2.1 Study site

Xishuangbanna is located in the southern part of Yunnan Province, China, at 21°09′–22°36′N latitude and 99°58′–101.50′E longitude. It is characterised by a western tropical monsoon climate, marked by distinct wet and dry seasons, ample warmth and favourable climatic conditions. These factors contribute to the region's significant productivity potential, fostering a rich diversity of biological resources and resulting in China's most expansive tropical forest vegetation.

Established in 1958, the Xishuangbanna Nature Reserve stands as one of China's earliest protected areas and serves as a typical tropical forest reserve. Spanning an area of 2425 km², it comprises five subreserves, including the Shangyong and Mengla subreserves, which border Laos and were chosen as focal points of this study. The Shangyong Reserve is located in the south of Mengla County, with an area of 311.84 km². The terrain is lower in the east and north and higher in the south and southwest. The main types of vegetation are tropical rainforests, mixed broad bamboo forests, evergreen broad-leaved forests, alpine meadows, shrubs and so forth (Lin et al., 2008). The Mengla Reserve lies to the south of the Shangyong Reserve; its topography features a central lowland flanked by elevated areas on both sides, resulting in large altitude variations. The total area of the reserve is 926.83 km². The main types of vegetation are tropical rainforests, subtropical green broad-leaved forests and monsoon forests (Lin et al., 2014).

Xishuangbanna is an ethnically diverse area inhabited by 13 distinct ethnic groups distributed in the reserve and its surrounding areas, including the Dai, Han, Jinuo, Tibetan, Yao, Hani and Jingpo groups. Over the last six decades, factors like population growth, economic markets and land policies have drastically altered the local agricultural landscape. Natural forests have been largely converted to rubber plantations, and there has been a marked increase in tea cultivation. The explosive growth of rubber cultivation has negatively impacted biodiversity, with the destruction of natural forests leading to the fragmentation of wildlife habitats (Liu et al., 2017). Cultural services have also changed due to economic development and social modernisation. For example, traditional conservation concepts have been abandoned by young people, affecting not only cultural inheritance but also the protection of local ecosystems and biodiversity (Infield et al., 2017). Therefore, an integrated assessment of CES is imperative for the conservation and management of tropical nature reserves from both cultural and ecological value perspectives.

Xishuangbanna is one of the most biologically and culturally diverse areas in China (Verschuuren et al., 2010). Numerous studies have documented the positive influence of indigenous culture in Xishuangbanna, such as sacred sites, in promoting biodiversity conservation (Corlett, 2018; Liu et al., 2002; Liu et al., 1992; Verschuuren & Furuta, 2016; Verschuuren et al., 2010). Evidence shows that despite their smaller size, these sacred sites possess biodiversity levels comparable to those of government-established nature reserves (Verschuuren et al., 2010). However, a research gap exists regarding the indigenous cultures and CES in Xishuangbanna, and decision-making in the management of nature reserves has largely overlooked CES.

2.2 Data collection

We modified the Guidance for the Rapid Assessment of Cultural Ecosystem Services (GRACE) (Infield et al., 2015) for data collection. There are three management zones in the Shangyong sub-reserve and five in the Mengla subreserve. Detailed information about the village interview is provided in SI (Supporting Information: SM.1). The interview was comprised of three sections: background information, CES information and the main threats to the reserve. The second section included the introduction to the survey, the description of the CES, and the importance score for each CES under current and alternative scenarios (see Supporting Information: SM.2). The questionnaire design was based on the TESSA toolkit (Peh et al., 2017). Two scenarios were described to the interviewees: the first depicted the current situation with the protection of the nature reserve in place, while the second described a situation without the protection of the reserve. For comparison, the alternative state that was proposed would primarily meet the villagers' cultivation and development needs. In this scenario, the area would no longer belong to a national nature reserve but instead belong to a more leniently protected area with reduced conservation effort and funding. A detailed description of the alternative state can be found in Liu et al. (2019).

Data on rivers, towns, settlement points and roads were obtained from digitised maps produced by the Yunnan Institute of Forest Inventory and Planning. The digital elevation model was downloaded from the Computer Network Information Center of the Chinese Academy of Sciences (Website: http://www.gscloud.cn). The vegetation map of the study area was generated by the Yunnan Institute of Forest Inventory and Planning by interpreting the SPOT-5 satellite remote sensing images from 2018. A 1:50,000 hydrographic network map and a digital elevation model were utilised as reference control images, and panoramic images were corrected for geometric biases using the ERDAS 9.2 software. The root mean square error was less than 1, indicating that the vegetation map met the precision requirements of the study. The vegetation in the study area was divided into 16 categories: tropical and monsoon forest, broad leaf forest, bamboo, coniferous forest, shrub rangeland, farmland, fruit garden, rubber plantation, tea plantation, other plantation, eucalyptus, water body, road, built-up development, bare land and unused land.

2.3 Data analysis

2.3.3 Conservation priority for CES

The following three spatial analyses were performed to determine the spatial distribution relationships: Getis-Ord General G, spatial autocorrelation (Moran's I) and average nearest neighbour. Hotspot analysis was used to obtain the distribution pattern of each CES type, while the environmental variables were analysed to predict their impact on the SolVES modelling distributions. The Wilcoxon signed rank test was conducted to explore the difference in importance scores between the two states, initially judging the impact of protected areas on different types of CES.

In this study, the impact of protected areas on the main types of CES identified by locals was considered, and the impact index (I_i) of protected areas on CES was defined. The derived formula is as follows:

where I_i represents the impact index of the given CES, CS_i (current score) represents the importance score of the cultural service value in the current state, AS_i (alternative score) represents the importance score of the cultural service value in the alternative state, RI_i is the relative importance of an ecosystem service, CS_max represents the maximum value of all CS_i and CS_min represents the minimum value of all CS_i. The larger the I_i, the greater the impact of the protected area on the given CES and vice versa.

Combining the impact index and the frequency of cultural service types, that is, the perceived frequency, the cultural service conservation priority index (PI) was defined as the impact of protected areas on each type of cultural service (I_i) multiplied by the perceived frequency of that cultural service. A larger PI indicates that the type of cultural service is affected by the protected area and should receive greater priority in local planning and management. Since the investigation phase in the Mengyuan zone was the wording adjustment phase, the score data for this zone were excluded to avoid result bias.

3.1 Description and characteristics of CES

A total of 147 valid questionnaires were obtained from village interviews, which covered all management zones in the study area. The responses represented 52 natural villages and farm production teams, the two predominant village types within the nature reserve. A total of 150 CES points were obtained from the participatory mapping. The ratio of male to female respondents was approximately 68:32, indigenous people accounted for 73% of the respondents and people aged 31 to 50 accounted for 71% of the respondents. The main ethnic groups among the respondents were Hani (37%), Yao (22%), Dai (17%) and Han (11%). The ethnic groups in the minority were Yi, Zhuang, Lahu, Bai and Bulang.

Through the trial interviews in Mengyuan, the following eight types of CES were determined: spiritual and religious, folk and custom, aesthetic and inspirational, educational, cultural heritage, sense of place and identity, leisure and recreational and wildlife. Definitions and descriptions of these types of CES facilitated the interviewees' understanding during the subsequent interview process in the whole study area. The CES types are shown in Table 1. The frequency of mentions for a particular CES type was used to determine perception frequency (Figure 1a). The main types of CES perceived by local residents were wildlife (78.23%), leisure and recreational (73.47%), folk and custom (57.14%) and sense of place and identity (54.42%).

We digitised and superimposed the 150 points of CES locations obtained during the village interviews to present the overall distribution of cultural service value (Figure 1). The CES points were distributed within the protected area or adjacent to the protected area. In addition, the CES types categorised as recreational, folk and custom and spiritual and religious exhibited a more expansive distribution across the study area. The distribution and score of cultural service value in the unprotected state was notably lower than in the protected state. Although the value of wildlife was frequently perceived, assigning a specific location was difficult; therefore, fewer wildlife points were observed on the distribution map. Under the alternative state, there was a significant decrease in scores pertaining to leisure and recreational value. Additionally, some scores for the CES types such as wildlife, aesthetic and inspirational and sense of place and identity were almost entirely absent. However, the scores for the CES types related to spiritual and religious, as well as folk and custom, remained almost unchanged.

In the current value assessment, CES points were randomly distributed. Conversely, in the alternative state value, the CES points and the changes between the current and alternative states showed a clustered distribution. This indicated that some CES areas were more significantly affected by the disappearance of protected areas (Supporting Information: Table S3).

3.2 Relationship between CES and environmental variables

The type of CES, leisure and recreational, earned the most points (N-COUNT) and the highest maximum value index (M-VI), followed by aesthetic and inspirational, folk and custom and spiritual and religious in the interviews Table S4. Similar to wildlife, it was difficult to define specific locations for the educational and sense of place and identity CES types; thus, these earned lower points. The majority of the CES types were in a relatively clustered distribution (R < 1).

The leisure and recreational, aesthetic and inspirational, folk and custom, spiritual and religious, sense of place and identity, and educational CES types were selected to build a value mapping model, as these had higher points and indices. The predicted distribution of the six CES types is shown in Figure 2a–f, while the main environmental variables affecting their distribution are presented in Table 2. The detailed impact of each variable is available in the Supporting Information: Figures S2–S7.

Among the six types of CES analysed, natural education had a training AUC value of 0.742, while the other types had training AUC values above 0.893 (Supporting Information: Table S5), indicating the consistency of the model with the study area. The degree of matching was high, and the results of the cultural value distribution exhibited significant reliability and reference value (Elith et al., 2010; Swets, 1988).

The leisure and recreational values were perceived in the scenic mountains, forests and rivers around the villages according to village interviews (Table 1). Consequently, the distribution of leisure and recreational value was related to water, farmland and cash forest, with the distance-to-water (DTW) variable having the greatest impact (Supporting Information: Figure S2). There was a direct relationship between the closeness of people to water and the perceived value of leisure and recreational activities. A similar relationship was observed for the distance to farmland and cash forest.

The distributions of the folk and custom type and the spiritual and religious type were closely related to the locations of the farmland and villages (Table 1). There was a decrease in values as the distance increased (Supporting Information: Figure S3 and S4). The environmental variable of land use type exerted an impact on the distribution of the spiritual and religious types such as tropical monsoon forests, broad-leaved forests, rubber plantations, farmland or roads (Supporting Information: Figure S3). A fluctuation in the value was observed for the folk and custom types. The value increased as the distance from the natural forest increased, reaching a peak at 200 m, after which it decreased. The rise in elevation was another environmental factor whose increase caused the value of the folk and custom type to decrease (Supporting Information: Figure S4).

The aesthetic and inspirational type value was distributed mainly in tropical rain forests, broad-leaved forests, bamboo forests and other areas with beautiful scenery and quiet environments. VI decreased as distances to forests and roads increased. The VI increased with rising slopes, stabilising within a certain range when the slope reached approximately 23° (Supporting Information: Figure S5).

The sense of place and identity type distribution was related to the location of forests, and respondants mainly perceived it in farmland, rivers and forests near villages. According to Supporting Information: Figure S6, the distribution is concentrated within 50 m of the forest and is affected by the distances to the village and the road. The farther the distance, the lower the value. The distribution around the farmlands was significant for the educational type value, and as the distance to water increased, the value decreased (Supporting Information: Figure S7).

Value transfer mapping models for the leisure and recreational and folk and custom types were selected to predict the whole distribution in Mengla County. The models fitted the data well, with the testing AUC of 0.714 and 0.831, respectively, indicating potentially useful performance to transfer social values to a larger area, that is, the whole of Mengla County (Sherrouse & Semmens, 2015). Distribution characteristics were consistent with the results in the study area (Figure 2g,h). The leisure and recreational service was related to the distance to water, farmland, villages and forests. The distribution of the type of folk and custom was found to be affected by the location of the farmlands and villages and the elevation.

3.3 Impact of nature reserves on CES and conservation priority

The results revealed that the average importance scores of all CES in the alternative state were lower than those in the current state (Table 3). Specifically, the importance scores for the types of spiritual and religious, folk and custom, aesthetic and inspirational and sense of place and identity in the current state were very high (above 4.5). The types of leisure and recreational CES and wildlife CES showed the lowest scores of 3.96 and 3.43, respectively. Comparing the scores between the current state and the alternative state revealed that all CES, except the spiritual and religious type and cultural heritage type, showed significantly higher importance under the protected state than under the nonprotected state.

The impact of the two subreserve areas is illustrated in Supporting Information: Figure S8. The types of spiritual and religious CES, folk and custom and cultural heritage had lower impact indices, indicating less impact from the nature reserve. Meanwhile, the other CES types had larger impact indices. The leisure and recreational and wildlife types showed different impact indices in the Mengla and Shangyong reserves, with those for Mengla significantly higher than those for Shangyong (paired t-test, p < .05).

The conservation PI values of various types of CES are shown in Figure 3. Conservation priorities for spiritual and religious, folk and custom and cultural heritage types of CES were the lowest (PI < 1.0), and those for the aesthetic and inspirational type were relatively low. The conservation PI of the sense of place and identity, educational, leisure and recreational and wildlife types was relatively high (PI > 1.0).

Using a five-point Likert scale, we gauged residents' attitudes toward the management practices of these nature reserves (Supporting Information: Figure S9). Residents were relatively supportive (4–5) of the protection and management of landscapes, heritage and species. The results also indicated the relatively positive attitude of local residents (3–4) toward farmland, economic income, community activity spaces and educational resources related to livelihood. Residents' attitudes towards entertainment venues and tourism industries were relatively negative (2–3) because they were unable to enjoy the cultural benefits of nature reserves due to restricted entry. Moreover, the majority of respondents expressed opposition to the restrictions placed on local residents' access to the protected area. They also indicated that current management practices limit the access of local stakeholders to entertainment and leisure venues.

4.1 Distributions of CES

The leisure and recreational CES type received the maximum points in this study. The most important influencing factor was the distance from water because the specific recreational activities that local residents frequently mention are fishing in the river, picnics and so forth. Talking to others after farm work is another recreational activity that indicates the distance to farmland as an influencing variable.

The folk and custom type and the spiritual and religious type are important for the residents of Xishuangbanna due to the multiethnic characteristics of this area (Corlett, 2018; Liu et al., 2002; Liu et al., 1992; Zeng & Reuse, 2016). Many sites in Xishuangbanna that are used for spiritual and religious CES and folk and custom CES vary according to different ethnic groups. Therefore, these two types of CES are mostly related to villages and nearby farmlands, mountains and forests where residents live. The interview results indicate that most of the sites that fall under these two types of CES are actually natural sacred sites (e.g., holy hills and cemetery hills) (Liu et al., 2002; Luo et al., 2001; Verschuuren et al., 2010) protected by local residents. Further studies on religious beliefs and ethnic conventions will aid in the identification of these sacred sites and the understanding of their role in biodiversity and ecosystem conservation (Infield, 2001; Infield et al., 2017).

Wildlife is the most frequently mentioned CES type, but it was not analysed in the SolVES model because the locals could not indicate precise locations, resulting in a lack of points. However, the locals exhibited strong awareness and recognition of the importance of wildlife protection for future generations (Supporting Information: Figure S9). This offers a conducive avenue to prioritise wildlife protection and highlight affected areas so that residents can better understand and participate in biodiversity conservation.

4.2 Prioritisation of CES conservation

The most frequently mentioned CES types are wildlife, leisure and recreational, folk and custom and sense of place and identity. The PI indicated a higher conservation priority for leisure and recreational, wildlife, sense of place and identity, and educational CES types, as these are prone to degradation without protection measures. To garner local support, reserve managers should prioritise these types of CES in their policy considerations (Infield, 2001).

Spiritual and religious, folk and custom and cultural heritage CES are very important to local residents (with an importance score >4 in the current state), which is consistent with the study area being an ethnic minority region. The related locations are mainly cemetery hills and holy hills. In general, local residents are highly aware of the religious significance and preservation of these natural sacred sites. Nonetheless, certain villages revealed that their revered hillsides have been destroyed and replaced with cash forests, driven by the quest for local economic development. Hence, it is necessary to strengthen local ethnic beliefs to protect natural sacred sites from degradation through the promotion and documentation of local religious and folk activities and cultural education.

Conservation PI values for both the spiritual and religious, and folk and custom types, are notably low, indicating that the presence of nature reserves does not significantly alter the perception of residents regarding these CES. In a real conservation initiative, leveraging such indigenous conservation practices can reduce protection expenses and expand the scope of the conservation network. In their early research, Liu et al. (1992) indicated that the sacred sites in Xishuangbanna play an important role in the preservation of dry monsoon forests. Vernacular conservation is based on traditional resource management, cultural taboos and township regulations and conventions. By integrating these indigenous protective practices, specifically the protection of sacred sites, into China's protected area system, conservation outcomes can be enhanced with a more efficient allocation of resources and efforts.

Thus, incorporating conservation priorities grounded in CES offers a novel and effective application, especially in China's proposed Asian Elephant National Park, which is currently being developed (Kang & Li, 2022; Yang et al., 2021). For effective planning and development of the national park system, it's imperative to prioritise leisure and recreational, wildlife, sense of place and identity, and educational CES types due to their heightened vulnerability. Conversely, CES types such as spiritual and religious, folk and custom and cultural heritage may be more sustainably overseen through organic local conservation initiatives.

Although the reduction in economic income and available farmland caused by conservation management affects local support, access to recreational value remains the most influential factor in endorsing nature reserves. Many residents expressed that restrictions on entering the reserve limited their ability to freely enter the forest and invite relatives and friends to engage in entertainment and leisure activities. Therefore, the diminished leisure and recreational value of the villages outweighed the loss of economic income. To address this problem, reserve managers should conduct a comprehensive assessment of the advantages and disadvantages, reconsider and adjust access rights for local people, and strengthen the management of community engagement activities. With the consent of the local people, a moderate payment mechanism for ecosystem services can be introduced, granting them access at a low price. Indeed, This approach will give locals more options without imposing excessive fees solely to profit from the protected areas. Furthermore, a comanagement mechanism could be introduced, permitting local people to enter as volunteers. This approach could be adopted as a part of daily patrol or nature education activities, allowing local residents to enjoy entertainment and leisure activities while contributing to the reserve's management.

4.3 Study limitations

First, our study only mapped the distribution of individual service types, given the limitations of the SolVES model being unable to integrate all services in a comprehensive service distribution graph. Second, there was a lack of comparability among different services. Therefore, we recommend the incorporation of an overall CES distribution analysis into future models.

In terms of importance assignment, we did not determine whether it should be assigned a zero value or ignored if a CES type was not mentioned by the respondent. If these null values are ignored, more attention should be paid to the perception of each participant; if these are assigned a zero value for averaging, the result is reflective of the overall population. The survey sample in each village was small; therefore, it may not reflect the perceived importance at the village level and in the surrounding areas. For this reason, we focused only on perceived importance at the individual level. Furthermore, there are chances of errors in the perceptions of the respondents due to the random sampling method. Hence, deriving important indicators for the entire village population based solely on individual perceptions is a hasty approach. Another problem is the potential trade-off in the perceived importance when viewed at an individual level versus a collective population perspective.

In subsequent studies, an intensive and comprehensive survey across a specified area should be conducted to determine whether variations in sampling densities have an impact on the overall distribution of CES values. An additional factor to take into account is whether the results from individual interviews should be used directly as input or whether the results for entire villages should be aggregated and averaged before running the model. In this study, we conducted a relatively intensive survey of the Mengyuan management zone (49 interviews in Mengyuan villages vs. a total of 147 interviews) and discerned that except for two services (i.e., folk and custom CES and spiritual and religious CES), the locations of all the other services varied. A large number of scattered CES points introduced complexities and ambiguities to the analytical process.

4.4 Moving forward for CES involvement in nature conservation

In the context of globalisation, it is necessary to prioritise collaborative research in biodiversity and cultural diversity, harnessing cultural diversity as a tool to protect biodiversity. Areas with high biodiversity are often inhabited by ethnic minorities, and cultural diversity is highly important (Pretty et al., 2009). One-third of the world's land and more than 40% of protected areas are owned and managed by traditional indigenous people (Inouye, 2014). Most nature reserves and rural areas in China have multifunctional landscapes where nature and culture integrate, but a corresponding biological and cultural diversity protection management system is lacking. Beyond policy frameworks and government-led initiatives, attention should be paid to the influential role of local culture, including village regulations and traditional ecological knowledge, to promote coordinated protection of regional biodiversity and cultural diversity, leading to sustainable community development (Bi et al., 2020).

In southeast and southwest China, informal rules and traditional customs govern many forested areas, such as sacred mountains (Shen et al., 2015; Xu et al., 2005), community forests (Gao et al., 2018), and Fengshui forests (Hu et al., 2011). Gao et al. (2018) suggested that government agencies incorporate culturally protected forests into policy and legislative frameworks and introduce ecological compensation mechanisms. However, Allendorf et al. (2014) indicated that strategies for directly integrating these areas into conservation strategies may be inappropriate, as people primarily view these sacred forests as religious sites and do not associate them with ecological value.

For a holistic approach to ecosystem services management, it is imperative to incorporate CES methodically. Many kinds of initiatives, including spatial planning, ecosystem-based management, integrated conservation-development schemes and payment for ecosystem services, stand to gain from a structured consideration of CES. However, despite their significance, many CES are overlooked in ecosystem services research, resulting in their limited incorporation in decision-making processes (Chan, Guerry, et al., 2012).

Integrating CES into the planning and management of nature conservation is not an easy task. An approach based on cultural and religious values tends to be more sustainable compared to those based on legislation or regulation (Liu et al., 2002). However, the deployment of local and informal regulations faces challenges due to globalisation and state-led economic modernisation. This is evident from the decrease in community-level customs and cultural practices as socioeconomic development progresses (Gao et al., 2018; Infield et al., 2017). In some cases, even though the small size of the CES sites might lead to minimal ecological value for local residents, these sites can serve as platforms to enhance understanding of the environment and to highlight the importance of protected areas for biodiversity (Allendorf et al., 2014). Our study contributes to the existing body of knowledge by paving the way to promote CES in nature conservation.

In conclusion, CES should be taken into consideration in our nature conservation strategies. Our study, based in two government-established tropical nature reserves in Xishuangbanna, SW China, identifies eight distinct types of CES. Notably, leisure and recreational and wildlife CES, both with high PI, rely on the formal protection of nature reserves. Conversely, spiritual and religious CES, and folk and custom CES, with lower PI, were protected primarily by community-based conservation. Leveraging and incorporating this vernacular conservation in our existing protected area system can reduce protection costs and extend the conservation network. Simultaneously, maintaining and enriching the cultural well-being of local residents through the safeguarding of vulnerable CES will help gain more local support.