The rural areas of Pukou District, Nanjing, are situated in the northwestern and western regions of the district, encompassing 178 natural village sites, covering a total area of 589.27 km² (Figure 1). Within the study area, substantial areas of farmland, forest, and other habitat types remain, which are essential for pollinator habitat conservation. Moreover, the study area includes key ecological nodes, such as Laoshan, Guanyinshan, and Lvshuiwan, which are vital for conserving pollinator diversity in Pukou District’s rural areas. The Laoshan Nature Reserve hosts 2,081 species, representing 47.66% of Jiangsu Province’s total species, including 718 insect species. Among them, Lvshuiwan has complex terrain and topography, with a variety of wetland types, providing favorable conditions for the growth of numerous plants and animals.

Pukou District, a core component of Nanjing Jiangbei New Area, faces a conflict between rapid urbanization and ecological conservation. According to the Nanjing Pukou District Territorial Spatial Plan (2021–2035) and related policies, major industrial projects in Jiangbei New Area have driven the expansion of construction land, with 61.9% occupying arable land, encroaching on pollinator habitats. Concurrently, modern urban development has intensified regional pollution and disturbances, exacerbating pollinator habitat fragmentation, thereby heightening the vulnerability of ecosystems.

Thus, selecting Pukou District as the study area not only captures the characteristic features of rural pollinator habitats but also serves as a critical case study for biodiversity conservation under intense urbanization.

Pollinator data were sourced from field surveys conducted in spring 2025 in Pukou District’s rural areas using the cross-sectional counting method (Natuhara et al., 1998), supplemented by the Global Biodiversity Information Facility (GBIF, https://www.gbif.org/, accessed February 2025), the Integrated Insect Specimen Database (https://insectaintegration.com/, accessed February 2025), and literature reviews, resulting in 300 records with precise geographic coordinates.

Habitat information was compiled from the Chinese Animal Theme Database (http://www.zoology. csdb.cn, accessed February 2025) and Fauna Sinica Insecta, summarizing the habitat characteristics of pollinators in Pukou District’s rural areas. Ultimately, habitat features for 37 families and 313 pollinator species were compiled, with key attributes extracted, including host plants, suitable temperature ranges, and altitude (Table 1).

Most crops serve as targets for pollinators. Syrphidae larvae prey on aphids, thrips, and other agricultural pests. Most LEP insects congregate in rice paddies and wetlands near small streams to access water. LEP host plants exhibit varied environmental requirements, necessitating consideration of factors such as forest canopy closure and terrain. Given the varied flight abilities of pollinators, habitat proximity is a key factor influencing their distribution. Concurrently, human activities significantly impact pollinators. Although all three insect orders are influenced by temperature, climate change was deemed insignificant due to the study area’s limited spatial extent, and thus excluded as an environmental factor. In summary, 13 environmental factors were selected (Table 2).

Terrain data were sourced from the Geospatial Data Cloud, Computer Network Information Center, Chinese Academy of Sciences (http://www.gscloud.cn). The Normalized Difference Vegetation Index (NDVI) was obtained from the National Ecosystem Science Data Center, National Science & Technology Infrastructure of China (http://www.nesdc.org.cn). Land cover data (LUCC) were sourced from the Resource and Environment Science Data Platform (https://www.resdc.cn/Default.aspx). Road network and river data were sourced from OpenStreetMap (https://www.openstreetmap.org/). All data were collected for 2020, with distance factors calculated using Euclidean distances to land use types. Data processing and analysis were conducted using ArcGIS 10.8, with all raster files standardized to a 30 m × 30 m pixel resolution.

Using ENMTools, Pearson correlation coefficients were calculated for 13 environmental factors. The results revealed that certain factors were highly correlated, with correlation coefficients of | r | = 0.834 for E3 and E1, and | r | = 0.804 for E2 and E7. Based on the Percent Contribution (PC) and Permutation Importance (PI) values from the preliminary experiment (Table2), factors E3 and E2 were excluded. Ultimately, 11 environmental factors were retained for modeling.

Using the ENMeval package, optimal parameters were determined for the three pollinator orders. The results indicated that the optimal parameters were: RM = 3 and FC = LQ for HYM; RM = 3 and FC = LQHPT for LEP; and RM = 2.5 and FC = LQHP for DIP (Figure 3). Optimized parameters were applied to the Maxent model, yielding improved accuracy compared to default settings. The average AUC values were 0.793 ± 0.023 (TSS = 0.53) for HYM, 0.803 ± 0.016 (TSS = 0.56) for LEP, and 0.853 ± 0.028 (TSS = 0.55) for DIP, validating the reliability of the predictions.

Maxent model results (Figure 4) indicate that pollinator habitats are primarily distributed within the Laoshan Mountain forest area, extending gradually to surrounding farmland and river habitats. Habitat area varies by pollinator order, with DIP exhibiting the smallest and LEP the largest. (1) HYM habitats are centered on the Laoshan forest area, with dispersed patches across rural areas in the region. The northern region, dominated by high-standard farmland, exhibits reduced ecological carrying capacity for HYM due to habitat homogenization. Conversely, the southwestern economic forest area supports HYM habitats due to its diverse vertical vegetation structure. (2) LEP habitats are primarily concentrated in the Laoshan forest area, with secondary aggregation centers in the northern Yongning-Tangquan ecological zone and southwestern village clusters. This distribution reflects ecological niche differentiation among LEP families: Pieridae predominantly occupy open agricultural areas, Papilionidae favor forest ecosystems with complex terrain and dense vegetation, and Lycaenidae are concentrated in grasslands with robust herbaceous layers and marginal habitats. The study area’s multidimensional habitat gradients effectively sustain LEP community diversity. (3) DIP habitats are primarily located in Laoshan Mountain forests, with scattered patches in rural areas and communities to the north and west. This distribution is attributed to the saprophytic nature of DIP larvae, as forests provide optimal saprophytic substrates and microhabitats for their development.

Landscape pattern characteristics of the 66 villages were quantitatively analyzed (Figure 5). The southwestern region displays high fragmentation, with the Shannon Diversity Index (SHDI) exhibiting a concentric pattern of low values in the central core and moderate values in the outer transition zone. The CONTAG index is inversely correlated with SHDI. In the northern region, fragmentation increases along the north–south axis, while SHDI decreases, with low CONTAG values indicating landscape dispersion in the paddy field–fish pond system. The central area shows low fragmentation and high CONTAG, but single land cover results in low SHDI. The southeastern region, however, exhibits both low fragmentation and high diversity.

Six landscape indices for habitat patches of the three pollinator orders were calculated across the study area and normalized (Figure 6). LEP habitat patches exhibit greater habitat advantage than those of the other two orders, corresponding to their larger suitable habitat area. LEP habitats also demonstrate higher integrity, with elevated normalized PD and 1/AREA_MN values. HYM habitats show reduced edge complexity and greater patch fragmentation. DIP habitat patches exhibit the lowest dominance among the three orders, yet display low landscape fragmentation, with patches concentrated in the central mountainous areas and surrounding regions.

PLAND, PD, LPI, ED, 1/AREA_MN, and FRAC_MN were assigned weights of 0.111, 0.147, 0.150, 0.405, 0.100, and 0.087, respectively. Higher normalized PD and 1/AREA_MN values indicate greater habitat integrity. Edge complexity, habitat integrity, and habitat advantage for the three insect orders were derived through weighted calculations. The results are presented below:

1. HYM core habitats are concentrated in the central mountainous forest area, exhibiting significant ecological advantages and structural integrity (Figure 7). These habitats are characterized by high vegetation cover, well-developed vertical structure, and low edge complexity, forming a stable ecological unit. The southwestern sub-core area displays heterogeneous habitat quality with variable dominance and integrity. The northern area exhibits a fragmented habitat mosaic. Although small continuous habitats exist, the overall pattern limits stable population development.

The weights for CONTAG, SHDI, and PD were 0.170, 0.214, and 0.616, respectively. The arithmetic mean of the suitability values for the three insect orders was calculated to generate the final suitability map (Figure 10).

1. High-suitability villages (0.612–0.713): 19 administrative units, including Chalukou and Huaguang. These villages are primarily located around mountains and forests or along the Lvshuiwan wetlands of the Yangtze River. Forest cover exceeds 20%, with undulating terrain and dispersed settlements. The water system comprises primarily scattered ponds and ditches. Landscape pattern analysis reveals low patch density and high SHDI, indicating high habitat heterogeneity and ecological integrity, which support abundant habitats and breeding resources for pollinators. Southeastern villages benefit from robust shoreline protection policies, further enhancing habitat stability.

The findings revealed a “three-tiered gradient pattern” of pollinator habitats in Pukou District, synergistic hotspots at natural–artificial interfaces, and differential impacts of roads and settlements on habitat distribution. These findings demonstrate pollinators’ ecological adaptability to habitat heterogeneity and underscore the critical role of human activities in shaping habitat patterns. Future research will investigate the ecological mechanisms driving habitat distribution differentiation and synergy, examine the influence of landscape patterns on insect diversity, and develop targeted conservation and restoration strategies based on village suitability evaluations to enhance rural biodiversity conservation in Pukou District.

By optimizing the Maxent model’s parameters, this study markedly enhanced habitat prediction accuracy compared to studies using default settings. This study revealed that diverse RM and FC settings are required for different pollinator orders, reflecting their distinct ecological niches and environmental adaptations. The high AUC and TSS values obtained demonstrate that the optimized model is highly applicable to the study area and serves as a reliable tool for habitat assessment in regions with comparable ecological conditions. However, the limited DIP sample size may constrain prediction reliability, and future studies should include larger sample sizes to enhance model generalization.

The findings demonstrate that pollinator habitat distribution follows a “three-tiered gradient pattern” (mountain–forest core, agricultural transition zone, urban fringe), closely tied to the ecological adaptability of different pollinator orders. For instance, LEP’s extensive suitable habitat area reflects its robust adaptability to diverse habitats, particularly those with high vegetation diversity and complex terrain. This aligns with Gomez et al. (2014), who reported that LEP species exploit diverse ecological niches due to their varied host plant requirements. Conversely, DIP habitats are primarily concentrated in forest areas, reflecting the saprophytic larval requirements for specific microhabitats. The synergistic pattern at natural–artificial interfaces highlights pollinators’ ecological flexibility in exploiting resources at habitat edges. This finding aligns with Caitano et al. (2020), Cavigliasso et al. (2022), who noted that habitat edges frequently serve as ecological hotspots for pollinators due to resource concentration. However, the urban fringe exerts a pronounced inhibitory effect on habitat distribution due to fragmentation and pollution, particularly for HYM. Thus, future conservation efforts should prioritize enhancing habitat connectivity at the urban–rural interface to mitigate urbanization’s adverse impacts.

Kernel density analysis (Figure 12) reveals that habitats of the three pollinator orders form synergistic hotspots at natural–artificial interfaces, including Xiangtang Village, the Laoshan–university campus transition zone, Hou Chong Wetland Park, and the Fenghuang Mountain area. The high SHDI index in the hot area suggests that habitat heterogeneity fosters interspecific coexistence. Moreover, Synergistic hotspots arise from resource sharing and interspecific interactions. The diverse landscapes of wetlands and transition zones offer varied food sources, supporting the foraging behavior of HYM and LEP. Water sources boost pollinator activity by sustaining humidity gradients and vegetation diversity. Conversely, DIP rely on saprophytic environments, restricting overlap with HYM and LEP, yet they exhibit Local synergies in ecological transition zones at forest edges.

Compared with European agricultural landscape studies (Kremen et al., 2007), Pukou District’s synergistic hotspots are predominantly located in areas of high ecological integrity, emphasizing the pivotal role of forests and wetlands over single agricultural areas. This synergy bolsters ecosystem stability and sustains pollination services for agriculture. However, urban expansion threatens connectivity in these hotspots, underscoring the need to prioritize protection of core habitats and ecological corridors.

Landscape pattern analysis reveals pronounced spatial heterogeneity in habitat suitability across the 66 villages in Pukou District. High-suitability villages, primarily located in mountainous and forested areas, exhibit low patch density and high SHDI, indicating that complex landscape patterns support pollinator diversity. This aligns with Marini et al. (2014), who reported that highly heterogeneous landscapes sustain greater species richness. Conversely, low-suitability villages, characterized by intensive agriculture, exhibit high patch density and low SHDI, resulting in habitat homogenization and diminished pollinator suitability. Moderate-suitability villages display varied landscape patterns, with some demonstrating landscape–habitat synergy, while others are constrained by structural deficiencies. K-means clustering results underscore the potential for targeted improvements in these villages, particularly those with high landscape heterogeneity but limited habitat functionality. These findings highlight the critical role of landscape pattern optimization in enhancing habitat suitability and offer a scientific foundation for village-level conservation planning.

This study finds roads and settlements are primary human disturbance factors, given their pronounced spatial gradient effects in Pukou District’s rural landscape. Within a 0–300 m road buffer, habitat proportions for the three pollinator orders were 69%, 65%, and 71%, respectively, suggesting that roadside vegetation strips serve as foraging and migration corridors for HYM and LEP. Conversely, within a 0–300 m settlement buffer, habitats accounted for 47%, 40%, and 33%, reflecting the suppressive impact of intense human activity (Figure 13). Gradient analysis reveals that anthropogenic activities shape pollinator distribution by modifying habitat integrity and resource availability, and that road buffers can function as ecological corridors. These findings offer a scientific foundation for village suitability assessments and conservation strategies.

To realize mutual benefits for pollinator habitat conservation and rural development in Pukou District, we suggest a four-tier strategy of core protection, ecological corridor connection, agricultural optimization, and community participation.

1. Core Protection Areas. For high-suitability villages (e.g., Chalukou, Huaguang), conservation should: (a) designate forests and wetlands as permanent ecological zones, prohibiting road expansion or mining; (b) establish 50–100 m wide vegetation buffers at forest–town edges; and (c) adopt sustainable land-use policies aligned with Nanjing’s 2021–2035 Spatial Plan.

In summary, high-, moderate- and low-suitability areas were proposed as core protection, target areas for quality improvement and pilot areas for restoration, respectively. Based on policy incentives, technological support and public participation, this study offers a scientific basis for the effective protection of pollinators in Pukou District and a methodological guideline for the biodiversity conservation in the rapidly urbanized areas through optimized Maxent modeling and landscape-scale approaches.