1 INTRODUCTION

Since 2019, the Chinese government has initiated the integration and optimization of protected areas to address four critical issues: multi-management, cross-overlapping, uncertainty of boundaries and numbers of NRs, and conflicts between nature conservation and social development (Wu et al., 2019, 2020). The primary challenge in these conservation efforts stems from land-use conflicts, particularly because territorial spatial planning and the farmland “red line” policy prioritize the protection of permanent basic farmland and scattered urban areas within nature reserves (NRs), hindering their conversion into protected areas (Ministry of Natural Recources of the People's Republic of China, 2022, 2023). This farmland “red line” policy is very restrictive and mandates the preservation of the current 1.87 billion mu (1 mu = 666.67 m²) of farmland, stipulating that plots of farmland larger than 3 mu in mountainous regions and 5 mu in plain areas should remain as permanent basic farmland (Ministry of Natural Recources of the People's Republic of China, 2022).

However, due to the historical context of agricultural land distribution and the initial approach to conservation planning, which favored large-scale and preliminary designs, many of China's NRs include farmland and human settlements (Chen et al., 2022; Shackelford et al., 2015; Xu et al., 2019). This has led to the presence of scattered nonconservation lands within NRs, disrupting the homogeneity, integrity, and continuity of their natural landscapes. These nonconservation lands or “skylights” represent a significant alteration to the intended ecological continuity of NRs.

“Skylights” or unprotected areas within NRs, act as conduits for human-induced stressors and pose significant challenges for long-term biodiversity conservation. Numerous studies have highlighted the growing “anthropogenic state” adjacent to NRs, undermining their efficacy and breaking the direct correlation between the extent of protected areas and successful nature conservation (Buxton et al., 2017; Fuente et al., 2020; Geldmann et al., 2019; Jones et al., 2018). Geographically, protected areas with higher exposure to these unprotected matrices are increasingly susceptible to current and future anthropogenic stressors, including land cover changes, habitat fragmentation, resource extraction, biological invasions, and chemical and physical pollution (Fuente et al., 2020; Güneralp & Seto, 2013; Laurance et al., 2012; Liu et al., 2020; Veldhuis et al., 2019). These anthropic stressors can alter both the community composition of flora and fauna and the functional integrity of ecosystems within NRs (Ries et al., 2017).

Guangdong, one of China's most economically developed provinces, exemplifies the pronounced spatial conflicts present within NRs. The types and current distribution of “skylights” in these NRs and their effects are not clear. To address this, our study focused on identifying “skylight” features in Guangdong and examining the potential benefits of their removal from NRs. Utilizing high-resolution remote sensing data, we characterized “skylights” and formulated four different management scenarios. We evaluated these scenarios based on two spatial attributes: exposure level and maximum depth. These attributes are crucial indicators of the effective size of NRs and can be used to determine “skylight” influences (Schauman et al., 2023). Our research provides quantitative evidence revealing the adverse effects of maintaining “skylights” within NRs and offers new perspectives on the strategic integration and optimization of NRs in China.

2 METHODS AND MATERIALS

3 RESULTS

3.1 “Skylights” increase the exposure levels of nature reserves

We identified two primary types of “skylights” from remote sensing data: urban land and farmland “skylights”. Urban “skylights” covered 68,702 grids (approximately 64.88 km²), forming 5.81% of the total “skylights” whereas farmland “skylights” spanned 1,113,755 grids (1051.76 km²), accounting for 94.19% of the total. These “skylights” were distributed in a scattered manner, ranging from as small as 1 grid cell to 18.14 km², with those smaller than 3 mu accounting for 63.98% of the total (<5 mu accounted for 66.27%) (Figures S2,and S3).

Using a gridded data set of internal distances to unprotected areas, we evaluated the statistical distribution using cumulative curves (Figure 1). In the existing configuration, which includes all urban and farmland grids, 82.83% of the NR areas were less than 1 km from the unprotected matrix; 92.23% were within 2 km, 95.39% within 3 km, 98.13% within 5 km, and 100% within 10 km.

When farmland was maintained in NRs, 82.74% of the NR areas were less than 1 km from the unprotected edges; 92.16% were within 2 km, 95.29% within 3 km, and 98.13% within 5 km. Maintaining only urban areas resulted in 73.12% of the NR areas being less than 1 km from the edges, with 90.53% within 2 km, 95.13% within 3 km, and 98.12% within 5 km. When both urban areas and farmland were removed, the proximity of NRs to unprotected edges reduced to 61.37% within 1 km, 82.31% within 2 km, 90.89% within 3 km, and 97.02% within 5 km. Statistically, removing “skylights” notably reduced the exposure level of NRs by 21.46% at a 1 km distance (contact zones) and by 9.92% at a 2 km distance.

3.2 Removing “skylights” increases the maximum depth of nature reserves

Only one NR polygon exhibited a maximum depth greater than 10 km, while 97.52% of the polygons had a maximum depth of less than 5 km (Figure 2). Through the K-W test and subsequent pairwise comparisons (Tukey Kramer Nemenyi and Dunn's all-pairs test), we evaluated the impact of two types of “skylights” on NRs' maximum depth. The results revealed significant differences among the four scenarios (chi-square (χ²) = 132.27, df = 3, p < 0.001). Compared with those in the control group, where both urban areas and farmland were removed, the maximum depth of NRs increased significantly (p < 0.001). Likewise, removing just farmland significantly increased the maximum depth of NRs (p < 0.001). However, the removal of only urban areas did not significantly affect the maximum depth (p = 0.983) (Figure 2). Therefore, “skylights” significantly reduce the maximum depth of NRs, and their removal, particularly of farmland “skylights” can substantially enhance this aspect (Figure 2).

In the mountainous region of Guangdong Province, considering the farmland “red line” policy, removing farmland “skylights” of less than 3 mu would lower the 1 km exposure level from 82.83% to 80.39%. This represents an 11.73% improvement, which, when combined with the optimal scenario-removing all “skylights” is notable, yet it does not significantly affect the maximum depth (chi-square (χ²) = 2.20, df = 1, p = 0.138) (for gradient analysis, see Tables S1 and S2). Conversely, when “skylights” of less than 5 mu were removed, there was a significant increase in maximum depth (chi-square (χ²) = 4.14, df = 1, p = 0.042), as shown in Table S2.

4 DISCUSSION

The exposure of NRs to unprotected areas has multiple implications for long-term nature conservation and human well-being. Compared to global levels, where one-third of protected areas are less than 2 km from their boundaries (Schauman et al., 2023), our study found that in Guangdong Province, more than 80% of the NR areas are within 1 km of human stressors. If “skylights” were removed, these high levels of exposure would largely decrease, and the maximum depth of NRs would significantly increase.

In the integration and optimization process of NRs, dealing with farmland “skylights” poses more significant challenges than urban areas. Farmlands and NRs often share similar distributions, being located in remote, less economically developed regions or at city peripheries (Margules and Pressey, 2000). A major issue is the Chinese government's commitment to maintaining 1.87 billion mu of farmland, preventing some of this land from being converted into protected areas (Ministry of Natural Recources of the People's Republic of China, 2023). Therefore, in terms of quantity, distribution characteristics, and restrictive policies, farmland has inevitably become the primary type of “skylight”.

Furthermore, from a landscape-human health connection perspective, the prevalence of farmland “skylights” within NRs may play a key role in future pathogen spillover and pesticide pollution across natural habitat-matrix boundaries. These interactions have historically led to disease outbreaks, population declines, and even pandemics among wild populations (Faust et al., 2018; Jones et al., 2008; Tilman et al., 2001; Wilkinson et al., 2018).

Additionally, “skylights” reduce the maximum depth of NRs, leading to discontinuous natural habitats and directly affecting species dispersal. Maximum depth is strongly correlated with the effective size of the protected area (Schauman et al., 2023), which indicates that protected areas with greater maximum depths usually exhibit better conservation performance. Urban areas and farmland, which are landscapes with high degrees of human modification, are generally unsuitable for wildlife and often obstruct species movement (Cao et al., 2020). Consequently, “skylights” perforate, disrupt, and fragment the continuous natural environment of NRs, potentially compromising the ability of governments to effectively safeguard nature in the long term (Game, 1980).

The small size and anomalous shape of Guangdong's NRs also affect their exposure levels and maximum depth. These NRs show a marked decrease on average size across national (225.68 km²), provincial (83.32 km²), municipal (39.43 km²), and county levels (19.59 km²). National NRs constitute a minor fraction (3.98%) of total NRs but cover 20.24% of the total protected areas. Smaller reserves inherently face higher exposure levels, regardless of proximity to other NRs. Typical long and narrow-shaped NRs, such as those for amphibians, reptiles, and fishes in Guangdong, which are normally established along rivers, usually have higher exposure levels due to their extensive boundary lines and lower maximum depths despite not always hosting the most biodiversity.

It is important to note that this study has several limitations. It mainly considers farmlands as exposure stressors within NRs, overlooking diverse stressors from outside, and uses maximum depth to represent effective NR size. It also excludes the “skylights” formed by mining areas due to limitations of remote sensing data, possibly leading to underestimated exposure levels.

Despite these limitations, our study highlights the increasing exposure levels produced by farmland “skylights” between NRs and unprotected areas during integration and optimization. The effect of these “skylights” on conservation effectiveness warrants further exploration. We recommend initially targeting “skylights” smaller than 3 mu (0.02 ha), prevalent within NRs, aligning with existing agrarian policies to reduce exposure levels. Addressing larger “skylights” should follow to achieve better conservation outcomes.

AUTHOR CONTRIBUTIONS

Haiyang Gao: Data curation; funding acquisition; methodology; resources; writing—original draft. Di Zhu: Conceptualization; formal analysis; supervision; validation; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The author(s) declare no conflicts of interest.