Wealth and Altitude Explain Urban Plant Diversity in Residential Areas of Hainan, China

Abstract

Rapid urbanization worldwide poses significant challenges to biodiversity, as urban habitat fragmentation coexists with diverse landscape forms. Residential areas, a critical component of urban ecology, are essential for understanding the mechanisms that drive biodiversity conservation and the harmonious coexistence of humans and nature. Additionally, the gradient distribution of biodiversity remains a focal point in ecological research, aiding in the comprehension of fundamental species–environment interactions. In this study, we sampled 269 residential areas across fifteen counties and municipal districts on Hainan Island to investigate biodiversity alongside residential characteristics, as well as socio-economic and environmental variables. Utilizing the Generalized Linear Model (GLM), we analyzed the differences and commonalities of plant driving factors through horizontal and vertical two-dimensional gradient models with box plots, Principal Component Analysis (PCA), Principal Coordinate Analysis (PCoA), and path models to examine the existence, distribution, and nature of these gradients. Our findings indicate the following: different plant types are driven by distinct mechanisms; cultivated plants are primarily valued for ornamental purposes, whereas in rural areas, their edible value is emphasized. Urban residential plant diversity was primarily influenced by altitude and fundamentally affected by economic factors. Our analyses identified distinct differences in the driving factors influencing various plant types and established two primary gradients of plant diversity distribution within residential areas: a horizontal gradient influenced by housing prices and a vertical gradient corresponding to changes in elevation. Both gradient models were found to be outcomes of socio-economic factors, highlighting the significant role of economic development in shaping urban biodiversity.

1. Introduction

The rapid pace of urbanization has drastically reshaped global development patterns, with urban populations surpassing rural populations since 2007, a trend that continues to widen [1]. This urban expansion has led to a shift from small rural communities to high-density urban settlements, driving multidimensional urbanization and diversifying urban landscapes [2]. The impact of urbanization on plant diversity is multifaceted. Habitat fragmentation and the increasing isolation of ecosystems present significant challenges to plant diversity. Additionally, urbanization exacerbates climate change, which further contributes to the decline in plant diversity [3,4]. However, urbanization also reshapes plant richness and species composition. Research suggests that urbanization, acting as an environmental filter, stimulates changes in plant phenotypic traits and enhances plant adaptability [5,6,7]. Given that urbanization is reshaping patterns of plant diversity and creating complex landscape dynamics, understanding the factors that influence plant diversity in urban residential areas is crucial for achieving sustainable urban ecosystems, creating green cities, and enhancing the well-being of urban residents [8].

Residential landscapes are key urban land cover types and have been the focus of numerous studies on plant diversity. Socio-economic factors, particularly the dynamics of surrounding communities and economic activities, are known to influence plant diversity. For instance, the “luxury effect” demonstrates that wealthier neighborhoods often exhibit higher species diversity [9,10,11], while increased road density and traffic flow can fragment landscapes and reduce species richness [12,13]. On the contrary, the luxury effect is not a universal concept, as it is influenced by multiple factors [14]. Some studies suggest that in certain cities, bird diversity may not exhibit a positive correlation with wealth levels [15]. Similarly, the assessment of plant diversity in West African urban areas does not provide comprehensive support for the luxury effect hypothesis [16]. As a result, it is important to further investigate and verify the luxury effect in these contexts. Additionally, factors such as the size of residential green spaces, the age of housing, and maintenance practices significantly impact biodiversity [17,18]. Environmental factors, particularly elevation, also play a crucial role in biodiversity, as seen in studies of microbial and insect communities [19,20]. However, the effects of elevation on plant diversity in residential areas remain underexplored.

The concept of gradient changes, which reflects variations in plant diversity, is central to ecological research. Previous studies on Hainan Island have identified horizontal distribution patterns of shrubland vegetation influenced by precipitation and soil conditions [21], while research on tropical rainforests has examined the impact of elevation gradients on biodiversity [22,23]. Socio-economic gradients, such as urban–rural and population-density gradients, have also been widely used to explain biodiversity patterns [24,25,26]. Despite the growing interest in understanding the driving factors of plant diversity, there is a notable gap in research on the plant diversity drivers in residential areas of Hainan Island.

Although the analysis of driving factors has become quite common in the study of urban ecology and species diversity in residential areas, However, we believe that much of the research has separated the driving mechanisms from the patterns of plant diversity changes, meaning that we cannot truly comprehend how these driving factors influence plant diversity. Therefore, we selected several variables to understand the changes in plant diversity based on the identified driving mechanisms. We boldly hypothesize that the plant diversity in residential areas exhibits both horizontal and vertical gradients of variation. This necessitates the need for a typical and diverse research area to more clearly elucidate the relationship between driving mechanisms and gradient changes.

In this study, we extend the traditional definition of residential land use to include both long-term residential areas (e.g., communities and standalone buildings) and short-term commercial accommodations (e.g., hotels). We categorize plants within these areas into three types: (1) trees, shrubs, and herbs, (2) cultivated or spontaneous species, and (3) introduced or native species. Understanding the ecological roles and utilization values of these plant types is vital for exploring their diversity and the driving mechanisms behind their distribution [27,28]. The differences in nutritional content and ecological roles between cultivated and spontaneous species highlight the importance of investigating these dynamics, as variations in plant diversity can have direct implications for urban ecology [29].

This study focuses on Hainan Island, the southernmost province of China, which represents a unique tropical climate with rich plant diversity. Although Hainan occupies just 5.29% of China’s tropical land area, it accounts for 42.5% of the nation’s total tropical land [30]. The rapid economic growth of Hainan, driven by its status as a free trade port, has led to an expansion of residential areas, offering a rich sample for our research. Our objectives are to: (1) investigate the driving factors behind different types of plant diversity in residential areas of Hainan Island, (2) explore the horizontal gradient distribution of plant diversity in these areas, (3) analyze the altitudinal gradient distribution of plant diversity, and (4) propose recommendations for the protection of plant diversity in Hainan Island’s residential areas.

2. Methods

2.1. Overview of the Study Area

The study was conducted in Hainan Province, located on Hainan Island, which spans approximately 3.54 million square kilometers. Positioned between longitudes 108°37′ to 111°03′ east and latitudes 18°10′ to 20°10′ north [31] (Figure 1), Hainan Island is China’s second-largest island, situated at the southernmost tip of the country. It is separated from Guangdong Province by the Qiongzhou Strait to the north, bordered by the Beibu Gulf to the west, and lies in proximity to the Nansha Islands to the south. The island experiences a tropical monsoon climate, characterized by consistently high temperatures and abundant rainfall throughout the year, with indistinct seasonal variation. Nonetheless, the climate can be broadly classified into a rainy season from May to October and a dry season from November to April [32]. These climatic conditions support a rich diversity of tropical flora and fauna, making Hainan a valuable area for studying plant biodiversity. Furthermore, the ongoing development of the Hainan Free Trade Port has invigorated the region’s economic growth [33], contributing to the expansion of the real estate sector. The burgeoning tourism industry has also attracted high-income groups, such as seasonal tourists [34], which has further diversified the landscape of residential areas on Hainan Island, providing numerous case studies for research on human–environment interactions in the region.

2.2. Sampling Protocol

Due to constraints imposed by factors such as privacy concerns and public safety in urban areas, random sampling was not feasible. Therefore, purposive sampling was employed [9]. To ensure representative coverage of the entire island, we utilized a zoning map of Hainan Island, ensuring uniform spatial distribution of sampling locations. To investigate variations in plant diversity across different types of residential areas, we categorized these areas into the following types: (1) multi-floor buildings, defined as residential buildings with more than three but no more than six floors; (2) small high-rise buildings, defined as residential buildings with more than 6 but no more than 11 floors; (3) high-rise buildings, defined as residential buildings with more than 11 floors; (4) villages, defined as residential settlements in rural areas where agriculture is the primary economic activity [35]; (5) hotels, which are residential buildings primarily serving as accommodations for profit; and (6) villas, defined as standalone or semi-detached houses typically associated with high-income groups. To ensure proportional representation, we aimed for each residential area type to account for at least 10% of the total sampling points. To minimize sampling errors, we followed two key principles in selecting sampling locations. First, at the scale of Hainan Island, we took into account factors such as urban size and terrain. Specifically, we prioritized selecting more sampling points in economically developed and larger cities, while fewer points were chosen in smaller, economically underdeveloped counties, with careful attention to maintaining a uniform spatial distribution across the entire island. Second, at the city level, we aimed to evenly distribute sampling points across various types of residential areas and around the city center (using the municipal government as the reference), considering accessibility and ease of sampling. However, to maintain data quality and accuracy, we excluded plots where the total plant species count was fewer than five, as well as plots with no recorded species of herbs, shrubs, or trees. As a result, out of 269 initially surveyed plots, 238 were retained for further analysis (

Table 1. Sampling sites in different types of residential areas.

Type of Residential Area	Number of Sites	Percentage of Total
Multi-floor	38	16%
High-rise	82	34%
Small high-rise	42	18%
Villa	20	8%
Hotel	21	9%
Village	35	15%
Total	238	100%

).

2.3. Field Investigation

The field survey followed a structured methodology. Initially, we identified suitable sampling locations within the residential districts based on accessibility and representativeness. Three 20 m × 20 m plots were established within each selected area, where we recorded the number of species and individuals in the tree layer. To capture diversity in the shrub and herb layers, smaller plots were also established: five 5 m × 5 m plots at each corner and at the center of the main plot for the shrub layer, and five 1 m × 1 m plots for the herb layer [36,37,38,39] (Figure 2). In addition to the biological data, we also gathered information regarding the management practices and socio-economic conditions surrounding each plot, which are further detailed in Section 2.4.

2.4. Plants Driving Factors

The factors influencing plant diversity in residential areas can be analyzed through multiple dimensions. First, plants can be categorized based on their characteristics. These categories include trees, shrubs, and herbs. From the perspective of plant origin, species can be classified as either introduced species (non-native species) or native species (species indigenous to China). Another important classification is based on cultivation practices, where plants are either cultivated species (intentionally planted for ornamental or edible purposes) or spontaneous species (naturally occurring without human intervention). The classification of plant characteristics primarily follows the Flora of China (https://www.iplant.cn/frps2019/, accessed on 1 October 2024).

To explore the driving factors behind plant diversity in residential areas, we categorize these variables into three main groups: residential area characteristics and management variables, socio-economic variables, and environmental variables. The first group, residential area characteristics and management variables, includes factors directly linked to the physical and management aspects of residential areas. These include the year of building construction (BUA), property fees (PRFs), floor area ratio, greening rate, and the number of parking spaces (NCPS). Additionally, the total number of houses (NHO), distance from the main road (DMR), as well as management-related factors such as the number of times watered per year (WT), number of times fertilized per year (FAT), and number of repairs (MVT), are also considered. These variables are all believed to play a role in influencing plant diversity by shaping the environment in which these plants grow.

The second group, socio-economic variables, refers to broader economic and social factors that may affect plant diversity on a larger scale. These variables include housing prices (HOPs), surrounding traffic flow (TFV), and the gross domestic product (GDP) of the region. Additionally, road network density (RND) and population density (POP) in the surrounding area are included as socio-economic variables. HOPs are a significant factor that has been closely linked to plant diversity in numerous studies. Research conducted in tropical cities in China, including Haikou, Zhanjiang, and Sanya, has shown that housing prices are an important explanatory variable for plant diversity [40,41,42]. Furthermore, several studies have demonstrated that HOPs, as a driving factor, are positively correlated with other variables such as GDP [43] and population size (POP) [44], with these factors interacting in a mutually reinforcing manner. Consequently, these factors not only influence the types of plants present in residential areas but also the level of human intervention and infrastructure development, both of which can significantly affect plant distribution. Lastly, environmental variables focus on the natural factors that influence plant diversity. While many studies have shown that socio-economic factors tend to have a greater impact on urban ecological diversity, environmental variables are still crucial for understanding plant distribution. In this study, two key environmental factors were selected the normalized difference vegetation index (NDVI) and the altitude of the sample plot (Height). The NDVI, commonly used to assess vegetation health and density, measures the difference between near-infrared and visible light reflectance. It provides valuable insights into vegetation coverage and vigor, aiding in the assessment of environmental factors’ impact on plant diversity. The altitude, on the other hand, plays an important role in studying vertical gradients in urban ecosystems, offering a deeper understanding of how altitude influences plant distribution and diversity.

In our analysis, HOPs, GDP, and PRFs are specifically highlighted as indicators of the economic status of residential area owners, reflecting their potential influence on plant diversity [45,46,47]. The sources of the aforementioned driving factors are provided in Table S1. Furthermore, to better understand the utilization of cultivated species, we collected data on their ornamental and edible values through both the Flora of China and field surveys. The proportion of cultivated species was calculated, and box plots were drawn to visualize and analyze their utilization values.

2.5. Plant Diversity Index

The biodiversity index is a crucial tool for assessing biodiversity, as it provides a quantifiable measure that is closely aligned with real-world ecological patterns and is more easily understood by the public [48]. In this study, we chose to combine the Margalef index and species richness to evaluate biodiversity. The Margalef index was applied to measure the total species richness, as well as for trees, shrubs, and herbs, which were analyzed using boxplots. Meanwhile, species richness was used to evaluate the diversity of cultivated, spontaneous, introduced, and native species. This dual approach aims to validate the authenticity and reliability of the biodiversity gradients from multiple ecological perspectives.

(1) Species Richness (S): It refers to the total number of different species present within a specific area or range. It is a straightforward measure of biodiversity, reflecting the diversity of species without considering their relative abundances.

(2) Margalef Index (Dₘₐ) [49]:

$${\mathrm{D}}_{\mathrm{m}\mathrm{a}}={\displaystyle \frac{\mathrm{S}-1}{\mathrm{l}\mathrm{n}\mathrm{}\mathrm{N}}}$$

where

• S represents species richness, or the number of species in the sampled area;
• N denotes the total number of individual organisms observed within the same area.

The Margalef index adjusts species richness by considering the total abundance (N), providing a normalized measure of species diversity that accounts for both the number of species and the size of the population.

2.6. Data Analysis

The data collected allowed us to assess species diversity and its associated indices within residential areas, along with the relevant driving factors influencing plant diversity. We used a Generalized Linear Model (GLM) to examine the driving mechanisms behind species richness for total species, trees, shrubs, herbs, cultivated species, native species, alien species, and local species.

First, we applied the R package “bestNormalize” to transform all variables to achieve the best normal distribution fit [9]. We then used IBM SPSS Statistics 23.1 for z-score normalization [50], filtering data within the range of −3 < z-score < 3. Next, we performed Pearson correlation analysis on the filtered data and used Origin to create a correlation heatmap (Figure S1). Variables with a correlation coefficient greater than 0.5 were considered collinear and excluded. From the collinear variables, we retained NHO (from NHO and NCPS) and WT (from WT, FAT, and MVT).

Using these filtered variables, we performed stepwise regression analysis with the GLM, selecting the model with the minimum AIC value [51]. The GLM regression results confirmed our hypothesis of horizontal and vertical gradient distributions of plant diversity in Hainan’s residential areas, with HOPs and Height as the primary influencing factors for these gradients.

For HOPs, we conducted a Principal Component Analysis (PCA) to explore the differences among the six types of residential areas. PCA reduces dimensionality while preserving the structure of the data, allowing us to classify the areas based on variables retained after correlation tests and total species diversity. This analysis helped identify differences between residential area types. We then used Origin to create box plots comparing the Margalef indices (for total species, trees, shrubs, and herbs) and species richness (for cultivated, native, alien, and local species) across the different residential types. These plots provided a multifaceted view of horizontal gradients in species diversity influenced by economic factors.

Regarding Height, we followed a similar approach by plotting boxplots at different altitudes to examine the vertical gradient distribution. Nevertheless, considering the substantial presence of outliers when categorizing data based on Height, we adopted a strategy of eliminating extreme values to improve the clarity of the boxplots. This approach helped ensure that the visual representation of the data was more accurate and focused on the central trends, reducing the distortion caused by outliers. The significant positive impact of altitude on species richness prompted us to hypothesize that altitude might serve as an intermediate variable. Studies have shown that altitude gradients are influenced by both environmental and socio-economic factors [19]. To test this hypothesis, we used two verification methods:

• Principal Coordinate Analysis (PCoA) was performed to examine the relationship between Height and socio-economic variables (such as DMR, TFV, POP, RDA, HOPs, and GDP), with the results plotted using Origin;
• A path model was constructed in R to hypothesize Height as an intermediate variable affecting species richness. The model incorporated socio-economic variables, with the AIC and BIC values used to select the best-fitting model [52].

3. Results

3.1. Analysis of the Driving Forces

After conducting a stepwise regression analysis of plant diversity in residential areas using a Generalized Linear Model (GLM) (

Table 2. Analysis of GLM driving mechanisms for species diversity in Hainan Island residential areas. The variables denoted by red characters correspond to “Live area characteristics and management”, those in blue represent “Social economy”, and the ones in green signify “Environment”. A value of 0 is used to indicate that a particular variable was not incorporated into the model. Signif. codes: 0 ‘***’, 0.001 ‘**’, 0.01 ‘*’, 0.05 ‘.’, 0.1 ‘ ’, 1.

Driving Force	Total		Tree		Shrub		Herb		Cultivated		Spontaneous		Introduced		Native
BUA	0	0	0	0	0	0	Positive		0	0	0	0	0	0	0	0
PRFs	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
PLR	0	0	Positive	*	0	0	0	0	Positive	*	0	0	Positive	.	0	0
GRR	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
NHO	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
DMR	Positive	*	0	0	0	0	Positive	*	0	0	Positive	**	Positive		Positive	*
WT	0	0	0	0	0	0	0	0	Positive	.	Negative		0	0	0	0
HOPs	Positive	***	Positive	**	Positive	***	0	0	Positive	**	NA	0	0	0	Positive	***
TFV	Negative	*	0	0	0	0	Negative	***	0	0	Negative	***	0	0	Negative	*
GDP	0	0	Negative	NA *	0	0	0	0	0	0	0	0	0	0	0	0
RDA	0	0	0	0	0	0	0	0	Negative	.	0	0	Negative		0	0
POP	0	0	Positive	*	Positive	*	0	0	Positive	*	0	0	0	0	0	0
HEIGHT	Positive	***	0	0	Positive	*	Positive	***	Positive	*	Positive	***	Positive	***	Positive	*
NDVI	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
AIC	−52.52		−29.47		−60.56		−53.63		−45.69		−52.97		−39.9		−44.99
Adjusted R-squared	0.13		0.09		0.08		0.16		0.15		0.18		0.13		0.09

), the model fit was found to be generally poor, with the best-fit model yielding an AIC of −52.97 and an Adjusted R² of 0.18. Despite subsequent efforts to validate the outcomes using multiple models, including Generalized Additive Mixed Models (GAMM), Ridge Regression, Random Forest, Gradient Boosting Machine (GBM), and Structural Equation Modeling (SEM), the Adjusted R-squared values of the best-fitting models among these five were still lower than that of the Generalized Linear Model (GLM) (Table S2). Since the GLM provided the best fit among the models tested, we speculate that this may be due to the inherent distribution characteristics of the data, suggesting that the models may not fully capture the underlying complexity of the plant diversity in residential areas.

For the total species diversity in residential areas, we observed highly significant positive correlations (p < 0.001) between HOPs (house prices) and Height with species richness. Additionally, DMR (distance from the main road) and TFV (traffic-flow volume) showed varying effects, with DMR having a positive and TFV a negative correlation with species richness. These results underscore the important roles that both economic and environmental factors play in influencing species diversity in residential areas.

Looking at specific plant categories, HOPs were positively correlated (p < 0.01) with tree species richness and showed a significant correlation with shrubs (p < 0.001), though no significant relationship was observed for herb species richness. POP (population density) had a weak positive correlation with both trees (p < 0.05) and shrubs (p < 0.05), but no significant correlation with herbs. Height showed a positive correlation with shrub- (p < 0.05) and herb-species richness (p < 0.001), but not with trees. Additionally, TFV was found to negatively impact herb diversity (p < 0.001), highlighting the role of urban traffic dynamics in shaping plant communities.

When analyzing cultivated versus spontaneous species, significant differences in the driving factors were observed. Notably, DMR had a stronger positive effect on spontaneous species (p < 0.01) than on cultivated species, where no clear relationship emerged. HOPs had a significant positive correlation (p < 0.01) only with cultivated species, pointing to the influence of housing prices on ornamental or edible plant choices. TFV was found to have a significant negative impact on spontaneous species diversity (p < 0.001), but no such effect was noted for cultivated species. Additionally, the frequency of watering (WT) showed a positive correlation with cultivated species richness (p < 0.05) but had a negative impact on spontaneous species diversity (p < 0.05), further emphasizing the distinction between managed and unmanaged plant types in residential areas.

The differences between introduced and native species were also striking. Height had a positive relationship with both groups, but its impact was more significant for native species (p < 0.001), indicating that altitude may have a more pronounced effect on the native species composition. DMR showed a stronger positive correlation with introduced species (p < 0.001) compared to native species (p < 0.05), highlighting the influence of urban infrastructure on the distribution of non-native plants. Interestingly, the diversity of native species significantly increased with Height (p < 0.001), suggesting that higher altitudes may offer more favorable conditions for native plant species.

Our results confirm the presence of both vertical and horizontal gradient distributions of plant diversity in the residential areas of Hainan Island. Specifically, Height demonstrated a vertical gradient in species richness, with higher altitudes supporting a greater diversity of native species. On the other hand, economic factors, particularly HOPs, were found to drive a horizontal gradient in plant diversity, creating a pattern where residential areas with higher housing prices had greater species diversity. This finding aligns with our hypothesis that economic factors shape plant diversity both vertically (via altitude) and horizontally (via house prices) across different residential areas.

An analysis of cultivated composition in various residential areas (Figure 3) revealed that, except for villages (where the proportion of edible plants exceeded ornamental plants), all other residential types exhibited the opposite trend, with ornamental plants outnumbering edible ones. The village had the lowest Orn/Cul ratio (ornamental to cultivated plants) among all residential types, while hotels and villas had the highest Edi/Cul ratios (edible to cultivated plants). These findings highlight the differences not only in species diversity but also in the types of plant species utilized in different residential areas. Economic factors, particularly housing prices, appear to influence the utilization of plant species, with high-end residential areas tending to favor plants with ornamental or edible value.

3.2. Horizontal Gradient Analysis of Plant Diversity: House-Price Gradient

In the previous analysis, we found that HOPs (housing prices) have an extremely significant positive impact on the overall species richness of residential areas, suggesting that as housing prices rise, plant diversity increases. This indicates a clear horizontal gradient distribution of plant diversity across residential areas based on housing prices. To further explore this relationship, we conducted stratified sampling across six different types of residential areas: low-rise, high-rise, mid-rise, villages, villas, and hotels. Based on our data collection and analysis (Table S3), we observed that housing prices follow this increasing order: villages < low-rise < mid-rise < high-rise < villas < hotels.

To further understand whether these differences in housing prices lead to significant variations in plant diversity across residential area types, we performed a Principal Component Analysis (PCA). As shown in Figure 4, the differences between the six types of residential areas fall within a reasonable range, validating our original aim of exploring whether the variations in plant diversity are indeed influenced by HOPs or if other factors are at play.

Examining Figure 5 and Figure 6, we observe that the villa residential areas exhibit the richest plant diversity across most plant characteristics, except for shrub plants. In contrast, village residential areas have the poorest plant diversity, with the only exception being the diversity of native species. Although the plant diversity rankings do not strictly align with the housing price order we proposed earlier, the general pattern still holds that higher housing prices correspond to greater plant diversity in residential areas. This trend is evident across several categories, with herb plants being more diverse than shrubs and trees, introduced species more diverse than native species, and cultivated species more abundant than spontaneous species.

An additional point to note is the use of the Margalef index, which yielded a more stable and consistent distribution of the overall data compared to using raw species diversity. The Margalef index helped minimize variance, thus making the results more robust and convincing. This analysis supports our conclusion that plant diversity in residential areas is positively influenced by higher housing prices, with villas showing the highest diversity and villages showing the lowest.

3.3. Vertical Gradient Analysis of Plant Diversity: Altitude Gradient

One of the critical issues in our study is understanding how Height influences plant diversity in residential areas. The driving mechanism analysis revealed a highly significant positive correlation between Height and species diversity, confirming the presence of a vertical gradient in plant diversity. However, the nature of this increase—whether linear or logarithmic—remains unclear. To explore this, we classified the 238 sample plots into five altitude categories: less than 10 m, 10–15 m, 15–20 m, 20–80 m, and above 80 m. From the box plots (Figure 7), it is evident that, except for tree species, plant diversity is significantly richer in plots above 80 m. Below 80m, species diversity shows minimal variation, suggesting that the diversity increase with altitude primarily occurs above this threshold. This pattern explains why the overall impact of Height on plant diversity appears relatively small.

To address potential sampling bias, we analyzed the distribution of sample plots across elevation ranges. Plots above 80 m were concentrated in Danzhou City, Wuzhishan City, Sanya City, Ledong County, Chengmai County, and Baisha Li Autonomous County, with over half located in Danzhou and Wuzhishan. This uneven distribution raises the possibility of sampling bias, warranting further investigation. Additionally, we propose an alternative hypothesis: the influence of Height on species richness may not be direct but rather an indirect effect mediated by other variables.

To test this hypothesis, we conducted a PCoA analysis to examine the relationships among various driving factors (Figure 8). Height exhibited the closest positive correlation with TFV (traffic-flow volume), as indicated by the smallest angle between them. Conversely, Height showed strong negative correlations with GDP, HOPs (housing prices), and RDA (road network density), as evidenced by the largest angles. These results suggest that in the selected residential areas, Height is positively associated with TFV but negatively correlated with socio-economic indicators such as GDP and housing prices.

To further investigate, we constructed a path model with species richness as the final variable, Height as an intermediate variable, and driving factors such as DMR, TFV, POP, RDA, HOPs, and GDP. Two models with the best explanatory power are shown in

Table 3. Results of AIC best model and BIC best model in path analysis.

Model	Height			Richness
	Independent Variable	Coefficient	p-Value	Independent Variable	Coefficient	p-Value
AIC (1464.59)	TFV	0.1628	**	Height	0.241	***
	GDP	−0.1207	.	TFV	−0.1231	*
	RDA	−0.0164		GDP	0.1348	*
	HOPs	−1.51 × 10−5		RDA	−0.0264
				HOPs	0.00003703	*
	Adjusted R-squared	0.05		Adjusted R-squared	0.10
BIC (1490.64)	TFV	0.1691	**	Height	0.24	***
	HOPs	3.77 × 10−5	***	TFV	−0.1312	**
				HOPs	3.772 × 10−5	***
	Adjusted R-squared	0.04		Adjusted R-squared	0.09

Signif. codes: 0 ‘***’, 0.001 ‘**’, 0.01 ‘*’, 0.05 ‘.’, 0.1 ‘ ’, 1. Coefficient represents partial regression coefficient.

. The best AIC model (Adjusted R² = 0.10) revealed that Height remains the most influential variable on species diversity (coefficient = 0.241, p-value < 0.001), followed by TFV, GDP, and HOPs. Height also showed a significant positive correlation with TFV (coefficient = 0.1628, p-value < 0.01), while GDP, HOPs, and RDA exhibited negative correlations. In contrast, the best BIC model (Adjusted R² = 0.10) indicated that HOPs had a surprisingly strong positive correlation with Height (p-value < 0.001), although its impact was weaker than in the AIC model. TFV also showed a significant negative correlation (coefficient = −0.1312, p-value < 0.01).

From the combined results of the PCoA and path analysis, we conclude that Height is closely linked to economic variables, with TFV demonstrating a reliable positive correlation in both models. In the AIC model, GDP, HOPs, and RDA were negatively correlated with Height, while the BIC model showed a strong positive correlation between HOPs and Height. The AIC model aligns more closely with the results of the PCoA analysis, supporting the hypothesis that TFV is the primary economic variable associated with Height. Overall, our findings confirm that Height indirectly influences species diversity through its relationship with socio-economic factors, particularly TFV.

4. Discussion

4.1. Differences in Driving Mechanisms of Plant Species Diversity

In this section, we examine the differences in driving factors among various plant types, while the commonalities will be addressed in a subsequent section. Understanding these differences is crucial, as recognizing the morphological and environmental functional variations in these plants is essential for fostering a harmonious relationship between humans and nature [53].

When classifying trees and shrubs, we observe that only trees exhibit an increase in size with the rise in the PLR, whereas shrubs and herbaceous plants do not demonstrate this trend. This phenomenon can be attributed to the morphological differences among these plants, as trees typically possess larger canopies compared to shrubs and herbaceous plants [54]. Large crowns can negatively impact neighboring streets and reduce light availability to buildings [55]. In areas with higher PLR, which indicates lower building density, selecting trees as the primary greenery is more appropriate due to their extensive canopy coverage.

Regarding cultivated and spontaneous plants, we are particularly interested in why POP distinctly impacts only cultivated species. In contemporary urban settings, regions with high population density often signify urban settlements where human activities exert a greater influence. This leads to a higher prevalence of cultivated species that serve human needs, such as ornamental plants and street trees [56]. The intense human presence and management in these areas favor the proliferation of plants that are intentionally planted for esthetic or functional purposes, thereby differentiating them from spontaneous species that grow without direct human intervention.

In the context of introduced and native species, our findings reveal that introduced species are adversely affected by RDA, which contradicts existing published research. Typically, non-native species tend to have a competitive advantage in areas with high road density due to increased disturbance and opportunities for dispersal [57]. This unexpected result suggests that in our study area, other factors associated with high road density may be inhibiting the success of introduced species. This discrepancy highlights the need for further investigation to understand the underlying mechanisms influencing the distribution of introduced versus native species in different urban contexts.

Combining our findings with Figure 3, we gain additional insights into species utilization across different residential area types. In urban areas—comprising villas, high-rise, multi-floor, small high-rise buildings, and hotels—cultivated species are extremely abundant. These plants are often planted as street trees or ornamental flowers, such as fox tail palms, bougainvillea, and plumeria. In contrast, rural areas like villages exhibit relatively sparse cultivation of species, which often compete ecologically with native species. In villages, cultivated plants are typically planted in backyards to meet farmers’ consumption needs or for sale, including species like betel nut, cassava, and rubber.

This pattern is a consequence of our long-term research on the residential areas of Hainan Island and likely extends to many cities worldwide. We interpret this phenomenon as an extension of the luxury effect, where economic conditions influence urban ecology by driving subjective modifications based on the needs and preferences of residents with varying economic statuses. This represents a typical case of the interaction between economy and urban ecology, where higher-income residential areas favor the cultivation of ornamental and economically valuable plants, thereby enhancing plant diversity.

4.2. Horizontal Gradient Analysis of Plant Diversity: House-Price Gradient

Current research has achieved significant theoretical advancements in analyzing differences in the spatial scale of plant diversity [25,58]. In our results section, we conducted a detailed analysis of two key factors: HOPs and Height. By categorizing our sample plots based on these factors, we uncovered spatial distribution differences that offer valuable insights into plant diversity patterns within residential areas.

Among the six types of residential areas classified according to HOPs, villas consistently exhibited the highest species diversity, while villages were the most species-poor. This observation aligns with the luxury effect, a phenomenon we have emphasized multiple times throughout our study, which posits that higher economic status areas tend to support greater biodiversity. Given that this finding reinforces our previously discussed hypothesis, we will not elaborate further on it here.

However, it is important to note that hotels often differ significantly in terms of location compared to other residential areas. Hotels are more likely to be situated in tourist-centric areas, which can interfere with their positioning in this ranking. Additionally, the average price difference between multi-story buildings (low-rise, mid-rise) and small high-rise buildings is relatively small, which could influence the variability within certain categories. Despite these factors, the overall trend suggests that higher housing prices correlate with higher plant diversity.

Some unexpected patterns emerged when examining specific plant categories such as herbaceous plants and native species. Unlike the general trend observed with total species diversity, high-end residential areas like villas did not show a significant advantage in these categories. Surprisingly, villages demonstrated relatively high species diversity for native species, with the richness of native species being comparable to that of cultivated species.

Our hypothesis, inspired by Rodenburg et al. [59], suggests that the reduction in native species is often primarily caused by agriculture and construction. However, in our study, the residential areas surveyed had fewer intersections with farmland. Instead, we propose that the high building densities in most residential types, except for villages, lead to significant habitat fragmentation. This fragmentation greatly reduces the living space available for non-cultivated species, thereby decreasing overall species diversity [60]. Consequently, the ecological niches of native species are rapidly replaced by introduced alien species. This replacement explains the predominance of cultivated and alien species in the residential areas of Hainan Island [56].

Building on the previous section and the relevant studies introduced when HOPs were first discussed, we have strong evidence to suggest that HOPs in this study serve as a proxy for the luxury effect. Fluctuations in HOPs reflect changes in plant diversity, as both are influenced by a range of socio-economic variables. This viewpoint is supported by numerous studies. For instance, the positive correlation between rising housing prices and tree management and greening levels has been empirically demonstrated in the streets of New York City [61]. Additionally, the land-use structure of communities plays a significant role in housing price fluctuations [62]. Furthermore, many studies have shown that greening management and land-use structure impact changes in plant diversity [40,63]. These complex interrelationships drive the emergence of housing price gradients.

These findings highlight the complex interactions between economic factors, urban development, and plant diversity. While higher housing prices are generally associated with greater plant diversity, specific plant categories such as herbs and native species may respond differently to urbanization pressures. Understanding these nuanced relationships is essential for developing strategies to preserve native biodiversity and manage introduced species in urban environments

4.3. Vertical Gradient Analysis of Plant Diversity: Altitude Gradient

Dividing residential areas based on elevation has also provided new insights into plant diversity. Our results show that only plots situated above 80 m exhibit a sharp increase in species richness, with this pattern observed across nearly all plant types. Related research on tree diversity in Hainan Island suggests a negative correlation with elevation [64]. However, this research often focuses on wild plants rather than residential area plants, prompting questions about the reliability of the observed elevation-related change.

As discussed earlier in our study, we cannot completely rule out the possibility that the differences in species richness at varying elevations could be attributed to sampling bias. Specifically, the sampling teams in Danzhou and Qionghai may have recorded more plant species or made procedural errors during data collection. Therefore, we maintain a skeptical stance on whether elevation is truly driving these differences in species richness.

To investigate whether elevation’s effect on species richness is independent, we analyzed potential confounding factors. The results suggest that Height (elevation) is influenced by economic factors, particularly TFV. Simultaneously, TFV itself is generally negatively correlated with species diversity. To further explore this, we used partial regression coefficients to calculate the indirect impact of TFV on species diversity through Height, as well as the direct impact of TFV on species diversity, comparing their magnitudes. The results show a clear negative correlation between TFV and species diversity.

The negative correlation between GDP, HOPs, RDA, and Height further supports the view that residential areas at higher elevations typically correspond to poorer socio-economic conditions. This relationship is consistent with findings from big data studies in China, which suggest that traffic volume characteristics (such as vehicle flow) can explain regional economic development indicators [65]. In our study, we observed a micro-environment in residential areas on Hainan Island that differs from ordinary road sections. This micro-environment is primarily driven by Road Traffic Nuisance, referring to the physical pollution and noise disturbance caused by public transportation in residential areas [66]. This phenomenon is especially pronounced in areas with poor noise management capabilities, which tend to be located in higher-altitude areas with relatively lower economic levels (e.g., Danzhou and Wuzhishan). This may explain why TFV is positively correlated with Height, as poorer noise management in high-altitude areas allows traffic volume to increase, thereby negatively impacting biodiversity.

Regarding the positive correlation between Height and species richness, we believe it is still related to the combined effects of GDP, HOPs, and RDA. In areas with lower economic development, land use for construction typically exerts less destructive impact on biological habitats, allowing the original ecological environment to be better preserved. This contributes to greater species richness and higher ecological diversity in these areas.

A somewhat subjective observation further enriches this argument: surveyors reported more mosquito bites in higher-altitude areas compared to lower-elevation regions, suggesting richer biodiversity, as mosquitoes are often more prevalent in diverse ecosystems. While economic variables provide valuable insights, environmental factors also play a significant role in the observed patterns of species diversity at varying elevations. Several environmental processes associated with increasing altitude likely contribute to greater species richness. Firstly, lighting conditions improve with altitude, enhancing photosynthesis and potentially supporting a wider variety of plant species. Secondly, higher altitudes tend to experience increased precipitation and humidity, creating more favorable conditions for plant growth. Lastly, intensified weathering and erosion at elevated levels lead to more fertile soils, which support diverse plant communities [67]. These combined environmental factors, alongside socio-economic influences, highlight the complex interplay shaping plant diversity in the residential areas of Hainan Island and underscore the need for comprehensive studies that consider both economic and environmental variables to fully understand and manage urban biodiversity.

4.4. Gradient Differences

While the factors influencing plant diversity have been extensively documented in broader ecological studies, their specific relevance to residential plants on Hainan Island requires further verification. Our analyses reveal that plant diversity in the residential areas of Hainan Island can be classified into two distinct structures: horizontal and vertical. Both structures are influenced by similar socio-economic factors but yield different results due to their unique observational focuses. The horizontal structure examines variability among different types of residential areas, such as high-end versus low-income neighborhoods, emphasizing how the socio-economic status of residents impacts biodiversity—akin to comparing the biodiversity of affluent neighborhoods with that of slums within the same city. In contrast, the vertical structure assesses variability associated with regional development, focusing on differences between economically developed coastal cities and less economically dynamic inland areas, similar to comparing a rapidly growing coastal city with an inland city experiencing lower economic momentum. In both structures, the underlying theme is unbalanced economic development. Differences in plant diversity are largely attributed to the varying levels of economic development across Hainan Island, where areas with lower socio-economic status (e.g., higher-altitude regions) tend to have better-preserved ecological environments. Conversely, more developed areas (e.g., coastal cities) experience extensive habitat destruction and urbanization, leading to lower species richness. These findings underscore the critical role of economic development in shaping urban biodiversity and highlight the need for targeted ecological planning to address the disparities in plant diversity across different residential and regional contexts.

4.5. Issues and Suggestions

In the current wave of urbanization, urban greening is widely recognized as a vital strategy to enhance residents’ quality of life and promote ecological balance. However, the management of urban vegetation faces significant challenges and shortcomings that impede its effectiveness. On the one hand, rapid urban expansion often comes at the expense of natural vegetation, resulting in a reduction in urban green spaces and a noticeable trend towards monoculture in vegetation types. This loss of diverse plant communities diminishes the ecological resilience of urban areas and reduces the benefits that diverse green spaces can provide. On the other hand, the management and maintenance of urban vegetation frequently lack systematic and scientific planning. Instead, these efforts often rely solely on experience-based approaches or aim to achieve short-term goals. Such practices neglect the essential aspects of vegetation diversity protection and restoration, as well as the maximization of ecological functions that diverse plant communities can offer. Without a strategic framework, urban vegetation initiatives may fail to sustain long-term ecological benefits and resilience. Additionally, urban vegetation management is confronted with multiple environmental and logistical issues, including improper water-resource management, soil pollution, and the invasion of alien species. Water-resource management is critical for the health of urban plants; yet, it is often inadequately addressed, leading to either water scarcity or excessive use. Soil pollution from urban runoff and industrial activities can degrade soil quality, adversely affecting plant growth and diversity. The invasion of alien species poses a particularly insidious threat to urban ecosystems. These non-native species can outcompete indigenous plants, disrupt local ecological balances, and reduce overall biodiversity. Despite their significant impact, the issue of invasive species in urban ecosystems remains understudied. For biogeographers and ecologists, understanding the processes and impacts of invasive species, as well as developing effective control strategies, is a critical challenge that demands urgent attention [68].

4.6. Implications and Future Research

Future studies should explore the complex interactions among multiple factors influencing urban plant diversity, including both human management practices and environmental conditions. Although economic factors such as housing prices and population density play a significant role, these relationships are often oversimplified and fail to account for the intricate dynamics between various drivers of biodiversity. For instance, green space management, microhabitat availability, and local environmental conditions (such as soil quality and water availability) can have profound effects on plant diversity, which warrants further investigation. Additionally, the role of urban planning and green space design, particularly in mitigating habitat fragmentation, should be more thoroughly explored. Research should also consider how road traffic and other disturbances impact biodiversity, particularly in terms of non-native species, which often thrive in areas with high disturbance but may show unexpected patterns in urban environments. Expanding the range of environmental variables and examining the dynamics between native and non-native species across different residential areas will provide a more comprehensive understanding of the drivers of plant diversity in urban ecosystems.

5. Conclusions

The acceleration of urbanization worldwide has presented humanity with the dual challenges of biodiversity loss and the diversification of landscape forms. In response, this study evaluated the characteristics and driving mechanisms of residential areas on Hainan Island by selecting residential zones from fifteen counties and cities within Hainan Province. We investigated the diversity of related plant species alongside thirteen variables encompassing live area characteristics and management, socio-economic factors, and environmental factors to thoroughly analyze the underlying driving mechanisms. Our analysis identified distinct differences in the driving factors influencing various plant types and established two primary gradients of plant diversity distribution within residential areas: a horizontal gradient influenced by housing prices and a vertical gradient corresponding to changes in elevation. Both gradient models were found to be outcomes of socio-economic factors, highlighting the significant role of economic development in shaping urban biodiversity. These conclusions offer valuable suggestions for the greening management and urban planning of residential areas on Hainan Island. However, the study acknowledges limitations, particularly in its examination of altitude, as it lacks a comprehensive consideration of environmental variables and does not address critical issues, such as the impact of invasive species. Therefore, future research should aim to incorporate these aspects to provide a more holistic understanding of the factors influencing plant diversity in urban residential areas.