Distribution and Change Characteristics of Ecosystem Services in Highly Urbanized Areas along Gradients of Human Activity Intensity: A Case Study of Shenzhen City, China
[ad_1]
1. Introduction
Human activity intensity (HAI) refers to the degree of disturbance caused by human activities in a certain area [1]; as a universal phenomenon in nature, HAI directly or indirectly affects the succession of ecosystems [2]. Ecosystem services (ESs) are defined as benefits or contributions provided by natural ecosystems that can be utilized by humans [3,4]; these services are the material basis for human survival and development, and are the bridge between natural ecosystems and human beings [5]. In recent years, the United Nations 2030 Agenda and the Sustainable Development Goals (SDGs) have aimed to reduce the impact of human activities on the ecological environment and improve human well-being [6]. It can be seen that understanding the relationship between HAI and ESs is an urgent problem that must be solved, in order to achieve the sustainable development of human society [7]. Currently, rapid population growth and urbanization have led to the substantial shrinkage of ecological land and serious environmental pollution, both of which have brought severe environmental pressure to the global ecological environment [8]. The impact of this will continue to expand and intensify in the future, with its obvious gradient characteristics, long-term evolution, and adaptation process [9]. Therefore, exploring and quantifying the spatial characteristics of ESs along gradients of HAI can provide theoretical guidance for the rational control of human activities and maintenance of service functions.
With the development of spatial analysis technology, the traditional non-spatialization methods (i.e., statistical analysis methods) are gradually facing more difficulty in meeting the needs of various types of applications; research on the spatialization of HAI has begun to develop, and various methods of direct or indirect spatialization have been put forward and widely used [10]. First, indirect spatialization methods include the vegetation index method [11], the human appropriation of net primary production (HANPP), and the global disturbance index method. Among these, the HANPP method has better evaluation results, can also reflect the positive impacts of human activities, and has been applied at different scales [12]. However, the method needs to be coupled with statistical data, resulting in the insufficient spatial refinement of the data [13]. The global disturbance index method couples surface temperature with the enhanced vegetation index, in order to characterize the intensity [14]. Nonetheless, this method makes it difficult to isolate the human factors, making it more suitable for large-scale vegetation belts. Second, direct spatialization methods include the land use type change method, the comprehensive indicator method [15], and the human footprint index method. Among these, the land use change method evaluates HAI using a model based on construction land, and then spatially expresses the evaluation results [10]. It has been applied at different spatial scales. The comprehensive indicator method [16] considers that HAI is the result of the combined effects of natural environmental conditions and human socio-economic activities. However, the data prepared by this type of method mainly take the county as the smallest spatial unit, with poor spatial refinement, and this applies to regions where administrative units are the object of study [17]. The human footprint index method was proposed by Sanderson and other scholars [18], and this method selects spatial data, such as population density, land use change, and roads, to quantify HAI [19], which can make up for the weaknesses of previous findings that were only calculated using a single indicator of landscape change or land use change. In general, the study of the human footprint index method has made some progress; this method is more suitable for quantifying HAI in land systems at a grid scale [20]. However, in quantifying HAI, the methodology has a wide accuracy range in terms of data selection, which means it has difficulties generalizing all human activities and may underestimate HAI to a certain extent; thus, it still needs to be improved and optimized [21,22]. Therefore, based on the research results of the human footprint index method, this study fully considers the economic and social development of the high-speed development area, and adds the urban facility density layer (POI data) to represent the spatial distribution of HAI.
Various researchers have conducted more elaborate studies on the classification of ESs, such as Costanza et al. [23], Ouyang et al. [24], Xie et al. [25], and Millennium Ecosystem Assessment [26], all of whom elaborated on the classification system. Among them, MA is an influential classification system, which mainly includes provisioning services, regulating services, supporting services, and cultural services; it is widely acknowledged and employed by scholars in the academic field [27]. Existing studies have found that ES evaluation has mainly focused on the evaluation of service provision and regulation, with many studies on land use change and ESs, but few studies on support and cultural services [28]. In the meantime, ES evaluation methods have undergone a long development process from simple ES measurement (quality and value) to the current composite model method [29]. In recent years, research on ESs has become widespread in countries around the globe, focusing mainly on service frameworks, developing value assessment methodologies, and analyzing internal trade-offs [30]. From the perspective of research objects and spatial scope, the research objects regarding ESs have become broader, their spatial scope has expanded, and their application fields have become more diversified. The research objects involve ecologically fragile areas [31], river basins [32], hills and mountains [33], etc. With the rapid development of remote sensing technology, InVEST, ARIES, MIMES, and other models have been applied for evaluating ESs [34], but the emphases of these models are different. Among them, the InVEST model is the most systematic; not only does it have a strong spatial analysis function, but its low application cost, strong open source, high assessment accuracy, etc., are also superior to other models [35]. Therefore, it is widely used in current ecosystem assessment models.
Changes in HAI affect the changes in the ecosystem structure, function, and service, posing a threat to the ecosystem’s balance and the sustainable development of society and the economy [2,36]. However, due to the complexity and uncertainty of human activities [37], the interaction mechanism between HAI and ESs is still being explored. Currently, many studies have discussed the changing trend in the relationship between HAI and ESs from different perspectives [20,38] and have reached different conclusions. For example, Qi et al. [39] used the Sanderson human footprint model and InVEST model to measure human activities and ecosystem services in the Qinghai Lake Basin, China, and simulated the influence curves of human activities on six types of ecosystem services, including carbon storage, habitat quality, soil erosion, and water yield, through the generalized additive model. This model also reveals the evolution of complex nonlinear relationships between human activities and these types of ESs. Existing studies show that most studies start from the perspective of exploring the correlation between them, and often fail to fully and specifically reveal their roles under different gradients. In terms of research methodology, scholars have further explored the quantitative correlation between human activities and environmental systems with the help of coupled coordination models [20], while less research has been conducted on the spatial correlation between HAI and ESs, as well as on the spatial distribution of the degree of correlation. There needs to be more research on whether a spatial autocorrelation exists between HAI and ESs [1], what kinds of characteristics that spatial aggregation may present, and whether the spatial relationship between the two evolves with spatial and temporal changes. In real contexts, there are different scales of action for the effect of HAI on ESs, and thus a single bandwidth cannot well reflect the spatial process of the action intensity between variables [40]. Based on this, Fotheringham et al. [41] proposed the use of multiscale geographically weighted regression (MGWR) for estimating the spatial bandwidths of variable differentiation based on traditional geographically weighted regression (GWR), which better solves the defects of a single bandwidth of the GWR model, and has become an important method for revealing the impacts and the scale of the effects between variables. Therefore, this study adopts the bivariate spatial autocorrelation method and the MGWR model to explore the relationship and clustering patterns between human activities and ESs under different gradients.
Thanks to the depth of research on this topic, the theoretical and methodological system for quantifying ESs and HAI has gradually improved. However, there are some shortcomings: (1) more attention has been paid to the study of ecologically fragile areas, but fewer studies have focused on areas with higher levels of economic development; and (2) most of the existing studies are limited to the quantitative relationship between HAI and ESs, which, to some extent, limits the disclosure of their complex interactions from a gradient perspective. In coastal cities with rapid economic development, such as Shenzhen, China, human development activities are more frequent and intense due to the need to source land from the sea in order to meet the social development requirements. ESs face a significant challenge, which greatly hinders the sustainable development of the economy in the Greater Bay Area [42]. In summary, we studied the distribution of HAI and ESs in Shenzhen by analyzing spatial and temporal geographic data with the InVEST model, human footprint index method, MGWR model, and the bivariate spatial autocorrelation model. We also aimed to divide HAI into multiple gradient bands from weak to strong, and explore the relationship between different gradients of HAI and ESs. The research objectives are as follows: (1) to understand the spatial and temporal characteristics of ESs and HAI in economically developed areas and (2) to explore the relationship between gradient differences in ESs and gradient changes in HAI. This study overcomes the deficiencies of existing research findings and explores the spatial relationship between HAI and ESs from a gradient perspective, in order to provide data support and a scientific basis for the sustainable development in high-utilization urban areas.
4. Discussion
4.1. Validation of ES Assessment Results
Currently, InVEST is used as an evaluation model to evaluate ESs; it has been widely used on various scales, and its scientificity and rationality have been verified [35,49]. Hence, this study used this model to carry out ES assessments. To further ensure the quality and accuracy of the ES assessment results, the NPP-based CASA model was selected for verifying the accuracy of the estimation results for water conservation, soil conservation, and habitat quality (ecosystem service diversity) [50,51]. In light of the different analysis results, we discuss the results from a spatial distribution perspective. The evaluation results of ESs in 2010 and 2020 were obtained using the InVEST model in this study. Among these results, water conservation was lower in the central and western regions, and higher in the southeast regions. Secondly, the spatial distribution of soil conservation in the east was significantly higher than that in the west, and was significantly higher in the south than in the north. Moreover, the distribution of biological mass gradually increased from the northwest to the southeast. This is basically in line with the research results of Chen et al. [52]. In summary, the resulting data regarding water conservation, soil conservation, and habitat quality in Shenzhen, as assessed using the CASA model, are similar to the spatial distribution results of high and low values of various ESs simulated using the model, indicating that the simulation results of ESs in the high-speed urbanization region obtained using various models are highly reliable. They can reflect the actual situation of ES changes [53]. The differences in values may be related to the spatial and temporal scales of the selected study area and the setting of model parameters. In addition, in terms of carbon sequestration, the parameters in the InVEST model of this study were determined by referring to the existing carbon density results of the Guangdong–Hong Kong–Macao Greater Bay Area [54]. To a certain extent, this also ensured the scientific results of this study.
4.2. Validation of HAI Results
Currently, the human footprint index method selects spatial data directly related to human activities (land use, population density, night lighting index, and road network data) for data preparation [19,20]. However, due to the differences in topography and regional development conditions in different regions, this method needs to be adjusted and optimized in different regions. Therefore, this study fully considered the economic and social development of the high-speed development area, and increased the density layer of urban facilities to represent the spatial distribution of HAI in the study area in 2010 and 2020. The areas with higher HAI were located in the western part of the study area, where various development and construction activities exist; meanwhile, the areas with lower HAI were located in the eastern part of the study area, which are mainly areas with higher vegetation cover. The results obtained in this study show that areas with a high level of economic development, low vegetation cover, and predominantly built-up land have high HAI, while areas with a low level of economic development, high vegetation cover, and nature reserves have low HAI. This is very similar to the results of Wang et al. [55] and Huang et al. [20]; to a certain extent, this proves the feasibility and scientificity of the human footprint index model constructed in this study.
In addition, a comparison was made between the various methods used to study the spatialization of HAI. From the point of view of methodological difficulties, the global disturbance index method is the simplest in that it directly uses MODIS product data, followed by the land type change method [56]; however, the global disturbance index method is only applicable to forested areas and can only evaluate the impacts of a few types of human activities, while the land type change method has obvious shortcomings in terms of the accuracy of the evaluation results [57]. From this point of view, the comprehensive index method is the best, followed by the human footprint index method [1], but data prepared using the comprehensive index method have obvious shortcomings in terms of spatial refinement and fineness [58]. In summary, we found that the human footprint index model is more suitable for the spatial quantification of HAI in terrestrial systems.
4.3. Exploring the Relationship between HAI and ESs
The scientific question regarding “the gradient response of HAI to ESs” originates from topographic gradients [59]. Based on existing studies, it has been shown that the qualities of an ecological environment, or its ESs, exhibit a certain regularity as the topographic gradient changes [60]. In addition, human activities are a central factor in ecological environment changes. With an increase in HAI, ESs are drastically perturbed. Moreover, there are significant spatial differences in the impacts of human activities on ESs, mainly negative impacts, which is similar to the results of Zhang et al. [36] and Sun et al. [2].
Currently, there are few studies on the gradient response between HAI and ESs and a lack of comparative exploration. In this study, the MGWR and bivariate spatial autocorrelation models [61] were introduced on the basis of gradient division results, and it was found that HAI is negatively correlated with ESs as the gradient changes. Firstly, from 2010 to 2020, the HAI in the study area showed a downward trend in the 1st–3th gradient bands and an upward trend in the 4th–10th gradient bands, indicating that the HAI region is advancing from a low-intensity region to a medium-intensity region. Secondly, in the low-intensity human activity regions (1st–3th gradient bands), the mean values of each ES were higher than the average level in the study area; as the gradient increased in the high-intensity human activity regions (6th–10th gradient bands), the mean values of each ES were at more similar levels. Thirdly, the L-L and H-L aggregation types are mainly distributed in the 1st–3rd gradient bands, while the H-H and L-H aggregation types are mainly distributed in the 6th gradient band, which is mainly because the ES levels in the 6th–10th gradient bands are closer to each other; the 6th gradient band is the area with a high intensity of human activities in which ESs reached a higher level. Finally, from the perspective of geospatial distribution, there were large spatial differences in the effects of ESs on HAI. In the first to third gradient bands, the absolute value of the regression coefficient of ESs on HAI was small, mainly because the region is dominated by forested land and other vegetated land. Such land is weakly affected by HAI, and the emphasis on green protection in ecological barrier areas has made the degree of influence of ESs on HAI in this region relatively stable. In the 6th to 10th gradient bands, the region is dominated by construction land, which is highly affected by HAI; the absolute value of the regression coefficient gradually increased, so that the negative effect of ES weakening, due to the increase in HAI, increased for this region. Therefore, to a certain extent, this shows that ESs can be maintained at a better level under the influence of a medium HAI level.
4.4. Measures Recommendations
In this rapidly developing region of China, frequent human activities affect the changes in the ecosystem’s structure, function, and services, resulting in an overall negative correlation between HAI and ESs [4]. Based on the distribution characteristics of ES changes along HAI gradients and on their functional relationships, corresponding suggestions have been put forward to alleviate the pressure of human activities on ecosystems and promote the sustainable development of highly urbanized areas [54,62]:
In the first to third gradient bands, the L-L and H-L types dominate. Of these, the L-L region should take ecological environment protection and restoration as its priority, avoid excessive development and construction activities, and promote the overall enhancement and optimization of ESs; the H-L region should continue to take ecological protection as its top priority, follow the principles of scientific planning, ecological priority, and strict protection, and pay attention to protecting ecological barrier areas and strengthening ecological financial support.
In the sixth gradient band, the region is dominated by H-H and L-H agglomeration types. Of these, the H-H region should pay attention to the advantages of high-speed economic development areas in terms of resources, should promote the construction of green spaces in neighboring areas, and should enrich urban ecosystem diversity; the L-H region should take into account ecological protection and resource utilization, optimize urban spatial layout, promote the efficient use of construction land, reduce the ecological land fragmentation caused by human over-development activities, and improve the intensive use of land.
In other gradient bands, HAI was negatively correlated with ESs as a whole. In terms of construction land, in these areas, more attention should be paid to the restoration of the ecological environment in terms of economic development; for non-construction land areas, we should reduce the damage to the urban ecological environment caused by human activities, and integrate the goals of urban economic development and environmental protection.
4.5. Limitations and Prospects
First of all, some of the parameters required by the InVEST model, for calculating water conservation, soil conservation, habitat quality, and carbon sequestration function, were obtained from previous studies, such as vegetation cover and crop management factors, soil conservation measures factors, etc. The values of the parameters in different regions may be slightly biased, which may have affected the calculation results. Secondly, in quantifying HAI, this study simply added up the indicators without considering the mutual influences between the various types of activities, resulting in potential uncertainty in the evaluation results. Meanwhile, this study selected six indicators with the help of spatial and temporal geographic data (e.g., night-time light remote sensing, POI, etc.), which, to a certain extent, satisfies scale unity for each indicator. However, the breadth and precision of data coverage are still urgent issues to be solved. Finally, HAI with different gradients has a bidirectional relationship with ESs. For example, rapid urbanization has a negative relationship with ESs, while ecological restoration projects improve habitat quality and have a positive relationship. However, in this study, only the bivariate spatial autocorrelation model and MGWR model were used to explore the action intensity and aggregation types of ESs and HAI at different gradients. This means that the bidirectional relationship between HAI and ESs may not be well reflected; such a problem could be the focus of future research.
After examining the shortcomings of existing studies, firstly, in future studies, the interactions between ESs and HAI should be analyzed through other relationship types besides spatial ones; additionally, the mechanism of influence of different HAI gradient changes on the interrelationships between different kinds of ES should be analyzed. Secondly, based on the human footprint index model and the InVEST model, data sources should be enriched, data accuracy should be improved, and the accuracy and precision of HAI and ES measurements should also be improved.
[ad_2]
