Impacts of different levels of urban expansion on habitats at the regional scale and their critical distance thresholds

With the rapid development of urbanization, natural habitats in many parts of the world have been seriously damaged by urban expansion. However, urban expansion is a complex process, and the impacts of different levels of urban expansion on habitats at regional scales and their distance thresholds are still unclear. We conducted a study in Hubei Province, China to evaluate the impacts of the expansion of prefecture-level cities and county towns on the quantity, area, and quality of natural habitats and the critical threshold distances affecting habitats. The results show that, at a regional scale, habitat degradation was driven primarily by the expansion of large numbers of county towns, but the expansion of prefecture-level cities affected habitat degradation over greater distances. Specifically, the impact of county town expansion on habitat first increased and then decreased with greater distance from built-up areas, the threshold distances for habitat quantity and quality being approximately 8 km and 80 km, respectively. The impact of expanding prefecture-level cities on habitat showed a similar nonlinear change with greater distance, but the distance thresholds for habitat quantity and quality rose to approximately 40 km and 130 km, respectively. These findings not only reverse the conventional view that the expansion of large cities dominates habitat degradation, but also draws more attention to the influence of the expansion of numerous small county and towns on habitat, when measured at the regional scale. Understanding the distance threshold of particular spatial impacts can be help to inform spatial decision-making with regards to habitat conservation.


Introduction
Over the past half century, the world has experienced unprecedented rapid urbanization [1][2][3]. A metabased study showed that urban land area increased by 58 000 km 2 worldwide between 1970 and 2000, and projected that global urban land cover would increase by a further 43 000 km 2 to 12 568 000 km 2 , by 2030 [4]. Novotny et al [5] also noted that between 2015 and 2030, expansion will remain the most important form of urban growth. With 68% of the world's population expected to live in cities by 2050 [6], urban expansion will continue into the foreseeable future. Urban expansion is the expansion of built-up areas beyond the existing periphery of settlements [7,8].
Numerous studies have revealed that, over the past 30 years, the disorderly expansion of built-up areas has directly encroached on natural habitats, contributing to rapid biodiversity decline [9,10] increased urban heat islands [11], environmental pollution [12], and food crises [13]. For example, Song et al [14] found that, between 1982 and 2016, global land use change resulted in significant forest loss, on the scale of 385 000 km 2 in Brazil, 113 000 km 2 in Argentina, and 79 000 km 2 in Paraguay. Moreover, Jantz et al [15] predicted that from 2005 to 2100, global land use change is expected to reduce natural habitats by between 26% and 58%. Habitat is the natural environment in which an animal or plant usually lives [16], and is the spatial basis for regional ecological security and biodiversity maintenance [17]. Assessing the impact of urban expansion on natural habitats is essential to the study of regional ecological security and sustainable development.
Previous studies have shown that the impact of urban expansion on habitats is reflected by the area, quantity, quality, and degree of biodiversity of habitat patches [18,19]. For example, Zhu et al [20] found that forests in urban agglomerations in the middle reaches of the Yangtze River decreased by 0.9% (1290.9 km 2 ) due to urban expansion between 1990 and 2020. From 1980 to 2020, land use changes in Guangdong Province, China, contributed to average habitat quality decreasing by 0.0351 or 4.83% per year [21]. These studies reveal a significant negative correlation between urban development and habitat. However, it remains unclear how exactly urban expansion affects habitat spatially, and to what extent. Few studies have explored the relationship between urban expansion and biodiversity from the perspective of the urban−rural gradient. Li et al [22] found that on the gradient of 'downtown-suburbs-outer suburbs' in Shanghai, there are fewer forest patches the closer an area is to downtown, and the area of the forest in the city center is far smaller than in other parts of the gradient. Although such studies have revealed the trend of urbanization impacting on biodiversity along a gradient, they have not identified the specific modes that generate this spatial impact. Different urban expansion modes, such as low-density sprawl, compact development, urban intensification, and infill development, can result in different patterns of land use and land cover change, and have different impact on habitat and biodiversity [23,24]. For example, low-density sprawl, the low-density expansion of dispersed urban settlements into surrounding rural and natural areas, generally reduces the size and connectivity of natural areas within a certain distance to the built-up urban area [25], while compact development, which entails denser and more efficient use of land, causes less loss of natural habitats around such areas [26,27]. Further, most studies have been limited to a single city, but the spatial impacts of urban expansion on habitats may extend far beyond the single city scale.
Urban expansion is a complex process, and its spatial impact on habitat obviously differs in different regions [28]. Early scholars summarized various urban expansion patterns based on the spatial morphological changes in built-up areas, such as concentric circles, multicores, and sectors [29,30]. Other urban expansion patterns have subsequently emerged in various regions, including edge expansion, corridor connection, single-core expansion, multicore expansion, internal filling, and external enclaves [31,32]. The diversity of urban expansion modes increases the difficulty of studying the impact of urban development on habitats and, at the same time, underlines the necessity and urgency of understanding the impact of different types of urban expansions on habitats at the regional scale.
Recent studies have found that city size is crucial in how cities expand and how this expansion affects the spatial development of habitat and biodiversity [33,34]. Concepción et al [35] found that species to diversity did not always increase with greater distance from urban areas, as different species displayed different patterns. Sorace et al [36]found that on the gradient of 'city center-inner periphery-outer peripheryscarcely built areas' , species frequency decreased along the gradient in small towns; but in medium-sized towns, the opposite was true. The impact of city size and different expansion patterns on habitats, particularly that of small cities, remains unclear.
Although there may be multiple types of urban growth, such as infilling, edge-expansion and spontaneous growth [37][38][39], in general the impacts of urban expansion on habitats extends from a few kilometers to tens of kilometers from the edge of the built-up area [40,41]. For example, Shi et al [42] found that between 2004 and 2008 the area approximately 9-12 km from the city center of Lianyungang, China constituted the intensive urban growth zone, with in which industrial development encroached on salt marshes and cultivated land. Recent research has suggested that setting critical distance thresholds may be key to striking a balance between urban development and regional biodiversity conservation [43]. By setting such distance thresholds, policymakers and planners can determine the limits of urban expansion in a way that minimize impacts on habitats and biodiversity. For instance, setting up priority protected areas or buffer zones within critical thresholds can minimize the disturbance and impact of urban expansion on the key species around the city [44]. More importantly, where habitats within critical distance thresholds become incorporated into urban areas, it is essential to ensure that their biodiversity and ecosystem service provision area maintained [45]. However, there is little research which explores the critical distance threshold of urban expansion impacts on habitat at the regional scale and better understanding of such thresholds needs to be built.
We conducted this study in Hubei Province, a typical area of diversified urban expansion in China, to evaluate the impact of urban expansion at different levels on regional habitat. We aim to explain how the expansion of large and small cities affects habitat degradation at the regional scale. Specifically, the objectives of this study were to: (1) identify the different impacts of different levels of urban expansion on habitat patches, quantity and quality; and (2) identify key threshold distances of the impacts of urban expansion on habitat.

Study area
Hubei Province covers an area of 185 900 km 2 , accounting for 1.94% of the total area of China. It contains 17 prefecture-level cities and 61 small cities at or below the county level, distributed across three landscape types: mountainous, hilly and plain landscapes. Hubei Province is abundant in natural habitats, including 93 000 km 2 of woodlands, 894 km 2 of grasslands, 13 000 km 2 of lakes and rivers, and 612 km 2 of wetlands. Between 2000 and 2020, the impervious surface areas of prefecture-level cities increased by 917 km 2 , and the impervious surface area of county towns increased by 731 km 2 . These new impervious surfaces are primarily concentrated in the central plain area, particularly in the urban agglomeration dominated by Wuhan (figure 1).

Data collection and processing
In order to identify areas of urban expansion, we first obtained boundary data from the Global Urban Boundary (GUB) dataset for urban built-up areas in Hubei Province in 2000 and 2018. The GUB dataset, published in Environmental Research Letters by Li et al [46], includes the boundaries of all cities with an area of over 1 km 2 , and their surrounding settlements, worldwide. The identification of urban boundaries is based on high resolution (30 m) global artificial impervious area data, and delimited by an automatic extraction algorithm which is able to capture the contour features of urban and rural marginal areas. Since the GUB dataset did not provide boundary data for 2020, we updated the 2018 data through unsupervised classification and manual visual interpretation of 2020 remote sensing image (Landsat 8 OLI, 30 m) to obtain the 2020 urban built-up area boundaries for Hubei Province.
Identification of habitat and its quality was based on net primary productivity (NPP) and normalized difference vegetation index (NDVI) data, which was obtained from the MOD17 and MOD13 products on the NASA MODIS data platform (https://ladsweb. modaps.eosdis.nasa.gov/), respectively, at a spatial resolution of 500 m. Because vegetation cover in Hubei Province varies significantly between summer and winter, the NPP and NDVI obtained for this study were annual average for 2000 and 2020.

Habitat quantity and quality assessment
Habitat quality refers to the ability of an ecosystem to provide suitable conditions for the continued survival of individuals and populations [47], and is the primary basis for habitat identification. In this study, the habitat quality index (HQI) model was used to assess habitat quality, and habitat patches were identified based on the results of this assessment. The HQI model was published by Peng et al [48], and mainly calculates habitat quality through the combined effect of the ratio of NPP and NDVI to their respective regional average levels. The formula is as follows: where HQI i is the HQI, NPP i is the NPP of vegetation in grid i, and NPP mean represents the average value of NPP in the study area. NDVI i is the vegetation index value of grid i, and NDVI mean is the average value of NDVI values in all grids. In this formula, all NDVI values need to be normalized to 0-1 before conducting the calculation.
To evaluate ecosystem quality consistently across the country, the Technical Specification for Investigation and Assessment of National Ecological Status-Ecosystem Quality Assessment (HJ 1172-2021) in China proposes a grading standard for ecosystem quality score based on a large number of field investigations: the best 25% of sites are deemed excellent, those between top 25% to 45% are good; 45% to 65% moderate; the bottom 20% to 35% low, and the bottom 20% are poor. In line with this grading standard, the results for HQI in 2000 and 2020 in this study were organized into five grades, and those which were evaluated as moderate, good or excellent were identified as habitats. Thus, the spatial distribution of habitats was extracted and the number and area of patches of habitat were calculated accordingly.

Analysis of the impact of urban expansion on habitat
To quantify the spatial changes associated with urban expansion, we took the spatial distribution of urban built-up areas in 2000 as the benchmark and used ArcGIS software to perform superpose analysis with the data from 2020 to identify urban expansion.
We analyzed the spatial impact of urban expansion on habitat in terms of quantity, area and quality. Habitat quantity refers to the number of habitat patches, while habitat area mainly refers to the total area and average size of habitat patches. Firstly, the Euclidean distance tool in ArcGIS was used to calculate the linear distance between each habitat patch and the urban expansion area. Then the Euclidean distance calculated was divided into intervals of 2000 m and changes of habitat patch quantity, total area, average size and quality between 2000 and 2020 were measured. To identify both positive and negative impacts of urban expansion on habitats in further detail, those habitats that changed between 2000 and 2020 were further divided into lost and new habitats, which was done by comparing habitat patch distributions in both two years. To identify the quantitative spatial impact of urban expansion on habitat, we analyzed the quantitative relationships between changes in the quantity, area, and quality of habitat patches and their distance from areas of urban expansion, using several regression models in the Origin software, for lost habitats, new habitats and unchanged habitats. Specifically, the linear regression model was used to analyze the relationship between average size of the lost habitat patch and the distance from areas of urban expansion, according to the distribution characteristics presented by the data. The piecewise linear regression model was used to analyze the relationship between habitat quality ratio of the unchanged habitats and the distance from areas of urban expansion. Habitat quality ratio in this study refers to the ratio of the quality of habitat patches in 2020 to that in 2020, which can reveal the change of habitat quality. The formula is as follows: where, R h is the habitat quality ratio. HQI 2020 and HQI 2000 are the HQIs in 2020 and 2000, respectively.
When R h is greater than 1, it indicates that habitat quality has improved from 2000 to 2020.
Considering the nonlinear distribution characteristics of the data, the polynomial regression models were used to analyze other relationships between changes of habitat patches and the distance from areas of urban expansion. The key inflection points of the fitted function curves were calculated using the Peak Seeking tool in Origin to identify critical distance thresholds for the impact of urban expansion on habitat. Based on these steps, we analyzed the impact on habitat of urban expansion in prefecture-level cities and county towns separately, and for all cities collectively.

Impact of urban expansion on habitat quantity
The spatial impact of county town expansions on degraded habitat patches was consistent with the impact of all cities in Hubei Province, indicating that county town expansions were the dominant driver affecting the degradation of habitat patches at the provincial scale ( figure 2). The number of lost habitat patches peaked at approximately 8.03 km away from the area of expansion around county towns. Although the expansion of prefecture-level cities did not dominate the spatial extent of habitat degradation regionally, its impact on the quality of habitat patches was much greater than that of the county towns, and extended much further approximately 130 km from the area of urban expansion. Moreover, in contrast to the impact of county town expansion, the impact of prefecture-level city expansion on the number of degraded habitat patches first decreased and then increased and then decreased again with greater distance from the expansion area. The two inflection points of this trend peaked at 5.23 km and 40.51 km from the expansion area, respectively ( figure 2(a)).
Similarly, the effect of county town expansion on the total area of degraded habitat patches reflected the total impact of the expansion of all cities in the province. The impact of county town expansion on the total area of lost habitat mainly occurred within 60 km from the urban expansion area, and the total area of disappeared habitat decreased with greater distance from the urban expansion area as a power function. However, it is worth noting that the total area of lost habitat first increased with distance up to 7.67 km from the expansion area of county towns. The impact of the expansion of prefecturelevel cities on the total area of habitat degradation differed from that of county towns. The total area of lost habitat rapidly decreased, then increased, and then decreased again with greater distance from the expansion area. The two inflection points of this trend appeared at 9.78 km and 42.45 km from the expansion area, respectively ( figure 2(b)). Unlike the total area of lost habitat, the average size of lost habitat patches was affected similarly by the expansion of prefecture-level cities and county towns. The average size of lost habitat patches showed a linear decrease with greater distance from the expanded area ( figure 2(c)).
Interestingly, the impact of urban expansion on the total area and number of new habitat patches had similar spatial characteristics to that of the lost habitat patches. The impact of county town expansion on the distance of new habitat patches mirrored that of all cities in the province. The number of new habitat patches peaked at a distance of 8.31 km from areas of expansion, and the decline in the number of new habitat patches tended to be stable at distances of more than 60 km from such areas. The impact of the expansion of prefecture-level cities on the number of new habitat patches first increased and then decreased with greater distance from expansion areas, with the turning point of the trend occurring at 41.72 km from expansion areas ( figure 3(a)). This differs from the pattern observed around county towns.
The impact of county town expansion on the total area of new habitat patches was similar to that of the total impact of the expansion for all cities in the province: that is, the total area of new habitat first increased, then decreased, and finally stabilized with greater distance from expansion areas, the respective inflection points occurring at 8.58 km and 60 km. The distance over which the expansion of prefecture-level cities impact on the total area of new habitat patches was much higher than that of county towns. The total area of new habitat patches increased with distance over the first 33.39 km from areas of expansion around prefecture-level cities ( figure 3(b)).
All types of urban expansion had similar impacts on the average size of new habitat patches, which showed a bimodal curve with two peaks. The difference was that the distances at which the two peaks occurred were much greater for prefecture-level cities (29.16 km and 117.7 km) than for county towns (17.68 km and 75.69 km, respectively) ( figure 3(c)).

Impact of urban expansion on habitat quality
The impact of county town expansions on the average quality of habitat patches was again consistent with the total impact of all urban expansion; that is, habitat quality first decreased linearly with greater distance from the expansion area and then showed increased rapidly once the distance exceeded 89.69 km. The impact of the expansion of prefecture-level cities on t habitat patches broadly showed an exponential increase with greater distance from the expansion area. At distances of less than 120 km from urban expansion areas the rate of habitat quality improvement was slow, but it accelerated at distances of t more than 125 km ( figure 4).
Between 2000 and 2020, the quality of the new habitat patches also showed a trend of slowly increasing and then rapidly decreasing with greater distance from urban expansion areas. The impact of county town expansion on the quality of new habitat patches was once again consistent with the total impact of all urban expansion. The inflection point for this impact occurred at approximately 72.47 km from the urban expansion area. The expansion of prefecture-level cities had a weaker impact on the quality of new habitat patches than that of county towns, but the impact distance was greater with the inflection point occurring at approximately 115 km from the expansion areas of prefecture-level cities (figure 5).

Impacts of urban expansion on habitat
The impacts of urban expansion on habitat have received significant attention in recent years. However, until now it has remained unclear which types of urban expansion play the dominant role in driving these impacts. To the best of our knowledge, this is the first study to show that the spatial impact of county town expansion on habitat is stronger than that of prefecture-level city expansion, when assessed at a regional scale. Our study found that the spatial impact of county town expansion on habitat patch area, quantity, and quality was broadly consistent with the total impact of all city expansion, and was therefore key to dominating habitat change at the regional scale. This is different from previous studies, which have placed more emphasis on the impact of expansion around large cities as drivers of severe habitat damage [49], and our results also highlight the collective impact of all urban expansion on habitat at the regional scale, rather than just the impact of expansion in individual cities. Overemphasizing the impact of individual large city expansion on habitat while ignoring the collective impact of all urban expansion may be detrimental to the conservation of habitats at a regional scale.
Although the expansion of prefecture-level cities did not dominate overall habitat change at the regional scale, the spatial distance of their impacts was larger than that of county towns, and the impact curve was more complex. The spatial impact curves of prefecture-level city expansion on habitat generally showed multiple trends. This may be explained by the heterogeneity of landscapes that occur in transitioning from urban to rural areas [50]. With regards to county town expansion, we noted that habitat loss in mountainous areas was significantly greater than that in plain areas. This might be because there are more high-quality habitats in mountainous areas, meaning that any urban expansion in such areas is more likely to lead to habitat degradation. There are relatively few natural habitats in the plains areas of the region under study: therefore, urban expansion has less overall impact in this regard. However, the magnitude of impacts mentioned here refers primarily to the quantity and area of habitats, rather than considering the ecosystem services that they provide.

Distance thresholds affecting habitat
Threshold distance is a key parameter for coordinating regional habitat conservation and urban development [51]. We found that the impact of county town expansion on the quantity of habitat at the regional scale occurred within 60 km of the expansion area. This is consistent with other research findings [52] which show that more than two-thirds of the negative impacts of urban expansion on habitat quality were concentrated within a distance of approximately 50 km. According to the transfer to land use types, the degraded habitat patches within this threshold distance of the study area were mainly converted to construction land, indicating that there was a 'jump' expansion in county town development. This might be indicative of a lack of proper spatial planning and strict land management systems in county town development. In line with a previous study, the area and number of degraded habitat patches increased trend within approximately 7 km from the expanded area of county towns, and then began to decline rapidly once the distance exceeded 8 km. Within this specified distance, Feng et al [53] observed that earlier trend towards suburbanization saw industry move to suburban areas and build physical infrastructure which replaced natural habitats with impervious surfaces. However, such impacts drop off markedly at distances over 10 km.
The spatial impact threshold of urban expansion of prefecture-level cities on habitat quality was greater than that of county towns. This may partly relate to the potential for urban expansion. In the study area, most prefecture-level cities were distributed in plains areas, where urban expansion was less affected by topography. However, 63% of the county towns were located in mountainous and hilly areas where urban expansion is more limited; thus, the distance over which their urban expansion impacted on habitat was smaller than that of prefecture-level cities.

Strategies for habitat conservation
How urban expansion affects habitat, and what the distance thresholds involved are, can guide spatial planning which aims to coordinate regional habitat conservation and urban development. In this study, we found that the overall impact of county town expansion on regional-scale habitats was more significant than that of prefecture-level cities. This implies that the focus of the regional spatial planning strategy should not be simply focus on large cities but also include the development of county towns. The impact of county town expansion on habitat was mainly concentrated within a 60 km radius, meaning that the relationship between habitat protection and urban development needs to be addressed over such distances. This may be particularly beneficial to the protection of species with short-and medium-distance dispersal capacities. For example, enhancing connectivity between habitat patches within 60 km of county town expansion areas and designating protected areas to strengthen the overall resilience of the habitat network would mitigate the negative effects of urban expansion.
Meanwhile, more proactive policies are also needed to manage the future urban growth of large cities [54]. For example, by encouraging vertical expansion of high-rise development, brownfield redevelopment, and infill development that fully integrates green infrastructure into urban design [55], the negative impact of urban expansion on habitat may be mitigated to some extent. Some studies have also shown that non-green open space in cities, such as wasteland and brownfield sites, have significant green potential to enhance urban biodiversity [56]. However, the scarcity of urban land, particularly in city central and densely populated areas, means that greening the remaining non-green open space requires careful policies [57]. Whatever the form of urban expansion, restoring damaged habitats through strategies such as reforestation, wetland restoration and vegetation coverage enhancement projects remains important.

Limitations
This study reveals the impact of urban expansion on habitat and identifies the critical distance thresholds involved, insight which can guide the coordination of habitat conservation and urban development at a regional scale. However, due to the limitations of research capacity, our focus on habitat was limited to the quantity, area, and quality of habitats, without considering the spatial pattern of habitat patches. Spatial patterns determine the connectivity between habitat patches and the overall function of regional habitat networks [58], and this has important implications for the conservation of species that spread over long distances. The impact of urban expansion at different scales on habitat patterns needs to be further investigated. Different patterns of urban expansion, such as dispersed and strip types of urban fringe expansion, may also have different impact processes on habitats [59], which again merits further in-depth study.

Conclusion
We investigated the impacts of urban expansion at different levels on the quantity, area and quality of habitat and their respective distance thresholds. To the best of our knowledge, this is the first study to identify the differences in impact of prefecture-level city and county town expansion on habitat at the regional scale. These results could be used to establish key parameters for coordinating regional habitat conservation with urban development. We found that, among all urban expansion impacts, county town expansion was the dominant force in terms of impacts on habitat quantity, area, and quality degradation at the regional scale, and those impacts mainly occurred within 60 km of the urban expansion area. This finding supports the need for territorial spatial planning of county towns. The distance over which habitat is affected by the expansion of prefecture-level cities was significantly greater than that of county towns, and their impacts on the area and quantity of habitat mainly occurring within 130 km of expansion areas. However, due to the limited number of prefecturelevel cities in this region, the overall impact of their expansion on habitat at the regional scale was less than that of the county towns. These findings not only reverse our conventional impression that large city expansion dominates habitat degradation but also provides new perspectives and key parameters for developing holistic habitat conservation strategies at the regional scale.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).