± Standard Deviation	Coefficient of Variation (%)	Mean ± Standard Deviation	Coefficient of Variation (%)	T	p
Shannon–Wiener index	1.206 ± 0.278	23.02%	0.534 ± 0.265	49.65%	4.141	0.001 ***
Simpson index	0.658 ± 0.060	9.06%	0.303 ± 0.168	55.49%	3.877	0.001 ***
Margalef richness index	0.761 ± 0.526	69.08%	0.422 ± 0.170	40.28%	1.665	0.115
Pielou evenness index	0.899 ± 0.071	7.86%	0.573 ± 0.606	35.89%	2.903	0.010 **

Note: ***: p < 0.001; **: p < 0.01.

. In Area A, the Shannon, Simpson, and Pielou indices were significantly higher than those in Area B. However, there was no significant difference between the Margalef indices of Area A and Area B (p = 0.115). The Margalef index is employed as a quantitative measure to assess the level of species richness present within a given ecological community. The variation in species abundance across different sites in Area A results in a substantial coefficient of variation for the Margalef index, reaching 69.08%. The coefficient of variation in this area is not statistically different from the coefficient of variation observed in Area B. The coefficients of variation for the diversity indices were higher in Area B. The Simpson’s index had the highest coefficient of variation at 55.49%. Variations in species richness and evenness among the same sites in Area B resulted in higher coefficients of variation for the diversity indices.

3.2. Basic Characteristics of Stand Structure

From

Table 4. Characteristics of stand structure index.

Index	Area A		Area B		t test
	Mean ± Standard Deviation	Coefficient of Variation (%)	Mean ± Standard Deviation	Coefficient of Variation (%)	T	p
Tree height (m)	23.32 ± 0.78	28.53%	17.10 ± 0.25	24.3%	2.478	0.015 **
DBH (cm)	39.62 ± 1.73	37.06%	20.53 ± 0.39	36.56%	7.587	0.000 ***
Mingling degree	0.09 ± 0.14	155.56%	0.07 ± 0.22	316.23%	0.210	0.837
Neighborhood comparison	0.50 ± 0.08	15.37%	0.49 ± 0.07	13.30%	0.281	0.783
Competition index	1.40 ± 0.67	47.73%	3.29 ± 1.94	13.11%	−2.279	0.039 **
Angle scale	0.50 ± 0.023	4.47%	0.51 ± 0.04	7.70%	−0.265	0.795
Opening degree	2.90 ± 1.92	91.51%	0.58 ± 0.31	53.23%	2.502	0.025 **

Note: ***: p < 0.001; **: p < 0.01.

, it can be seen that the coefficient of variation for non-forest spatial structure parameters (tree height, DBH) in Area A and B is significant; meanwhile, the non-forest stand spatial structure indicators in A were all significantly more extensive than those in B (p = 0.015, p = 0.000). Referring to Table 1, analyzing the spatial structure indicators of each forest stand in Area A, we can see that the neighborhood comparison was 0.495, which is moderate. The opening degree was 2.903, indicating that the stand growth space is mainly sufficient; the angle scale was 0.503, which is close to random distribution, thus showing that the distribution state is ideal. The mingling degree was weak at 0.009, indicating a high degree of isolation between tree species. The neighborhood comparison of Area B was 0.485, and the forest differentiation was moderate; the opening degree was 0.58, the light transmission condition of the stand was good, and the growth space of the stand was sufficient. The angle scale was 0.508, and the horizontal distribution pattern of the trees conformed to a random distribution. The mingling degree was 0.069, showing weak mingling. The competition index values for Area A and B were 1.403 and 3.292, respectively, and the competitive pressure of Area B was significantly greater than that of Area A (p = 0.039). At the same time, the differences in mingling degree, neighborhood comparison, and angle scale were not significant. Moreover, the coefficient of variation for most of the indices in Areas A and B was relatively large, among which the coefficient of variation for the mingling degree was the largest at 155.56% and 316.23%, respectively (both of which are greater than 100%, thus showing substantial variability).

3.3. Correlation Analysis between Stand Structure Factors and Herb Diversity Index

The results of the Pearson’s correlation analysis conducted on the stand structural factors and diversity indices (shown in Figure 3) reveal a significant positive correlation between the Margalef richness index, Shannon evenness index, and Simpson diversity index of herbaceous plants, and the stand competition index (CI) in area A (p = 0.025, p = 0.011, p = 0.006). Furthermore, the Pielou evenness index exhibited a significant correlation with the stand mingling degree (M) (p = 0.045). Moreover, the Margalef richness index of herbaceous plants in area B demonstrated a noteworthy inverse relationship with the neighborhood comparison (U) (p = 0.013). Additionally, the Pielou evenness index and Simpson diversity index demonstrated a significant correlation with the tree height (H) (p = 0.032, p = 0.044). Conversely, the Shannon evenness index showed a significant negative correlation with the neighborhood comparison (U) (p = 0.049). Nevertheless, no significant correlation (p > 0.05) was found between the opening degree (K), angle scale (W), DBH, and the four diversity indices.

3.4. Canonical Correlation Analysis between Stand Structure and Understory Herb Diversity

The results of Pearson’s correlation analyses showed that the diversity of understory herbaceous plants was significantly correlated (p < 0.05) with the mingling degree (M) and the competition index (CI) in Area A. Additionally, the diversity of understory herbaceous plants was significantly correlated (p < 0.05) with the neighborhood comparison (U) and tree height (H) in Area B. To gain a deeper understanding of these correlations, we assign the four diversity indices in Area A (Margalef, Pielou, Simpson, Shannon) as the composite variable U₁, while the indices for mingling degree and competition index are designated as the composite variable V₁. Similarly, the four diversity indices in Area B are denoted as the composite variable U₂, while the indices for neighborhood comparison and tree height are denoted as the composite variable V₂. Canonical correlation analysis is subsequently conducted on the composite variables U₁, V₁, and U₂-V₂ to elucidate the linear associations between the biodiversity indices and forest structure variables in Areas A and B. This analysis provides a quantitative representation of the associations.

Table 5. Typical correlation coefficient and its significance test.

Sample	Typical Variable Group	Canonical Correlation Coefficient	p
A	1	1.000	0.000 **
	2	0.915	0.004 *
B	1	0.947	0.000 **
	2	0.710	0.013 *

Note: **: p < 0.01; *: p < 0.05.

presents the correlations between vegetation indices and environmental factors, offering valuable insights into their relationships. The correlation coefficients between the first group and the second group of typical variables in Area A were 1 (p = 0.000) and 0.675 (p = 0.004), respectively. The canonical correlation coefficients between the two groups of typical variables in Area B were 0.947 (p = 0.000) and 0.710 (p = 0.013), respectively, which reached extremely significant levels, indicating that the two groups of variables passed the canonical correlation coefficient test and that there was a strong correlation. The variables composed of two stand structure indicators can be used to explain the variables of understory herbaceous diversity composition. Generally speaking, the significance of the first pair of canonical correlation variables is evident and represent variables that explain a great deal of variation [38]. Therefore, only the first set of canonical variables was analyzed.

Combining the information presented in Figure 4 and

Table 6. Standardization coefficient and cross load coefficient of typical variables.

	Variable		X1	X2	X3	X4	Y1	Y2	Y3	Y4
A	Standardization coefficient	U1	−0.934	0.269	−0.934	−0.959	-	-	-	-
		V1	-	-	-	-	0.213	−0.984	-	-
	Cross-load coefficient	U1	−0.940	0.323	−0.934	−0.963	-	-	-	-
		V1	-	-	-	-	0.244	−0.978	-	-
B	Standardization coefficient	U2	−0.810	−0.343	−0.521	−0.561	-	-	-	-
		V2	-	-	-	-	-	-	0.971	0.198
	Cross-load coefficient	U2	−0.766	−0.325	−0.493	−0.531	-	-	-	-
		V2	-	-	-	-	-	-	0.920	0.188

demonstrates that, in the linear combination U₁ of the first group of typical variables in Area A, the negative loadings of X₁, X₃, and X₄ are larger, (−0.934, −0.934, and −0.959, respectively), and the cross-load coefficients are −0.940, −0.934, and −0.960, respectively. This indicates that the Margalef, Simpson, and Shannon–Wiener indices were the most responsive indices for the stand structure comprehensive index change; furthermore, they play a decisive role in the comprehensive index of understory herbaceous plant species diversity. The maximum negative loading of Y₂ in V₁ was −0.984, and the cross-load coefficient was −0.978, indicating that V₁ is mainly composed of the competition index. Based on the relationship between the two, it was considered that the Margalef, Simpson, and Shannon–Wiener indices would increase with the increase in the stand competition index. In the linear combination, U₂, of the first group of typical variables in Area B, the maximum negative load of X₁ was −0.810, the cross-load coefficient was −0.766, the minimum negative load of X₂ was −0.343, and the cross-load was −0.325, indicating that the Margalef index had the largest weight in U₂, and the Pielou index had the smallest weight. The maximum load coefficient of Y₃ in V₁ was 0.971, and the cross-load coefficient was 0.920, indicating that V₂ is mainly composed of the size ratio. Therefore, the larger the stand size ratio, the smaller the Margalef index.

3.5. Multiple Stepwise Regression Analysis of Stand Structure and Understory Herb Diversity

The multiple linear regression model was used to find the stepwise regression approach, with the species diversity index as the dependent variable and the stand structure index as the independent variable. In order to determine the relative importance of stand structure factors related to the species diversity index. It can be seen from

Table 7. Stepwise regression equation of species diversity.

	Species Diversity Index	Regression Equation	R2	p	D-W
A	Shannon–Wiener index	y = 0.911 × CI	0.831	0.011	1.898
	Simpson index	y = 0.936 × CI	0.876	0.006	1.552
	Margalef richness index	y = 0.852 × C − 0.442 × W	0.946	0.012	1.899
	Pielou evenness index	y = 0.831 × CI	0.690	0.041	2.588
B	Shannon–Wiener index	y = −0.634 × U	0.402	0.049	1.023
	Simpson index	y = −0.645 × H	0.416	0.044	0.760
	Margalef richness index	y = −0.736 × U + 0.466 × M	0.778	0.005	2.157
	Pielou evenness index	y = −0.677 × H	0.458	0.032	0.957

Note: Competition index—CI, angle scale—W, neighborhood comparison—U, tree height—H, and Durbin–Watson index—D-W.

that all equations passed the significance test (p < 0.05). The regression equation coefficients were well fitted; furthermore, the Durbin–Watson value was around 2.0, indicating that the model can be used to test the dominant stand structure factors affecting the diversity of understory herbaceous species. The four diversity indices of Area A and B were dominated by different stand structure factors. The regression equations of the Shannon–Wiener index, Simpson index, and Pielou evenness index in Area A (R₂ = 0.831, R₂ = 0.876, R₂ = 0.946) all retained the stand competition index, and the standardized coefficients were 0.911, 0.936, and 0.831, respectively. Therefore, it was considered that the competition index was the main influencing factor of the Shannon–Wiener, Simpson, and Pielou evenness indices. The regression equation of the Margalef richness index (R₂ = 0.946) retained the two factors of competition index and mingling degree, and the standardized coefficient for the mingling degree was the largest (0.852); thus, it was considered that the mingling degree had a great influence on the Margalef richness index. The regression equations of the Simpson index and Pielou evenness index in Area B (R₂ = 0.402, R₂ = 0.778) retained tree height, and the standardized coefficients were −0.645 and −0.677, respectively, indicating that tree height was the dominant factor influencing the Simpson index and Pielou evenness index. The Shannon–Wiener index regression equation (R₂ = 0.402) retained the stand size ratio, and the standardized coefficient was −0.634, indicating that the size ratio was the dominant factor of the Shannon–Wiener index. The regression equation of the Margalef richness index (R₂ = 0.778) retained the size ratio and mingling degree, and the largest standardized coefficient was the size ratio (−0.736), indicating that the size ratio had the greatest influence on the Margalef richness index.

4. Discussion

In an ecologically fragile area such as the desert–oasis transition zone, herbaceous plants play a crucial role in ecosystem restoration; furthermore, they also help to stabilize the ecological environment of the oasis. Studies have shown that the growth of herbaceous plants is often affected by the upper stand [39]. Therefore, it is of great significance to explore the relationship between herbaceous plant diversity and the upper stand structure factors for ecological restoration.

The structural parameters of shelter forests near the oasis (Area A) and near the desert (Area B) in the transitional zone were analyzed. It was considered that they all met the ideal standard, the tree species were open, the viability was good, the degree of forest differentiation was moderate, and the distribution pattern was close to random distribution. In Area A, tree height and diameter at breast height (DBH) were significantly greater (p = 0.015, p = 0.000), while the competition index was significantly lower compared to Area B (p = 0.039). The herbaceous plants found in Area A comprised a total of 11 species belonging to 6 families and 10 genera, indicating a relatively diverse range of species. The dominant species in this area were Phragmites australis, Rumex acetosa, and Eleusine multiflora. Phragmites australis exhibited a significant dominance in Area B, as indicated by its high importance value of 0.569. Two desert plant species, Alhagi sparsifolia and Karelinia caspia, were observed in Area B, whereas they were not present in Area A. Analyses of the understory herbaceous vegetation’s biodiversity revealed a statistically significant increase in the Shannon, Simpson, and Pielou indices in Area A compared to Area B. Furthermore, although the Margalef index did not demonstrate a significant difference in comparison to Area B, it displayed a mean value of 0.761, which was higher than that of Area B (0.422). Hence, it can be inferred that the vegetation growth environment in the transitional zone closer to the desert region is relatively poorer compared to the transitional zone closer to the oasis area. This region is more suitable for the growth of desert plants that are drought-tolerant and adapted to harsh environments.

Pearson correlation analysis and canonical correlation analysis indicate that there was a correlation between the stand competition index, mingling degree, and species diversity index (variable U₁) in Area A, and the correlation coefficient of the competition index was the largest (−0.984). It was considered that the species diversity of understory herbaceous plants in Area A was affected by the combination of the stand competition index and mingling degree. The multiple regression equation showed that the Shannon–Wiener index, Simpson index, and Pielou evenness index equations were positively correlated with the mingling degree of the stand. The Margalef richness index equation was affected by the competition index and the mingling degree, and the coefficients were 0.852 and −0.442, respectively. Therefore, it was considered that the stand competition index in Area A was the main influencing factor of the biodiversity index, and the diversity index will increase with the increase in the stand competition index. This may be due to the increase in the stand competition index, thus resulting in the increase in suitable species and the increase in species diversity. Hence, it is considered that the stand competition index is a significant factor affecting the species diversity of the herb layer. As the stand competition index in Area A is between 0 and 2, it belongs to weak competition. It is believed that an increase in the stand competition index within a certain range will lead to an invasion of shade-tolerant organisms, thereby increasing the value of the species diversity index [40].

The results of canonical correlation analysis in Area B showed that there was a correlation between the stand size ratio, tree height, and species diversity (U₂), and the load coefficients were 0.971 and 0.188, respectively. It can be seen from the analysis of the multiple regression equations that the Simpson index and Pielou evenness index in Area B are significantly affected by tree height, and the coefficients were −0.645 and −0.677, respectively. Previous research conducted by Zhu et al. [41] found a significant negative correlation between the biodiversity index and tree height, which is consistent with the results of our study. This correlation can be explained by the concentration of the root system of Xinjiang poplar primarily within the 0–40 cm soil layer, as suggested by Zhang et al. [42]. As tree height increases, transpiration rates also increase, leading to higher water consumption. This intensifies competition for water and nutrients among herbaceous plants in the shallow soil layer [43]. Consequently, the lack of favorable growth conditions for undergrowth plants results in a decrease in species richness and biodiversity index. The dominant factor of the Shannon–Wiener index equation was the stand size ratio, and the coefficient was −0.643. The Margalef richness index equation was affected by the size ratio and the mingling degree, and the coefficients were −0.736 and 0.466. It is believed that an increase in stand size ratio will lead to a decrease in biodiversity, which is different from the results [44] where the stand size ratio has no significant effect on the species diversity of understory herbaceous plants (p > 0.05). The reason for this discrepancy may be that the size ratio is mainly used to analyze the degree of differentiation for individual tree sizes [45]. The larger the value, the weaker the advantage of Populus alba var. pyramidalis species. The stronger the tree competition, then the lower the overall stability of the stand [46], which in turn affects the degree of the uniform distribution of understory herbaceous plants, resulting in a decrease in the Shannon–Wiener index. Bazzaz found that when the stand size was relatively small, the diversity index of understory herbaceous species was larger [47], which is similar to the results of this study. The Margalef richness index was not only negatively affected by the size ratio, but also positively affected by the mingling degree. Therefore, the higher the mingling degree of trees in the tree layer, then the greater the evenness index and diversity index of the understory shrub species [12,48]. It may be due to the fact that the canopies, branches, and trunks of different tree species in mixed forests form a complex structure. This structure can effectively intercept and attenuate wind forces compared to pure forests, providing a suitable growing environment for understory herbs.

This paper found no significant correlation between stand openness, angular scale, and four diversity indices (p > 0.05). Studies have shown that there is a positive correlation between lower herbaceous plants and upper stand openness in a small spatial scale in closed forests [7]. However, in this study, the shelter forest plots in Areas A and B were low-closed stands; thus, the way in which the lower vegetation obtains light is not limited to the upper stand openness. The abundance and diversity of herbaceous plants are not necessarily related to the effectiveness of light. The size of the angle scale reflects the distribution of the stand. Certain studies suggest that, when the angle scale is large, it is an example of aggregated distribution, which may have a direct or indirect impact on the understory herbaceous plants—thus affecting their species diversity. However, many environmental factors also play a role in interfering with the influence of the light spot distribution of upper trees when the stand with a small angular scale conforms to random distribution [48].

The main dominant factors of biodiversity—which were stand competition index, tree height, and size ratio—in Area A and B were compared. Plot B is adjacent to the desert and is vulnerable to wind and sand. Herbaceous plants in this area have evolved traits to adapt to drought, such as thicker leaves, thicker cuticles, and more complex, deeper, and wide-ranging root systems [49]. Herbaceous plants and trees compete for shallow water, which is easily affected by the size and distribution of trees. Therefore, the main influencing factors of herbaceous biodiversity in Area B are the tree height and size ratio. The sample plot of Area A is close to the oasis, and the natural environment is more suitable for the growth of species than that of Area B. At this time, a moderate increase in stand density and competition index will improve the soil environment and promote the growth of understory vegetation. Therefore, the main influencing factor of understory herbaceous diversity in Area A was the stand competition index.

In summary, through Pearson correlation analysis, multiple linear stepwise regression analysis, and canonical correlation analysis, this study demonstrates the primary and secondary factors of stand structure that impact understory biodiversity of herbaceous plants. In subsequent protection forest construction and ecological restoration work, the stand structure can be adjusted according to the local conditions in order to improve the biodiversity of understory herbaceous vegetation. In the follow-up study, environmental factors such as climate, light, and water can be included in the impact category of understory herbaceous plant diversity. This should be performed so as to more comprehensively understand the main and secondary influencing factors of understory herbaceous diversity. This can also help us to understand the formation mechanism of understory herbaceous plant diversity more accurately and to provide a scientific basis for the protection and utilization of forest resources.

5. Conclusions

A total of 13 species of herbaceous plants were found in the Taklimakan desert–oasis transitional zone in the Alar region of southern Xinjiang, involving 6 families, 12 genera, 9 species of xerophytes, 1 species of mesoxerophytes, and 3 species of xeromesophytes. The results showed that there were differences in the dominant factors in the biodiversity of understory herbaceous plants in the two experimental plots near the desert and near the oasis in the transitional zone. For the shelter forest plots near the oasis, the stand competition index was the main factor affecting the biodiversity of the understory herbaceous layer, and the stand competition index was positively correlated with the biodiversity of the understory herbaceous layer (within a reasonable range). In the shelter forest plots near the desert, there was a significant negative relationship between the biodiversity of the understory herbaceous layer, the stand size ratio, and tree height. In addition, in the investigation of the study area, it was found that there were a few shrub layer species under the forest. Therefore, while reasonably improving the stand structure through pruning and thinning, it is particularly important to increase the construction of understory shrubs. This can not only promote the restoration and growth of vegetation in the transitional zones of the sandy area, but it can also strengthen the windbreak and sand fixation of the tree, shrub, and grass structures so as to improve the soil quality of the transitional zone and improve the biodiversity and ecosystem stability.