Protected areas (PAs) are crucial for biodiversity conservation, but their effectiveness varies. While previous research often focuses on PA extent, representativeness, and governance, functional connectivity—the ease of animal movement within habitats—is frequently overlooked. This study addresses this gap by evaluating the effectiveness of China's NNRs in sustaining bird and mammal populations, considering functional connectivity at both intra- and inter-population scales. The Kunming-Montreal Global Biodiversity Framework emphasizes the need to enhance functional connectivity, highlighting the importance of this research. Most analyses of PA effectiveness have overlooked habitat quality, assuming PAs provide suitable habitats. However, habitat quality, influenced by natural landscape heterogeneity and human activities, limits wildlife movement and reduces functional connectivity. Maintaining well-connected habitats within and between PAs is crucial for long-term wildlife survival. Previous studies often focus on a single scale, lacking comprehensive inter-regional comparability. This study aims to address these limitations by integrating spatial and temporal scales to evaluate functional connectivity within NNRs across China. Habitat availability and type diversity are key factors regulating functional connectivity, but human-dominated habitat loss and fragmentation pose significant challenges. NNRs, while covering 10% of China's land, are often established based on administrative boundaries rather than biodiversity needs, leading to geographical variations in size and effectiveness. This research uses graph theory to model functional connectivity, assessing the impact of NNRs on 11,424 bird and terrestrial mammal populations with varying habitat preferences and mobility. The study specifically aims to determine the effectiveness of NNRs in conserving populations, identify factors regulating their effectiveness, and quantify the impact of artificial landscapes on connectivity.

Existing literature highlights the multifaceted challenges of evaluating protected area effectiveness, focusing primarily on total area, representativeness, and governance. Functional connectivity, a fundamental dimension impacting ecological effectiveness, has been largely ignored in previous studies. While some studies have explored functional connectivity within PA networks and individual PAs, there's a lack of comprehensive, inter-regionally comparable analyses across different biological taxa. Research emphasizing the establishment and expansion of PAs to improve functional connectivity overlooks the impact of habitat quality on wildlife populations. Studies have shown that habitat loss and fragmentation significantly reduce population flow and gene exchange, increasing extinction risk. The impact of transportation infrastructure on biodiversity, particularly within PAs, is also a well-documented concern. Several methodologies exist for modeling functional connectivity, with graph theory emerging as a valuable tool requiring less data input while providing reliable visualizations of ecological networks. The use of virtual species based on ecological niches allows for more standardized comparisons between regions and habitats.

This study utilized graph theory, specifically the Graphab software, to model functional connectivity within China's 474 NNRs. The methodology involved creating ecological networks where nodes represent habitat patches and links represent potential movement between them. A least-cost path (LCP) analysis was employed, requiring two primary inputs: a landscape resistance surface and species movement ability. The landscape resistance surface incorporated land cover (from ChinaCover 2010 and 2020 data, supplemented by OpenStreetMap data), topography (from ASTER GDEM v2), and human activities (from HumanFootprint v2 and Global Human Modification data). Resistance values were assigned to different landscape features, with higher values for artificial landscapes (roads, settlements) reflecting greater impedance to movement. The movement ability of species was represented using a virtual species approach based on six virtual species groups (birds and mammals in coniferous, broadleaf, and grassland/wetland habitats). Dispersal distances (10 km and 50 km representing low and high mobility) were determined based on the 25th and 75th percentiles of NNR minimum enclosing circle diameter. Allometric relationships were used to derive ecological traits like median daily movement distance, key patch area for meta-populations, and species geographic range area. NNR effectiveness was defined based on whether the minimum area within the NNR was sufficient to support a minimum meta-population size for long-term survival, considering both intra- and inter-population connectivity. The proportion of effectively protected populations within each NNR, biome, and nationally, was calculated. Random forest and generalized additive models (GAMs) were used to analyze the influence of habitat characteristics (habitat area, artificial landscape intensity, ecosystem diversity, NNR area) on functional connectivity. Sensitivity analyses were conducted to evaluate the robustness of the findings by examining the impact of varying dispersal distances and diets. The study compared functional connectivity with and without artificial landscapes to assess their impact on conservation effectiveness.

China's NNRs effectively protected 57% of bird and 42% of terrestrial mammal populations, totaling 50% for both groups. However, for high-mobility species, this dropped to less than 25% for birds and 13% for mammals. Functional connectivity varied across habitat types and biomes, with forest and grassland habitats showing higher connectivity than wetland habitats. The effectiveness of NNRs was significantly impacted by their size and the intensity of human activity. Smaller reserves with high human pressure displayed lower conservation effectiveness. Artificial landscapes, comprising less than 2% of NNR areas, caused nearly 40% of connectivity loss for highly mobile mammals. Random forest analysis revealed that the proportion of habitat area was the most important factor influencing functional connectivity, but the relative importance of other factors varied regionally. In eastern China, artificial landscape intensity was more influential, while in western China, NNR area and ecosystem diversity played larger roles. Increasing habitat area improved functional connectivity, while increasing artificial landscape intensity decreased it. Ecosystem diversity exhibited a quadratic relationship with functional connectivity, peaking at medium levels. The impact of human activities negatively correlated with NNR effectiveness. Reserves with lower human footprints showed substantially higher effectiveness, particularly for high-mobility species. Removing artificial landscapes improved functional connectivity but did not fully restore conservation effectiveness. Conservation effectiveness was lower for high-mobility mammals than for birds with the same dispersal ability, likely because birds rely more on flight. However, non-flying birds remained sensitive to habitat fragmentation. Eastern China, despite having a high population density and high human pressure, showed poorer protection for high-mobility species compared to western regions.

The study's findings highlight the limitations of China's current NNR system in safeguarding highly mobile species. The strong correlation between small reserve size, high human activity, and low conservation effectiveness underscores the need for a paradigm shift in PA management. The disproportionate impact of artificial landscapes on connectivity emphasizes the crucial need to mitigate human activities within NNRs. Regional variations in the factors influencing functional connectivity suggest that a common-but-differentiated management strategy tailored to specific geographic contexts is necessary. Habitat area is consistently identified as a crucial factor, but managing the impacts of human activities is equally critical, especially in densely populated areas. The study provides important insights into the dynamics between human activity, habitat quality, and the long-term survival of wildlife populations, challenging the assumption that simply expanding the area of existing PAs will automatically improve biodiversity protection. The study also highlights the importance of considering species-specific traits when evaluating conservation effectiveness, recognizing that different species have varying vulnerabilities to human activities.

This research demonstrates the inadequacy of current China's NNRs in effectively protecting highly mobile bird and mammal populations. The findings emphasize the critical need for a more comprehensive approach to PA management that considers functional connectivity, reserve size, and the impact of human activities. Future conservation strategies should prioritize increasing habitat area, minimizing human disturbance within reserves, and implementing region-specific management strategies. Investing in larger reserves and optimizing existing PAs is vital, alongside enhancing connectivity between reserves and developing inter-regional cooperative efforts. Further research should focus on more detailed modeling of species-specific movement patterns, the synergistic impacts of multiple human activities, and more effective methods of mitigating the negative effects of infrastructure on wildlife movement.

The study's reliance on virtual species, while enabling broader comparisons, may not fully capture the nuances of individual species' responses to environmental changes. Data limitations, particularly concerning the precise distribution of roads and other infrastructure, may lead to underestimation of their impacts on connectivity. The temporal aspect of the study focuses on the year 2020, and future changes in land use and human activities may significantly alter functional connectivity. The model's assumptions concerning resistance values and species movement may also introduce uncertainties. Further research is necessary to investigate these uncertainties and refine the model.