1. Introduction
Land-use/cover change (LUCC) resulting from the interaction between human activities and the natural environment on temporal-spatial scales is directly expressed in the form of changes in surface landscape patterns [
1]. Landscape patterns are defined as the spatial composition and configuration of land use. However, the rapid growth of industrialization and urbanization has intensified changes in land use and ecological issues, such as landscape fragmentation, occupation of natural habitat, environmental pollution, and loss of biodiversity [
2], leading to a drastic reduction in habitat quality [
3]. Habitat quality refers to the ability of the environment to provide conditions for human sustainable development and is an important support for species diversity and reproduction [
4,
5]. Numerous studies have shown that landscape pattern changes have a strong effect on habitat quality [
6,
7,
8,
9]. For example, landscape continuity, which is important for species to exchange materials, information, and energy flows, can lead to an increase in habitat quality. The increase in patches leads to landscape fragmentation, which is detrimental to animal migration and plant pollen dispersal and leads to a decrease in habitat quality. Most studies have utilized only the traditional linear equation approach to examine the influences of landscape pattern changes on regional habitat quality [
10], ignoring the spatial autocorrelation and spatial spillover properties of habitat quality. However, spatial regression models are able to solve this problem, which can quantitatively analyze the spatial relationship between habitat quality and landscape pattern by considering spatial autocorrelation. The Yellow River Basin, an important ecological barrier in China’s ecological security strategy pattern, plays an important role in biodiversity conservation and healthy ecosystem maintenance. However, rapid socioeconomic development has led to the expansion of construction land and the loss of natural habitats in the Yellow River Basin. Therefore, it is necessary to simulate landscape patterns and analyze the spatiotemporal characteristics of habitat quality. More importantly, quantitative analysis of the spatial relationship between habitat quality and landscape patterns is of great importance for maintaining biodiversity and promoting sustainable development.
In recent years, many assessment methods have been applied to evaluate habitat quality, such Social Values for Ecosystem Services (SolVES), ARtificial Intelligence for Ecosystem Services (ARIES), Multi-scale Integrated Models of Ecosystem Services (MIMES), and Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) [
11,
12,
13,
14,
15]. Among these models, the InVEST model is becoming a popular tool because it is more mature and easier to operate [
16]. For example, Moreira et al. [
17] adopted the InVEST model to assess the conservation status of Azorean natural habitats. Nematollahi et al. [
18] evaluated the roads’ effects on the natural habitats of wild sheep based on the InVEST habitat quality module. Hack et al. [
19] used the InVEST model to evaluate the impacts of built-up areas, roads, and water pollution on habitat quality. To conclude, the InVEST model can be applied to evaluate habitat quality in combination with habitat suitability and human activities and provides more detailed information about biodiversity [
13,
20]. Thus, the InVEST model is used in this study to evaluate the habitat quality of the Yellow River Basin.
Meanwhile, landscape patterns and biodiversity conservation have become one of the most popular issues in landscape ecology. The correlation of land use, landscape patterns, and ecosystems has drawn the attention of international scholars [
21]. For example, Zhu et al. [
22] used gray correlation analysis to explore the correlation between habitat quality and landscape pattern indexes in the eastern Qinghai-Tibet Plateau. Wu et al. [
23] used Pearson’s correlation analysis methods to investigate the influential factors of habitat quality and showed that vegetation cover, intensity of human activity, and land-use change can cause a decline in habitat quality. Yushanjiang et al. [
24] found that landscape pattern indexes were positively and negatively correlated with the ecosystem value in the Ebinur Lake Basin by using multiple linear regression models. However, most studies have ignored the spatial spillover effects of ecosystems, which are influenced not only by their own unit but also by the habitat quality of neighboring units. This will reduce the validity of conclusions. Noteworthy to mention is that most recent studies have begun to explore the spatial association between landscape patterns and ecosystems [
1]. For instance, Zhu et al. [
25] explored the effects of urbanization and landscape pattern changes on habitat quality in Hangzhou by using ordinary least squares (OLS) and geographically weighted regression (GWR) models. Chen et al. [
26] used a multiscale spatial panel regression analysis approach to explore the impact of landscape patterns on ecosystem services. Thus, research on the mechanism of landscape pattern influence on habitat quality is gradually shifting from traditional linear correlation analysis or regressions to spatial econometric models. Quantitative analyses of the spatial association between landscape pattern and habitat quality can help to better understand the impact of changes in landscape pattern on habitat quality.
The abovementioned studies are very important guidelines for advancing habitat quality research, but they are all from the perspective of the past to analyze the spatiotemporal characteristics of habitat quality in the region. There is a growing need to explore the evolution of habitat quality from a simulation perspective, which can provide insights for ecological conservation planning and sustainable development. Cellular automata (CA) is the basis of many land-use simulation models. Researchers have proposed constrained CA, CA-Markov, the Conversion of Land Use and its Effects at Small regional extent (CLUE-S), and Future Land-Use Simulation (FLUS) models by improving the algorithms and techniques of CA and used these methods to predict habitat quality in the future. For example, Ding et al. [
27] used the FLUS model to assess habitat quality changes in Dongying city in 2030 under multiple scenarios. Gomes et al. [
28] simulated land use and habitat quality by using CA in Lithuania. Tang et al. [
29] combined the CA-Markov and CLUE-S models to predict the evolution of habitat quality in Changli city. Li et al. [
30] simulated urban growth and integrated habitat quality by using the SLEUTH model. However, most of the models were simulated based on each meta-cell scale and lacked the ability to simulate the evolution of patches with multiple land-use types. In this study, a patch-generating simulation (PLUS) model developed by Liang et al. [
31] is adopted to simulate landscape pattern change. Compared with other CA-based models, PLUS has higher simulation accuracy and more realistic indicators of landscape patterns, which could provide more accurate quantitative assessment of the impact of landscape pattern on habitat quality.
The Yellow River Basin has been an important part of achieving balanced east-west and north-south development in China and plays an irreplaceable role in overall ecosystem health and biodiversity conservation in China. In the context of ecological protection and high-quality development, the evolution characteristics and spatial relationships of the landscape pattern and habitat quality in Yellow River Basin deserve unprecedented attention. Therefore, taking the Yellow River Basin as an example, this study quantitatively analyzes the spatial relationship between landscape pattern and habitat quality using a simulation approach to (1) simulate the future land use of Yellow River Basin based on the PLUS model and analyze the dynamic changes of the landscape pattern; (2) assess the spatiotemporal characteristics of habitat quality in Yellow River Basin using the InVEST model; (3) identify the spatial clusters of habitat quality and its rate of change from 2005 to 2031 based on univariate spatial autocorrelation; and (4) quantitatively evaluate the effect of the landscape pattern index on habitat quality based on the spatial lag model.
4. Discussion
4.1. Spatiotemporal Characteristics of Habitat Quality and Landscape Pattern
In this study, we examined spatiotemporal evolution characteristics of landscape pattern and its impact on regional habitat quality and then combined the PLUS and InVEST models to predict future habitat quality levels in the Yellow River Basin. The evaluation of current habitat quality and projection of future habitat quality in the Yellow River Basin are of great significance for ecological protection and high-quality development in the Yellow River Basin.
In general, the landscape of the Yellow River Basin was dominated by grassland and unused land. The area of construction land in the east was significantly larger than that in the west. From 2005 to 2018, the area of arable land and grassland decreased, while the area of construction land, forest, water body, and unused land increased. Meanwhile, PD and SHDI increased significantly in the Yellow River Basin, while ED, LSI, AREA_MN, and CONTAG decreased, and COHESION remained basically unchanged. The predicted trend of landscape pattern changes from 2018 to 2031 is basically the same as it was from 2005 to 2018. Although the proportion of built-up land area in the total landscape area of the Yellow River Basin is low, the expansion of built-up land in recent decades has caused the destruction of forest, grassland, water, and other habitat landscapes. This phenomenon is more obvious in the population and economic agglomeration areas in the lower reaches of the Yellow River Basin (e.g., Henan and Shandong, etc.). This is because areas with more intensified human activities and massive land-use changes have rapid population and economic growth and great demands for housing, transportation, and public facilities, leading to the occupation of many natural resources such as grasslands, water bodies, and forests and increasing the degree of landscape diversity and fragmentation.
In addition, we assessed habitat quality in the Yellow River Basin. It was found that the habitat quality of the western and northern Yellow River Basin along the Qinghai-Tibet–Inner Mongolia was relatively high. The main reason is that the region has good natural endowment and less construction occupation. It has also gradually established a nature reserve management system with national parks as the main body, nature reserves as the basis, and various nature parks as supplements. More importantly, due to the intervention of afforestation, water conservation, and other ecological protection measures, the number of forests and water bodies with high habitat quality in this area continues to increase. However, habitat quality was at a low level in the middle and lower reaches of the Yellow River Basin. The population and economic agglomeration effect is more obvious in this region. The urban expansion continues to occupy natural resources, leading to the degradation of habitat quality. The influence of the continuous expansion of construction land on the degradation of habitat quality is mainly the reduction of cultivated land, forest, and other landscape land area [
49]. At the same time, severe landscape fragmentation reduces landscape connectivity and affects the overall regional habitat, which is particularly critical to the quality of regional habitat.
4.2. Impact of Landscape Pattern Change on Habitat Quality
The results showed that the influence of landscape pattern change on regional habitat quality could not be ignored, and its impact direction and magnitude vary largely in different regions. Therefore, it is of great significance to analyze the effect of landscape pattern on habitat quality for regional landscape planning and ecological sustainable development. Spatial regression was used to quantitatively analyze the correlation between landscape pattern index and habitat quality in the Yellow River Basin. The results showed that the change of landscape pattern had an important effect on habitat quality in the Yellow River Basin. Landscape pattern indices (PD, ED, LSI, and cohesion) had significant effects on habitat quality. The regression coefficients for LSI and ED were both positive, indicating that increased LSI and ED improved habitat quality, while the regression coefficients for PD and cohesion were negative, indicating that increased PD and cohesion resulted in decreased habitat quality. Despite the positive contribution of LSI and ED to habitat quality, both LSI and ED values fluctuated during the study period. There was the greatest negative impact of PD on habitat quality, and the PD value increased over time, which was a major factor in the decline of habitat quality in the Yellow River Basin. In general, the effect of PD on habitat quality reduction was greater than that of LSI and ED enhancement, and the increase of PD implied that the landscape was more fragmented, and the landscape connectivity was weakened, which was related to the decrease of biodiversity and habitat quality. Some relevant studies support our findings. For example, Hu et al. used Geographically and Temporally Weighted Regression (GTWR) and Multiscale Geographic Weighted Regression (MGWR) methods to analyze the driving mechanisms of landscape patterns on habitat quality and found that an increase in landscape connectivity in the urban center of Nanjing significantly improved habitat quality, while an increase in fragmentation in high habitat areas reduced habitat quality [
50].
As part of spatial planning and land-use construction in the Yellow River Basin, it is necessary to coordinate the relationship between development and protection to improve regional habitat quality. It is important for the government to maintain the landscape integrity of natural habitats (such as forests, rivers, and wetlands) as much as possible, arrange agricultural landscapes reasonably, and improve the landscape diversity of urban construction areas [
25]. Specific measures can be adopted, including delineating ecological protection red lines and delimiting permanent primary farmland, managing high- and low-quality areas of ecological space [
51], and establishing pocket parks and green corridors.
4.3. Strengths and Limitations
In this study, the PLUS model was used to simulate the LUCC of the Yellow River Basin in 2031, with 2018 as the base period. Numerous studies adopting the CA-based model have focused mainly on improving technical modeling procedures rather than simulating the detailed patches of multiple land-use types that evolve over time. The PLUS model developed by Liang et al. [
31] has a powerful ability to simulate the evolution of land-use types at patch scale. It has been confirmed that the PLUS model has higher simulation accuracy and landscape pattern indicators that were closer to the real landscape than the other CA-based models. This is essential for accurate quantitative assessment of the impact of future landscape patterns on habitat quality and thus the development of policies to manage future land use and landscape patterns in the Yellow River.
At the same time, spatial autocorrelation models and spatial regression models are used to analyze the spatiotemporal characteristics of habitat quality and its response to landscape pattern changes. Various ecological processes often lead to nonrandom spatial distributions of land use, landscape, and biodiversity and show some dependence on spatial patterns. Thus, spatial autocorrelation analysis is crucial for understanding how ecological variables are related and vary in time and space, which can then be used to understand and predict ecological processes and functions. In addition to traditional factors, spatial autocorrelation is also an important factor that influences habitat quality and landscape pattern, but this factor is often overlooked. In previous studies, linear models are often used to analyze the relationship between landscape pattern and habitat quality, which cannot capture the spatial dependence and spillover effects due to spatial autocorrelation. Spatial regression models and spatial autocorrelation models are used to overcome this shortcoming in this study.
However, there are several limitations in this study. First, the InVEST model was used to evaluate the habitat quality of the Yellow River Basin by accumulating the effects of threat factors. Despite this, InVEST does not take into account the interaction between the threat factors, as their cumulative impact on habitat quality is not the same as their simple accumulation [
29]. Second, this study only analyzed the impact of seven landscape pattern indexes that were recognized as significant, while other related landscape pattern indexes were not comprehensively considered.
5. Conclusions
This study analyzes the spatiotemporal characteristics of landscape patterns and habitat quality, explores the spatial association between the landscape pattern indexes and habitat quality, and proposes reasonable suggestions to protect and improve habitat quality from the perspective of landscape pattern protection.
Firstly, the results showed that the landscape of the Yellow River Basin is dominated by grassland and unused land, and the area of construction land in the east is significantly greater than that in the west. From 2005 to 2031, the areas of cultivated land and grassland decreased, while the areas of construction land, forest, water bodies, and unused land increased. Then, it was found that a significant increase in PD and SHDI will occur in the Yellow River Basin, while ED, LSI, AREA_MN, and CONTAG will decrease, and COHESION remains almost unchanged. In general, landscape heterogeneity increases, and landscape connectivity decreases. In addition, the habitat quality in the Yellow River Basin shows a continuous decrease trend during the study period, but the change is not drastic. This is because the landscape pattern evolution has both enhanced and diminished effects on habitat quality, which offset each other to a certain extent. Forests, grasslands, and water bodies have the highest habitat quality among landscape types, while construction lands have the lowest. Finally, a spatial lag regression model was further applied to quantitatively assessed the effects of the landscape pattern on habitat quality. The results show that PD and COHESION have significant negative impacts on habitat quality, whereas ED and LSI have significant positive impacts. Landscape fragmentation due to high PD exerts the most significant negative effect on habitat quality. Therefore, we should consider enhancing the connectivity of habitats in landscape planning and limiting the fragmentation of ecological land caused by the uncontrolled expansion of construction land in order to achieve biodiversity conservation and ecological sustainability.