The Construction and Optimization of Habitat Networks for Urban–Natural Symbiosis: A Case Study of the Main Urban Area of Nanjing

Abstract

Maintaining ecological balance relies on biodiversity, and habitat network construction plays an imperative role in preserving biodiversity in regional areas. Nevertheless, there is a problem with the current habitat network construction, which focuses exclusively on ecological benefits without taking other benefits into account as well. In this paper, six species of birds with varying habitat types and varying adaptabilities to city life are selected as target species to build a habitat network based on the InVEST model, Circuit Theory, and Linkage Mapper, focusing on nuclei, patches, corridors, and islands for the harmonious coexistence of human-green space-birds in the most densely populated area of Nanjing, and to refine landscape design techniques for habitat creation. Below is a summary of the main results. Firstly, there is a direct relationship between species distribution and migration capabilities and the urbanization adaptation capabilities of species. Meanwhile, habitat quality has a significant impact on bird species distribution. Furthermore, the habitat network in Nanjing’s main urban area has a distributed and partially degraded core area, a single connectivity structure with poor functionality, and significant fragmentation of habitat patches. Finally, as a result of the above results, two perspectives of ecological landscape planning and design are proposed to optimize the relevant green space landscape in Nanjing’s central urban areas based on biodiversity and satisfying the tripartite symbiosis of humans, green space, and birds in the city. By planning and implementing habitat networks, it is possible to enhance the habitat quality of urban green spaces to a certain extent and provide new ideas for the overall planning of urban–natural systems.

1. Introduction

Changes in land use and human activities lead to declining urban biodiversity, which in turn weakens urban ecosystem functions and reduces self-recovery, resulting in negative impacts on human well-being [1]. A growing body of research is focusing on how to resolve the conflict between the need for urban expansion and the ecological problems caused by urban development. In urban environments, traditional biodiversity conservation measures are still mainstream, but they are not optimal because it is costly and difficult to restore degraded habitats [2]. In addition, biodiversity can only be conserved by protecting single habitats and fragmented habitats [3], as it ignores the advantages of communicating between multiple habitats. In order to conserve biodiversity, habitat network construction has been shown to be an important method of interweaving fragmented habitats that have remained blocked and governed due to urbanization development into networks and accelerating ecological processes. While urban green space is an integral part of the urban ecosystem, more attention has been paid to its other ecological services in the past, including climate regulation, air purification, and recreation. In recent years, more and more researchers have focused their attention on the role of urban green spaces in preserving biodiversity. The emphasis has shifted from the conservation of individual green spaces to the creation of integrated green spaces. A habitat network is constructed in conjunction with this concept. Accordingly, it may be viewed from the perspective of landscape architecture as an alternative effort to deepen our understanding of the concept of symbiosis between cities and nature by combining urban green space and habitat networks organically, as well as exploring urban green space landscape design strategies that contribute to the development of habitat networks.

The construction of habitat networks should be conducted in a scientific and reasonable manner in order to identify specific and important urban green spaces as landscape habitats within the habitat network. In order to construct habitat networks, a number of research methods have been developed, and these are generally conducted in relation to the target species, focusing broadly on two aspects. Firstly, changes in the habitat network structures of target species use infiltration theory and various methods such as graph theory and morphological spatial pattern analysis (MSPA). Secondly, identifying the functional connectivity of habitat networks is based on experimental methods of species tracking, current theories, and least-cost pathways, or involves investigating factors that influence species dispersal. Currently, there are several methods available for the construction of habitat networks, including species tracking experiments, infiltration theories, graph theories, MSPAs, and current theories [4]. There is a wide variation in the applicability of the different methods and in the results obtained. It was jointly organized and developed by Stanford University, The Nature Conservancy, and The Century Nature Foundation as part of the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) project. The Habitat Quality module in the model has been widely used by scholars for evaluating habitat quality. For example, some scholars have utilized the InVEST model to evaluate the temporal and spatial variation in habitat quality in the Nile Basin of the Ethiopian Plateau and its relationship with landscape characteristics [5], the change in habitat quality on Jeju Island [6], and amphibian habitat in the prairies of the North American continent [7]. Presently, assessing habitat quality based on land use change and identifying the impact of land use change on habitat quality are research hot spots [8]. The majority of existing habitat network construction studies are based on the minimum cumulative resistance model (MCR) to extract least cost paths (LCPs) as ecological corridors and manually discriminate habitat patches. However, this does not take into account the exchange process of energy flow and information flow in ecological processes, and results in certain deficiencies in habitat network structural functions. The Linkage Mapper tool has been used by a few scholars to extract ecological corridors with width information [9,10], which essentially calculates the weighted cost distance to the nearest source. In terms of ecological corridor identification, the Linkage Mapper tool operates on a similar principle as the MCR. However, it does not need to eliminate duplicate redundant corridors artificially and is capable of obtaining ecological corridors with certain width information by setting corridor exceedance thresholds and other operations. It has been observed that most of the current research methods and ideas are based upon a particular method of identification for ecological corridor extraction and ecological node discrimination, whereas little research has been conducted on multi-method and multi-objective comprehensive identification.

A greater focus is being placed on habitats in urban green spaces in research on designing green spaces based on habitat network construction. Moreover, much of the research on the habitat network focuses on the macro level, such as an urban green space pattern. There is little research conducted on the specific urban green space landscape design. Landscape elements have been broken down into various factors that affect habitat or habitat connectivity, and researchers have evaluated the landscape variables that contribute to the quality and connectivity of urban green space landscape habitats. According to Matthies et al., stratified random sampling was conducted in 32 urban green spaces in Hannover, Germany, and they concluded that the size of green space patches and the heterogeneity of habitat was most important for the diversity of plant, bird, and overall species richness. Due to this, it is important for urban green space layout to preserve large and highly diverse habitat green spaces in order to maintain a high species richness [11]. In a study conducted by Wangmo Kang et al. in Gyeonggi Province, Korea, 44 small- and medium-sized urban woodlands were examined, and it has been concluded that vegetation complexity and habitat connectivity are positively correlated with species abundance and diversity, but they have not received much attention in landscape design for green spaces [12]. A new method was applied by Wang J et al. to construct a network of habitats and an urban block unit structure for ecological restoration at the city-block scale, and four design approaches were extracted from the study. In addition, this study explained that the “crosswise radiation” network had a significant impact on butterfly biodiversity, while the “dendritic zoning” network had a significant effect on enhancing ecological restoration [1]. Meanwhile, current landscape design research on habitat creation is focused on: landscape design research based on maintaining habitat suitability and an abundance of target species, as well as plant habitat creation. Culbert et al. found that parks with multiple vertical vegetation landscapes provide more potential habitat and function better as buffers against microclimate changes, thus providing a greater diversity and richness of species [13]. Bonebrake et al. sampled butterfly species and abundance in four landscape spaces in Hong Kong’s urban parks, and based on the findings of this study, the high level of butterfly diversity, the presence of rare species, and the use of vegetation (particularly nectar resources) suggest that Hong Kong’s urban parks possess some conservation value [14]. Monica Kaushik et al. sampled vegetation and bird communities in 52 densely sampled sites in 18 urban green spaces in Dehradun, India, during breeding and non-breeding seasons. They demonstrated that urban green space size and tree species richness per unit area became important factors in predicting bird species richness and density, while built-up areas and barrenness of green space influenced community composition more than the size or density of green space. They suggested that community composition could be determined by the increased compositional and structural heterogeneity of vegetation and the conservation of large and old native trees to reduce habitat fragmentation [15]. Through a series of measures, including plant habitat design, Yukihiro Morimoto restored the natural forest containing the zoned top community within Osaka Manbo Memorial Park, including the design of urban plant communities, the design of plantings, and the modification of soil environments in an urban area, in conjunction with the forest window theory and the introduction of topsoil from the surrounding forest [16].

It is worth mentioning that research on urban green space landscape design based on habitat networks has generally concluded that habitat diversity, plant diversity, and insect diversity are all positively correlated with bird diversity in green spaces. In the food chain, birds take a prominent position. In a certain sense, the species and numbers of birds can determine the status of organisms in other ranks, while their abundance can be regarded as a measure of ecosystem health and species diversity [17,18]. Blair’s simplification of species urbanization adaptation gradients indicates that birds are the most studied animal taxon in terms of adaptation gradients [19] and are distinguished by their mobility, dispersal capacity, and ability to simulate networks at various scales. There are many ecological processes that are dependent upon birds, such as seed dispersal [20], pollination [21], and the regulation of herbivorous arthropods [22], as well as how birds respond to different levels of urbanization [23]. It is therefore common to use birds as indicators for monitoring biodiversity [24] and the restoration of habitats [25]. A report published by the Intergovernmental Science—Policy Platform on Biodiversity and Ecosystem Services estimates that 3.5% of birds had become extinct by 2016 [26]. It was suggested, however, that the risks of extinction for birds and other vertebrates would have been at least 20% greater if conservation action had not been taken in recent decades. Therefore, it is of the utmost importance to study bird biodiversity in relation to urbanization and to protect it. According to some studies, the fragmentation and isolation of urban habitats have adversely affected urban birds [27,28,29,30], leading to the homogenization of bird populations in cities. Nevertheless, the results of other studies have indicated that moderate urbanization has the potential to increase bird abundance and species diversity, due to the increased availability of resources and the reduction in predators [31,32,33].

Despite the fact that cities are dominated by humans, they provide habitats for many other species [34]. In order to assess the impacts of habitat fragmentation and degradation, it is necessary to understand species–habitat relationships. Until now, habitat modeling in fragmented landscapes has relied on metrics related to landscape composition and configuration. The importance of habitat quality in determining species distributions has not been sufficiently examined [35]. Moreover, mainstream methods used for the construction of habitat networks have several shortcomings. Furthermore, there is a lack of research on the optimization of urban habitat networks from the perspective of urban green space landscape habitat design and habitat quality improvement. It is hypothesized in this article that the ability of species to migrate and adapt to urbanization will affect both the range of species’ distribution and the breadth of habitat types. In addition, the construction and optimization of habitat networks can contribute to the restoration of urban biodiversity and improve the effectiveness of urban ecological systems.

In order to test the above hypotheses, this paper proposes a method for simulating habitat networks using the InVEST model and Linkage Mapper. There are six species of birds in Nanjing that have been selected as the target species in order to investigate the habitat network and green space landscape habitat quality in the main urban area of Nanjing. With the construction and optimization of urban habitat networks, it is hoped that harmony between people, green spaces, and birds can be promoted, and that symbiosis between humans and nature can be achieved.

2. Materials and Methods

2.1. Experimental Area

There are 9.31 million residents in Nanjing, the capital of Jiangsu Province, and the city is 86.8% urbanized. This is a mega-city within the Yangtze River Delta city cluster, shaped like a long north–south strip with a total area of 6587.02 km². A primary urban area located in Nanjing is located between the latitudes 31°80′ and 32°25′ N and the longitudes 118°56′ and 119°30′ E. The major urban area comprises an area of 787.45 km² and is home to 4,349,900 residents. This is the most urbanized area of the geographical landscape within Nanjing, representing 12% of the city’s area and 46.7% of its resident population as of the end of 2020 [36]. Accordingly, the main urban area located in the central part of Nanjing was chosen as the subject of this study. (Figure 1). According to the Flood Control Emergency Plan for the Main Urban Area of Nanjing issued by the General Office of Nanjing Municipal Government in 2018, the main urban area is west and north of the Yangtze River, and south of the Qinhuai New River, which contains the Jiuxiang River in Xianlin. To facilitate statistical analysis, this paper focuses on the six districts of Xuanwu, Qinhuai, Jianye, Gulou, Qixia, and Yuhuatai as the main urban areas of Nanjing.

2.2. Experimental Object

For the construction of urban habitat networks, indicator species should be observable by the naked eye, have certain regional mobility, and have high habitat quality. These species include birds, fish, amphibians, reptiles, and small animals. From the Ecological Quality Indicator Species List of Jiangsu Province (

Table 1. Bird species selected for the simulation of the habitat network of Nanjing’s main urban area.

Living Habits	Code	Species	Common Name	Food Guilds	Resident
Forest bird	1	Accipiter soloensis	Red-bellied hawk	Carnivorous	Migrant bird
	2	Falco tinnunculus	Kestrel		Migrant bird
	3	Otus sunia	Red-horned owl		Resident
	4	Garrulax canorus	Wood thrush	Omnivorous	Resident
Waterfowl	5	Aix galericulata	Mandarin duck		Migrant bird
	6	Hydrophasianus chirurgus	Water pheasant		Resident, Summer types

), six species of birds were chosen as target species based on the faunal resources of Nanjing; local observation reports from the China Birding Records Center; and the habitat types, urban adaptation, and migration capabilities of birds. Four forest birds and two water birds were included in the target species. Among these six species, there are three types of birds: urban avoiders, urban adapters, and urban symbionts [37].

2.3. Data Sources and Experimental Processing

In this paper, core habitat vector patches were used in conjunction with resistance rasters to determine the least-cost linkages between core areas to simulate habitat networks. Therefore, the higher the resolution of the raster data, the more accurate the network will be. As part of this analysis, the following data were used: the DEM digital elevation model with 30 m spatial resolution and SRTMASPECT slope orientation data with 90 m resolution (Geospatial Data Cloud); China’s maximum NDVI data for 2020 with a spatial resolution of 30 m (National Ecological Science Data Center); 2020 LULC land use classification data in Nanjing with a spatial resolution of 30 m (Chinese Institute of Geoscience and Resources for Science https://www.resdc.cn/, accessed on 28 July 2022); and the unified geographic and projection coordinate system (WGS_1984_UTM_Zone_50N) for all data based on land use data. According to this paper, the secondary classification of the LULC data corresponds to the habitat type of the target species. Data on bird observations were collected from Nanjing and neighboring cities (China Birding Records Center http://www.birdreport.cn/, accessed on 17 August 2022), and as a result of the analysis, a kernel density map of the spatial distribution of birds was produced using ArcGIS tools.

According to habitat descriptions and observation reports of reference target species [38], the InVEST User Guide [39], and other relevant literature [40,41,42,43,44,45], parameters for variables such as the distance of impact, the attenuation mode, the weight of threat sources, and the suitability of different land use types to serve as habitats for target species and sensitivity of habitats to threat sources were determined (

Table 2. The impact distance, attenuation mode, and weight of the threat source.

Threat	Max_dist/km	Weight	Decay
Dry land	1	0.7	Linear
Water field	1	0.7	Linear
Other land	12	1	Exponential
Village	5	0.5	Exponential
Town	10	0.9	Exponential

and

Table 3. Habitat connectivity and sensitivity of various land use to threat sources for target species.

Land Use Type and the Corresponding Land Utilization Data Code	Habitat Suitability						Threat
	Accipiter soloensis	Falco tinnunculus	Otus sunia	Garrulax canorus	Aix galericulata	Hydrophasianus chirurgus	Dry Land	Town	Village	Other Land
Water field 11	0.4	0.3	0.4	0.3	0.3	0.3	0.8	1	0.8	1
Dry land 12	0.5	0.5	0.5	0.4	0.2	0.2	0	0.5	0.4	0.5
Wooded land 21	1	1	1	1	0.6	0.8	0.5	0.7	0.7	0.9
Shrubland 22	1	0.8	0.8	0.8	0.4	0.4	0.4	0.6	0.6	0.8
Open woodland 23	1	0.6	0.6	0.6	0.3	0.3	0.4	0.6	0.5	0.8
Other woodland 24	0.6	0.6	0.6	0.6	0.3	0.3	0.3	0.5	0.4	0.8
High-coverage grassland 31	0.6	0.6	0.5	0	0.3	0.3	0.3	0.5	0.3	0.6
Low-coverage grassland 33	0.5	0.5	0.3	0	0.2	0.2	0.3	0.5	0.3	0.6
River and canal 41	0.8	0.7	0.3	0.7	1	1	0.7	0.9	0.7	1
Lake 42	0.5	0	0	0.7	1	1	0.7	0.9	0.7	1
Reservoir pond 43	0.5	0	0	0	1	1	0.7	0.9	0.7	1
Beachland 46	0.3	0	0	0.3	0.8	1	0.7	0.9	0.7	1
Urban land 51	0.6	0.7	0.5	0.2	0.3	0.5	0	0	0	0
Village 52	0.7	0.5	0.3	0.3	0.2	0.7	0	0	0	0
Other land 53	0	0	0	0	0	0	0	0	0	0
Bare land 65	0	0	0	0	0	0	0	0	0	0
Bare rocky ground 66	0	0	0	0	0	0	0	0	0	0

).

2.4. Experimental Method

The biodiversity-oriented urban green space habitat network simulation method was proposed, which used green space in Nanjing’s main urban area as the carrier. As target species, species with a variety of habitat types and adaptability to urbanization were selected. The InVEST model and species observation data were combined to screen and identify potential habitat source sites and to construct resistance surfaces. Using the ArcGIS Linkage Mapper tool, migration corridors and ecological flow paths of target species were simulated. In addition, the simulation results were combined with processing to provide a comprehensive habitat network of the main urban area. The results of the analysis of the simulated habitat network were used to develop recommendations for optimizing green spaces both in terms of overall planning and landscape design in the main urban area.

2.4.1. Ecological Source Site Identification Based on the InVEST Model

Species habitats were assessed using the Habitat Quality of the InVEST model. The purpose of this module is to assess the quality of species habitats using land use and cover (LULC) raster data in accordance with the following considerations: a. the threat source (distance, weight, and attenuation mode), b. the suitability of different land use types as habitat, and c. the degree to which the habitat is sensitive to the threat. We obtained the results as a raster plot of the spatial distribution of habitat mass fraction (values ranging from 0 to 1) [46]. First, the attenuation of the habitat impact of the threat source caused by the change in distance was calculated using Equation (1). [39]:

$$\left\{\begin{array}{c}{i}_{rxy}=1-\frac{d_{xy}}{d_{r_{max}}} if linear\\ i_{rxy}=e^{-\frac{2.99}{d_{r_{max}}} · d_{xy}} if exponential\end{array}\right.$$

(1)

where $d_{xy}$ denotes the linear distance between raster $x$ and raster $y$. $d_{r_{max}}$ represents the maximum distance of action of the threat source. A threat source’s impact on the habitat decays (linearly or exponentially) with increasing distance between them. According to Equation (2) [39], the degree of degradation $D_{xj}$ is calculated based on the effect of the threat source $r$ on the land use type $j$ raster $x$:

$$D_{xj}={\displaystyle \sum }_{r=1}^R{\displaystyle \sum }_{y=1}^{Y_r}\frac{w_r}{{{\displaystyle \sum }}_{r=1}^R{W}_r}{r}_y{i}_{xy}{\beta }_x{S}_{jr}$$

(2)

where $y$ represents the entire raster set of the threat source on the raster map, $Y_r$ represents the set of rasters of the threat source $r$ on the raster map, $W_r$ represents the weight of the threat source, $r_y$ represents the threat source’s influence value, ${\beta }_x$ represents the accessibility level of the raster $x$, and $S_{jr}$ represents the sensitivity level of the land use type to the threat source $r$. In general, $S_{jr}$ ranges from 0 to 1, and if $S_{jr}=0$, the land use type is not sensitive to the threat source. As a final step, the habitat quality score $Q_{xj}$ was calculated according to Equation (3) [39]:

$$Q_{xj}=H_j(1-\frac{D_{xj}^z}{D_{xj}^z+k^z})$$

(3)

where $Q_{xj}$ represents the habitat quality for raster $x$ in land cover type $j$, $H_j$ represents the suitability of land use type $j$ as a habitat for the target species, $k$ represents the half-saturation constant, and $z$ represents the specified value of 2.5.

Ecological source sites are the source of ecological land conservation, the habitat of existing species, and the source point of species exchange and dispersal [47]. In this paper, the following conditions and steps were used to identify habitat source sites: firstly, rasters with an InVEST habitat quality assessment score greater than 0.7 were extracted and converted into facets (vectors); secondly, the facets which correspond with the habitat characteristics of the species and have an area greater than 60 hm² were determined; thirdly, the observed coordinate data of the species were imported, and the facets whose points were within the interior or whose edges were shifted outward by 1100 m were identified as habitat sources [35].

2.4.2. Ecological Resistance Surface Parameter Setting

Environmental resistance is the expenditure or consumption necessary to overcome resistance in the process of ecological flow from a source to a landscape substrate of a different resistance type. A higher patch resistance value indicates greater difficulty in traversing patches during ecological processes. To quantify this metric, multiple resistance factors should be selected to weigh the overlay and analyze the data in a comprehensive manner. A combination of land use LULC and elevation data (DEM), slope data, and normalized vegetation index data (NDVI), based on the actual situation and vegetation cover of the study site, was used to construct ecological resistance surfaces (

Table 4. Integrated resistance surface parameter setting in the study area.

Resistance Factor	Classification Criteria	Resistance Coefficient	Weight
NDVI	$>0.90$	10	$12/25$
	$0.82~0.90$	20
	$0.76~0.82$	30
	$0.67~0.76$	40
	$0.14~0.67$	50
LULC	Cropland	50	$6/25$
	Forest	1
	Grassland	40
	Wetland	30
	Urban and built-up	100
	Bare ground	70
Slope	$0~1.41$	10	$4/25$
	$1.14~4.37$	20
	$4.37~9.16$	30
	$9.16~16.06$	40
	$16.06~35.93$	50
DEM	$0\le \mathrm{h}<15 \mathrm{m}$	10	$3/25$
	$15\le \mathrm{h}<50 \mathrm{m}$	20
	$50\le \mathrm{h}<100 \mathrm{m}$	30
	$100\le \mathrm{h}<150 \mathrm{m}$	40
	$150\le \mathrm{h}\le 255 \mathrm{m}$	50

). Using the Reclassify tool, the land use type, elevation, slope, and NDVI were used as resistance factors. The Reclassify tool was used to regrade these data and assign ecological resistance values to construct ecological resistance surfaces.

Each influencing factor was assigned a weight, and the weights were calculated using the sorted reciprocal method [48], which implies that all influencing factors are arranged in ascending or descending order, and the factor with the greatest influence receives the lowest value of 1, and so on. Based on Equation (4), the weights of each influencing factor were calculated:

$$L_i=\frac{1/k_i}{\sum \left(1/k_i\right)}$$

(4)

where $L_i$ represents the normalized weight value of the influence factor and $k_i$ represents the sequence number of the th influence factor in the influence factor sequence. In the above equation, the numerator represents the weight of each influence factor, while the denominator represents the sum of the weights of multiple influence factors. In terms of habitat quality, vegetation cover is the first priority, followed by land use; there is a certain amount of mountainous terrain in the main urban area, and slope has a considerable impact. Therefore, according to this formula, the weight values of each influence factor are $12/25$ for NDVI, $6/25$ for land use, $4/25$ for slope, and $3/25$ for elevation. Eventually, using the Raster calculator in the ArcGIS spatial analysis module, the individual layer data were analyzed to produce a comprehensive ecological resistance surface.

2.4.3. Ecological Corridors Extracted with Linkage Mapper

In Linkage Mapper, the MCR, circuit theory, and graph theory were integrated to construct ecological corridors (line and surface), important ecological nodes (stepping stones and obstacle nodes) were identified using circuit theory, and ecological source sites and corridor centrality were calculated using graph theory. Using the Linkage Mapper tool in ArcGIS 10.8, the least-cost path (LCP) was constructed as the optimal corridor for species movement between source sites based on circuit theory in the main city of Nanjing by utilizing ecological source map patches and resistance surface rasters.

2.4.4. Habitat Network Construction

An important pattern for biodiversity conservation is habitat networks, which are web-like landscape systems that consist of core areas, habitat patches, semi-natural habitat patches, and connectivity structures such as corridors and stepping stones. In this study, core areas are habitats that meet the criteria of having a score greater than 0.7, contain three or more target species, and cover at least 60 hectares. High-quality habitats that have no species observations at the present time, but are greater than 60 square meters or bear the presence of target species and an area greater than 30 square meters, are used as habitat patches [49]. As an alternative to natural habitat patches, paddy fields are used. Patches of fragmented habitat distributed throughout core areas and patches of habitat that facilitate species migration and high-quality habitat patches serve as stepping stones. In this study, the least-cost pathways between connected habitat sources, extracted using Linkage Mapper, were used as primary corridors and ecological flow paths as potential migration routes for the target species. On the basis of the extracted core areas, habitat patches, semi-natural habitat patches, stepping stones, and connectivity structures, for the main urban area, we simulated a habitat network and angled it with satellite images obtained in recent years, downgraded fragmented core areas into habitat patches, removed stepping stones and habitat patches that were nonexistent or severely fragmented, and finally obtained a habitat network for Nanjing’s main urban area with a core–patch–corridor–island structure after correction.

3. Results

3.1. Assessment Results of Habitat Quality

The spatial distribution of birds and habitat quality evaluation results were superimposed to form a spatial distribution pattern of biodiversity that combined biological aggregation and survival suitability, and complemented by evaluation grading and spatial distribution. Based on the habitat quality scores for target species and the overlay analysis of kernel density of observation site coordinates, it was determined that several distribution points and a wide distribution range were found for the six bird species in the main urban area, in accordance with the results of the habitat quality assessment conducted in the main urban area (Figure 2).

Waterbirds such as Mandarin ducks and water pheasants can be found in a variety of high-quality water habitats in the main urban area and in other suitable habitats, primarily in the Yangtze River coastal wetlands, Jijiang, the Qinhuai River, and other small waters. Mandarin ducks are migratory birds and have been observed to form small groups around small bodies of water, such as the large and small lake scenic areas on Zijinshan Mountain and Biwa Lake, Qianhu Lake, Yanqiu Lake, Yueya Lake, and the moat at the foot of the mountain. Water pheasants, which require a higher level of habitat quality, have a smaller distribution range than Mandarin ducks in the main urban area and can only be found in riverside wetlands, the Qinhuai River, and Mochou Lake. Due to their size, flight distance, and altitude advantage, raptors such as red-bellied hawks and kestrels are able to distribute in numerous high-quality mountain forest habitats as well as habitats of low-to-medium quality. Despite its small size, the actual distribution ranges of painted buntings as both resident and forest birds within the main urban area are much greater than the findings of the wood thrush assessment, owing to the distribution of paddy fields and villages as stepping stones around the main urban area, which reduces the wood thrush’s migration resistance. In contrast, the distribution range of the small raptor red-horned owls is much smaller than that of the red-bellied hawks and kestrels (fewer observation record sites), as red-horned owls are sensitive to human activities and habitat changes in cities, as well as their low adaptability to urban expansion. In addition, the mountain woodlands in the eastern part of the main city are severely fragmented by construction sites, and the reduced habitat quality has resulted in a decline in the size of its population [38].

Further, based upon the proportion of high-quality habitats in the actual distribution of the target species (

Table 5. Habitat types and total area of species distribution, the area of high quality habitat types, and the area share of both in the main area.

Species	Habitats for Which Observations Are Recorded			Habitats of High Quality (Two of the Highest Levels)
	The Types of Habitat and Their Corresponding Land Use Data Codes	Total Area /hm2	Percentage in the Main Urban Area of Nanjing/%	The Types of Habitat and Their Corresponding Land Use Data Codes	Area/hm2	Percentage of Habitats with Observed Records/%	Percentage in the Main Urban Area of Nanjing/%
Accipiter soloensis	12, 21, 23, 24, 31, 41, 42, 51, 52, 53	62,392.14	79.23	21, 41, 23	13,888.35	22.26	17.64
Falco tinnunculus	11, 12, 21, 23, 41, 42, 51	70,006.14	88.90	21, 22	7370.17	10.53	9.36
Otus sunia	21, 24, 41, 42, 43, 51	53,629.47	68.11	21, 23, 22	7800.57	14.55	9.91
Garrulax canorus	12, 21, 41, 42, 51, 52, 53	61,343.01	77.90	21, 22, 23, 41, 42	14,273.82	23.27	18.13
Aix galericulata	12, 21, 41, 42, 43, 51	58,885.65	74.78	41, 42, 43, 46	7475.31	12.69	9.49
Hydrophasianus chirurgus	41, 43, 51	46,194.75	58.66	21, 41, 42	14,832.63	32.11	18.84

), red-bellied hawks and kestrels have lower quality requirements for suitable habitats, and are more able to adapt to urban environments. Thus, symbionts such as kestrels, whose dense distribution corresponds to the degree of urbanization, are widely distributed in built-up areas and low-quality habitats.

3.2. Simulation Results of the Habitat Network

Using the method described in the previous section for calculating the resistance surface, according to the results of the research, the main city of Nanjing has an integrated resistance surface distribution, as shown in Figure 3. As a result, it is possible to obtain further simulation results for the habitat network based on this information. Using the comprehensive simulation of the habitat network in Nanjing’s main urban area (Figure 4a), eight core areas are identified, including Zhongshan Mountain National Park and Xuanwu Lake in the central core area; the Muyan Riverside Landscape Area and Qixia Mountain Scenic Area on the Yangtze River; Guishan Mountain, Longwang Mountain and Shewu Mountain in the east; and the General Mountain and Niushou Mountain in the southwest. The area consists of five habitat patches, including Jubao Mountain Park on the north side of Zhong Shan, Wulong Mountain, the campus green areas of Nanjing Normal University and Nanjing University, and Ling Mountain. In the connectivity structure, there are 23 major habitat corridors that make up the habitat network, each of which has a truncated ecological flow path width of 2 km (Figure 4b). The major corridors from west to east are located along rivers such as the Jiejiang River, the Yueya Lake, and the Yangtze River, as well as along roads such as North Central Road, Yuhua Avenue, East Inner Ring Road, South Inner Ring Road, Daming Road, and Xuanwu Avenue. Core areas, habitat patches, and habitat corridors make up the primary structure of the network, along with semi-natural habitat patches located north, east, and southwest of the main urban area. As a result of fragmented high-quality habitats around the core area and habitat patches that serve as pathways for species migration, the main urban area is composed of an overall habitat network. Compared with the southern and northern areas of Nanjing, the central part of the main urban area has more core areas, but except for the Zhongshan Mountain National Park, all of the other core areas are relatively small in size. The connectivity structure between the core areas separated by the city is relatively single and concentrated, and the main habitat corridors are mainly located along the rivers or on routes that are the shortest distance between the main areas and the patches. Generally, the habitat network in the main urban area has an overly fragmented and partially degraded core, a single and poorly functioning connectivity structure, and a severe fragmentation of habitat patches.

4. Discussion

4.1. Habitat Quality and Bird Distribution

Local habitat affects the fitness of animals through variation in resources and environmental conditions. Spatial–temporal variations in habitat conditions thus generate strong selective pressure for habitat selection, which in turn influences the reproduction and survival of individual birds, and contributes to the regulation of bird populations [50]. It is not surprising that researchers have long recognized the need to understand variation in habitats for birds. Some studies that modeled biodiversity using only habitat quality measurements have suggested that habitat quality influences species distribution in fragmented landscapes [51,52,53,54,55]. Other studies have compared the explanation power between the metrics of habitat composition and habitat quality, finding that habitat quality can override habitat composition influences [56,57]. Even so, habitat loss and habitat degradation are interdependent ecological processes that operate simultaneously [58] and few studies have examined how habitat quality and quantity interact to influence species distribution patterns in fragmented landscapes. Many hypotheses indicate that intermediate disturbance and landscape heterogeneity may enhance bird diversity [59]. However, bird diversity responses to urban environmental changes are complex and varied with different landscape features [60]. Habitat natural factors such as greenness, habitat quality, and proximity to waters have a positive impact on richness by improving vegetation diversity and coverage to enhance the quality of habitats [61,62]. Previous studies have shown that the main factors affecting urban birds include the natural habitat environment, area, configuration, and the surrounding urban matrix [63,64]. Studies have focused on urban green spaces, although few have reported on other suitable habitats for birds, such as urban waters [65]. An understanding of the effects of urbanization on birds are limited by only considering the composition and structure of habitats at patch level, such that research on integrating multiple perspectives of social ecosystems is lacking. To date, habitat modeling in fragmented landscapes has relied on landscape composition and configuration metrics based on the patch-corridor-matrix and heterogeneous mosaic theoretical frameworks [66]. However, the response of birds to their environment is mainly nonlinear, which makes it not fit well. In addition, the importance of habitat quality and its relationship to the distribution of bird diversity has not been sufficiently explored.

As a result of our analysis, we discovered that birds with high migration and urbanization adaptations can be found in a wide variety of habitats of varying quality. Birds with high migration and low urbanization adaptations are distributed only in high-quality suitable habitats with a large range, while birds with both low migration and urbanization adaptations are distributed only in a specific range of high-quality suitable habitats. The results of this study are in agreement with our previous hypothesis that the actual distribution of species and habitat quality assessment results correlate with the migratory and urban adaptation abilities of target species. This issue must be discussed in the broadest possible context. There may also be a discussion of future research directions. Furthermore, the results indicated that bird species are clustered, that the frequency of species occurrence in the surrounding landscape is influenced by ecological characteristics (i.e., breeding habitat and nest location) regardless of habitat area, and that bird habitat networks are structured in different ways according to the habitat type. Consequently, this study rectifies the lack of inquiry previously mentioned into the relationship between species and habitat and clarifies the importance of habitat quality in determining species distribution. There are two types of habitats that have the greatest impact on bird distribution: woodlands and wetlands. Wetlands were used primarily by migrants, while forests were primarily inhabited by residents. As shown in the results, both wetlands and forests make a significant contribution to the structure and biodiversity of the bird–habitat network, and should therefore be given more attention during conservation efforts.

This study bridges the gap between bird distribution and habitat quality in the context of rapid urbanization. On the one hand, it combines the InVEST model and ArcGIS software to model and analyze the bird habitat identification method. On the other hand, it adopts a habitat network perspective and integrates theories of landscape ecology and landscape architecture to optimize bird habitat strategies. The results of this study are essential for the sustainable implementation of ecological space restoration and management to maintain a high level of avian diversity.

4.2. Habitat Network Optimization Recommendations

In order to balance the needs of growing urban development with long-term ecological interests, the construction and optimization of urban green space habitat networks are aimed at improving connectivity between urban green spaces, enhancing people’s recreational experience while simultaneously promoting the exchange of species, and optimizing the landscape patterns of urban green spaces. Arevalo et al. have highlighted the importance of creating large networks of green spaces in rapidly urbanizing areas that are surrounded by smaller urban forms and regulating noise levels in order to ensure the protection of native bird populations in cities, particularly those in danger [67].

According to the results of the comprehensive network simulation, the habitat network in Nanjing’s main urban area has a distributed and partially degraded core area, a single connectivity structure with poor functionality, and significantly fragmented habitat patches. As a result of these current problems with the habitat network, the following recommendations are made in order to improve its structure and function systematically.

4.2.1. Protecting Core Areas to Drive the Operation of Urban Habitat Networks

The protection of core areas is essential to the operation of urban habitat networks. As habitat networks operate, core areas provide an impetus for species migration and can drive species populations high enough to drive them to migrate to look for more habitats. Expansions of urban construction land or other land uses that fragment the core area may have a significant impact on its long-term conservation [68]. Thus, when planning urban green spaces, it is best to maximize the habitat area and continuity of the core area and minimize the edges in order to reduce fragmentation of the core area. For example, Zhongshan-Xuanwu Lake’s important core area is connected to the Muyan Riverside Landscape Area on the north side of the city, the Qixia Mountain Scenic Area, Guishan Mountain, Shewu Mountain on the east side, and General Mountain and Niushou Mountain on the southwest side. There are six species of birds in the mountains and woodlands above, and they are widely distributed and have good dispersal conditions. To buffer the disturbance of species within the core area from the construction area, other types of green spaces should be used as transition areas between the core area and the construction area. For instance, the green space on Nanjing University’s campus between Qixia Mountain and Guishan Mountain can be used as a transition zone.

4.2.2. Valuing the Role of Cultivated Land as Semi-Natural Habitat Patches

Cultivated land is valued as a semi-natural habitat patch. In Nanjing, cultivated land is the predominant habitat type. It is widely believed that cultivated habitats may become additional habitats for some species as a result of activity disturbance and agricultural pollution [69]. As discussed above, the painted brow is one example. The cultivation of paddy fields can also help reduce the resistance of its species to migration and dispersal, reduce the impact of other habitat destruction on the species, and help it adapt to the urban environment in order to better survive and reproduce.

4.2.3. Protect and Plan Connectivity Structures at Various Scales

At various scales, connectivity structures should be protected and planned. In habitat networks, corridors provide important connectivity structures, but their importance is often overlooked. As a result, they are unable to fulfill their intended functions and roles. There are only 16 habitat corridors along the river and along existing roads in this paper that are effective, while the remainder are interrupted or blocked to some extent by roads, overpasses, tall buildings, or construction sites. Greenways of 12–30 m wide can be used at the regional level, particularly in the core urban areas, as a means of connecting the core areas, meeting bird migration, and protecting invertebrates [70]. In groups such as the Laoshan-Lushuiwan wetland group and the Shishu-Wuqi Mountain group, the core areas and habitat patches in these groups are closely interconnected and should be protected as a whole. In order to reduce the blockage between core areas and habitat patches, construction development within the groups should be reduced and building heights should be limited.

5. Conclusions

This paper adopts a habitat network simulation method based on the InVEST model, circuit theory, and Linkage Mapper in order to fully understand the effectiveness of habitat networks in enhancing urban biodiversity and improving urban ecological functions, as well as achieving the goal of symbiosis between cities and nature. The study based on six bird species as indicator species in Nanjing’s main urban area confirmed that there is a direct relationship between species distribution and migration capabilities and urbanization adaptation capabilities of species, and also that the habitat quality has a significant impact on bird species distribution. By planning and implementing habitat networks, it is possible to enhance the habitat quality of urban green spaces to a certain extent and provide new ideas for the overall planning of urban–natural systems.

It is imperative that decision makers and urban planners join forces and think about cities in a smarter way in order to create a healthier urban environment for humans as well as other species. A further objective is to increase the permeability and connectivity of urban green spaces, as well as the accessibility and connectivity of species and people, by constructing and conserving habitat networks in cities. As a result, the urban species observation database is constantly updated and revised, the methodology is continuously improved, and a virtuous cycle is created.