Introduction

In recent years, people’s preference for rural tourism has gradually increased with the rapid pace of life and growing pressure in urban areas1, especially during the COVID-19 pandemic2,3. Rural tourism, in various forms4, can provide a range of experiences, such as natural environments, rural scenery, historical and cultural heritage, and folk customs5,6,7,8. Currently, there is no consensus on the definition of rural tourism9. In this study, rural tourism is a small-scale business operation run by farmers, with rural landscape resources as the operating assets and providing services to tourists as the operating means. The development history of rural tourism in many countries illustrates its positive role in promoting rural economic and social development10,11,12,13. Furthermore, scientifically developing rural tourism can preserve heritage landscapes that were previously unknown or were on the brink of extinction and irreproducible14,15,16.

A traditional village is a settlement with profound historical and cultural heritage and preserved traditional production and living patterns17. These villages generally possess significant historical, cultural, and ecological value and are essential components of rural heritage. However, the development of traditional village tourism in remote areas has been slow due to underdeveloped tourism and poor road accessibility during historical periods. This phenomenon is particularly evident in high-altitude mountainous areas18. Traditional villages in mountainous regions have unique natural scenery, pleasant climates and rich cultural resources19,20, but they are small in scale and low in agricultural mechanization21. In recent years, some traditional villages in mountainous areas have gradually become tourist destinations, relying on rural revitalization and the planning and construction of transportation networks11. Villages with good road network connectivity are particularly favored by tourists18. In terms of road network density, people tend to engage in rural tourism and leisure activities in traditional villages within a 1-h highway traffic circle, which is 10–20 km away from the central cities and counties11. Tourism has promoted economic development and improved people’s living standards in villages. At the same time, more employment opportunities have been provided to slow village population loss through the development of homestays and catering22,23, tourist attractions11,24, and leisure agriculture25.

However, while rural tourism has positive effects on the protection of traditional village landscapes, it may also have negative impacts due to excessive tourism development26,27. Therefore, it is crucial to balance landscape protection and tourism development in traditional villages28. Traditional village landscapes include ecological, production, and living landscapes (EPLLs)7,29, which are mutually dependent and influenced. In the context of rural tourism, EPLLs in traditional villages undergo a certain degree of evolution while meeting tourist needs30. However, the trend of landscape evolution and its relationship with rural tourism remain unclear. Previous studies on traditional villages mainly focused on spatial patterns and morphological evolution12,31,32,33,34. Nevertheless, research on landscape evolution from the perspective of Traditional Village Landscape Carrying Capacity (TVLCC) is still inadequate.

In this study, the TVLCC refers to the maximum intensity of human activities (such as tourism, construction and development) that the landscape of traditional villages can endure35. Some scholars have applied the carrying capacity method to study built-up areas like cities36,37, individual villages38, as well as scenic areas such as national parks39 and coastal zones40,41. However, there is a lack of research on the landscapes of traditional village groups. Additionally, in the quantitative measurement of village landscape evolution, most studies have focused on specific landscape element, such as the fragmentation evolution of farmland landscapes42, the structural and functional evolution of rural homesteads43, or changes in landscape pattern indices44. As a result, there is insufficient comprehensive analysis of the carrying capacity level of EPLLs. Conducting research on the TVLCC can fill these research gaps, clarify the landscape evolution trend and the impact of rural tourism on it, and improve the research system of traditional village landscapes.

This study takes the Xiangxi Tujia and Miao Autonomous Prefecture (hereinafter referred to as Xiangxi Prefecture), a typical mountainous area with a wide distribution of traditional villages, as an example to study the TVLCC. The research objectives cover three aspects. First, it aims to explore quantitative methods for delineating the boundaries of traditional village landscapes in mountainous areas and apply these methods in TVLCC evaluation. Second, it attempts to reveal the spatiotemporal evolutionary characteristics and influencing factors of the carrying capacity of EPLLs, and explore the relationship between the changes in landscape carrying capacity and the development of rural tourism as well as the protection policies. Third, it intends to propose suggestions for the protection of traditional village landscapes and the promotion of tourism development at both the regional and village scales, which will help enhance the TVLCC level in the future.

Methods

Study area

Xiangxi Prefecture is located in the western part of Hunan Province (Fig. 1a), covering an area of approximately 15,500 km2 (109°10′-110°23′ E, 27°44′-29°38′ N). The elevation of the region ranges from 80 to 1755 m, and it belongs to the Wuling Mountains within the Yunnan-Guizhou Plateau. The Tujia, Miao, and other ethnic minorities lived here for generations, resulting in the formation of numerous historically significant traditional villages.

Fig. 1: Study area, tourism data and distribution of traditional villages.
figure 1

a Map of the study area; b Schematic diagram of the annual number of tourists and tourism revenue; c Distribution map of traditional villages (LS Longshan County, YS Yongshun County, BJ Baojing County, GZ Guzhang County, HY Huayuan County, FH Fenghuang County, LX Luxi County, JS Jishou City).

Full size image

The period of 2016–2017 was important for rural tourism construction in Xiangxi Prefecture. Before 2016, the number of tourists and total revenue in Xiangxi Prefecture increased slowly. After 2018, it showed a continuous upward trend. From 2020 to 2022, affected by the COVID-19 pandemic, the rural tourism industry experienced a significant decline (Fig. 1b). Table 1 shows the statistical data of the number of tourists and revenue in Xiangxi Prefecture over the past 20 years, obtained from the Statistical Yearbook of Xiangxi Prefecture (https://tjj.xxz.gov.cn/tjsj_187/tjnj/). In order to reveal whether the spatiotemporal evolution of the TVLCC is affected by rural tourism and the COVID-19 pandemic, the period from 2017 to 2022 was selected as the research time span. Five batches of traditional village lists published up to 2022 were chosen, comprising a total of 172 villages (Fig. 1c). The traditional village lists published after 2022 were excluded because of their low popularity and fewer rural tourism activities before 2022.

Table 1 The statistics of tourist numbers (unit: 10000 people) and revenue (unit: 100 million yuan) in Xiangxi Prefecture over the past 20 years
Full size table

Delineation methods for traditional village landscape boundaries in mountainous areas

Traditional villages are embedded in the natural environment of rural areas and do not have clear landscape boundaries. The farming radius is the maximum suitable distance for farmers to engage in cultivation activities45, affecting the spatial layout of the village landscape46. The EPLLs within the boundary range delineated based on the farming radius are the core components of traditional village landscapes, so they are involved in the carrying capacity evaluation in this study. However, the farming radius of traditional villages in mountainous areas is affected by the spatial resistance of complex mountain terrain, cultivation conditions, and agricultural production levels, resulting in irregular boundaries.

Currently, there is no standard method for delineating the landscape boundaries of traditional villages in mountainous regions. In this study, firstly, the ideal value of the farming radius and the comprehensive resistance surface that affects cultivation activities are calculated separately. Then, these two are spatially overlaid to determine the minimum cumulative farming cost at the end points of the farming radius, and the range of its contour lines is defined as the landscape boundary of traditional villages. The steps are as follows:

  1. (1)

    Calculation of the ideal values of the farming radius

    Limited by mountainous terrain and inconvenient transportation, traditional village production activities still rely mainly on conventional agricultural production modes, where population is an essential factor in determining labor supply. Therefore, the ideal values of the farming radius were calculated based on the relationship between population and per capita cultivated land47, ignoring the influence of various factors on farming activities under actual conditions. The formula is as follows.

    $$M\,\cdot \,P=K\pi {R}^{2}$$
    (1)

    where \(R\) is the ideal value of the farming radius; \(M\) is the per capita cultivated land; \(P\) is the population; and \(K\) is the wasteland reclamation index (percentage of cultivated land area to total village area), with a range of 0–1. Due to the significant differences in topography and scale of traditional villages in mountains areas, it is difficult to determine and unify the \(K\) value. Considering that cultivation costs need further consideration in this study, the K value is set to 1.

    The mean per capita cultivated farmland of each period is 724.45 m2, which are 793.3 m2, 686.7 m2, 746.7 m2, 786.7 m2, 653.3 m2, and 680 m2 in sequence respectively, from 2017 to 2022. The data are from the Statistical Yearbook of Xiangxi Prefecture. The population data of 172 traditional villages, ranging from 190 to 3279, were obtained through the official websites of Xiangxi Prefecture and its prefectural governments, the Digital Museum of Traditional Chinese Villages (http://www.dmctv.cn/), traditional village registration forms provided by government agencies (Xiangxi Prefecture Natural Resources and Planning Bureau, Jishou Municipal Bureau of Housing and Urban‒Rural Development) and on-site surveys. Finally, a dataset of the ideal values of the farming radius for the 172 traditional villages was established. When the farming radius of two or more traditional villages overlaps due to their proximity, they are extracted separately to participate in the calculation of the minimum cumulative cost (Fig. 2a). The ideal values of the farming radius of traditional villages in Xiangxi Prefecture are relatively small, ranging from 209 to 870 m.

  2. (2)

    Constructing the comprehensive resistance surface

    The geographic spatial resistance factors affecting cultivation activities are mainly terrain and land use factors46. In this study, considering the actual situation in Xiangxi Prefecture, the elevation, slope, aspect, and land use type indicators were selected and assigned resistance values according to the degree of resistance (Table 2). The larger the value is, the greater the resistance to cultivation activities. Among them, low altitudes, gentle slopes, and good sunlight conditions are most favorable for cultivation. The aspect is measured in a clockwise direction, with an angle ranging between 0 (north) and 360, and flat areas without a downward slope are assigned a value of −1. In terms of land use, snow and ice areas are not suitable for cultivation, so the resistance value is the largest; next are water areas that require detours and flooded vegetation, bare ground that is difficult to cultivate; certain agricultural planting cultivated under forest, so the resistance value of tree land is in the middle; built areas with a large labor force has a certain promoting effect on the expansion of the cultivation radius, so the resistance value is relatively small. The elevation, slope, and aspect data with a precision of 30 m were obtained from the Geospatial Data Cloud site, Computer Network Information Center, Chinese Academy of Sciences (http://www.gscloud.cn). The land use data are provided by the ESRI 10 m Annual Land Use Land Cover (2017–2022) (https://livingatlas.arcgis.com/landcoverexplorer/)48.

    Then, considering the slow change in topography and land use of the mountains areas, the resistance surface was calculated every other year. The study selected resistance indicator data from 2018, 2020, and 2022. The 2017 land use data are greatly affected by cloud cover, which significantly impacts the resistance results. Using MATLAB R2021a, the entropy weight method was used to calculate the objective weights of the indicators for each period, and the mean values were taken as the final weights (Table 2). Finally, the comprehensive resistance surfaces and their mean values for 2018, 2020, and 2022 were calculated.

  3. (3)

    Delineation of landscape boundaries

    Based on the mean value of the resistance surface, the cumulative cost of traditional village agricultural activities is calculated using the minimum cumulative resistance (MCR) model (Fig. 2b). Then, the minimum value of the cumulative cost at the endpoint of the ideal farming radius of each traditional village is taken as the ideal farming cost, and its contour line is used as the landscape boundary of the traditional village (Fig. 2c).

Fig. 2: Process of delineating the landscape boundaries of traditional villages.
figure 2

a Distribution map of the ideal farming radius; b Resistance surface and cumulative cost of agricultural activities in traditional villages; c Schematic diagram of the delineation of the landscape boundaries.

Full size image
Table 2 Grades, resistance values and weights of resistance indicators
Full size table

Selection of indicators of TVLCC

In mountainous areas, ecological landscapes play a role in maintaining ecosystem balance, providing environmental products and services, and improving the living environment. They are mainly composed of mountainous terrain, water systems, and forest ecological elements. The ecological landscape indicators include the woodland coverage index and the woodland fragmentation index, which have certain impacts on the structure, function, and ecological process of the landscape pattern. The topographic indicator is not selected because the change is not significant. In addition, the water index is not chosen. On one hand, it is because villages located on mountain slopes or mountaintops are a certain distance away from rivers, resulting in the water bodies of some villages being outside the scope of the landscape boundary. On the other hand, it is because the water body area of some villages is too small, causing the water body elements not to be identified in their land use data. Therefore, to ensure the comparability of the evaluation results of the TVLCC, the river factor was not selected uniformly in the end.

Production landscapes are dominated by farmland and supplemented by pasture. In these areas, villagers engage in productive activities, obtain various agricultural and livestock products, and develop featured agriculture and rural leisure tourism. The scale and shape of production landscapes are reflected through cultivated land coverage and fragmentation indicators. It should be noted that due to the limitations of the classification system of the annually available land use data (with a precision of 10 m), it is impossible to identify the intersection area between forestry and agriculture, as well as the agricultural landscapes of different cultivation types.

Living landscapes showcase the architectural style and ethnic customs of villages, providing essential venues for villagers’ daily living and social interactions. The living landscape indicators include building coverage and ethnic minority population density, which indirectly reflect village vitality and ethnic minority culture respectively. Unlike woodlands and cultivated land, buildings enhance tourism attraction through various morphological characteristics, such as scattered, ribbon, star, and grid shapes. Therefore, the building fragmentation index was not considered. Except for the fragmentation indicators, which are negative in direction, the other indicators are positive.

The carrying capacity indicators are shown in Table 3. The index values of WC, CLC and BC were obtained by calculating the area proportions of woodland, cultivated land and building land within the landscape boundaries of traditional villages using Arcmap10.8 software and land use data from 2017 to 2022. The index values of WC, CLC and BC are between 0 and 1. The larger the value is, the higher the coverage degree is. The moving window method of Fragstats was adopted to conduct patch density analysis on woodland and cultivated land. Then, the mean value within the landscape boundary is extracted as the value of the WF and CLF indicators. The larger the values of WF and CLF are, the more patches there are per unit area, and the more fragmented the landscape is. Here, determining the window threshold is a prerequisite. To reduce the number of computations, a 4000 m buffer zone is created with the traditional village as the center point. Then, moving windows of 100 m, 400 m, 600 m, and 800 m are selected for preliminary experiments. Among them, the 600 m moving window with moderate fluctuations in the index was selected, which not only retains the spatial information pattern but also reflects the landscape gradient characteristics.

Table 3 TVLCC indicators and their weights
Full size table

The EMPD data are obtained by multiplying the population density data by the percentage of ethnic minorities in each county and city and then extracting the mean density within the landscape boundary. The larger the value of EMPD is, the greater the population density of ethnic minorities per unit area is, and the more remarkable the cultural characteristics are. The population density data (100 m, 2017–2020) are from WorldPop (https://hub.worldpop.org/geodata/listing?id=29), and the data from 2021 and 2022 are obtained by multiplying the 2020 population density data by the population growth rate. The ethnic minority population proportion data are from the Statistical Yearbook of Xiangxi Prefecture.

Due to factors such as cloud cover and data accuracy, some indicators may show significant anomalies compared with the data of the previous and next years. To reduce errors, this study used the method of removing these anomalous values and replacing them with the average values of the indicator in other periods. At the same time, by combining the judgment of remote-sensing images, the difference value after adjusting the abnormal data is compensated to the actual land use types to ensure that the total area remains unchanged.

Carrying capacity evaluation method

Before conducting the carrying capacity evaluation, the indicator data are standardized using the min–max normalization method. The calculation formulas for positive and negative indicators are as follows.

$${\rm{Positive\; indicators}}:{x}_{{ij}}={(x}_{j}-{x}_{\min })/{(x}_{\max }-{x}_{\min })$$
(2)
$${\rm{Negative\; indicator}}:{x}_{{ij}}={(x}_{\max }-{x}_{j})/({x}_{\max }-{x}_{\min })$$
(3)

here \({x}_{j}\) is the value of the jth indicator; \({x}_{\min }\) and \({x}_{\max }\) represent the minimum and maximum values of the jth indicator, respectively; and \({x}_{{ij}}\) is the jth indicator value after standardization in the ith year.

The evaluation of TVLCC adopts the method of combining CRiteria Importance Through Intercriteria Correlation (CRITIC) and the technique for order of preference by similarity to ideal solution (TOPSIS) in SPSSAU online software49,50. The CRITIC method uses the volatility (contrast strength) and correlation (conflict) of data to calculate the weight and is suitable for stable data with a certain correlation between indicators. In this study, the carrying capacity indicators of EPLLs have a clear spatial distribution correlation within the traditional village landscape boundary. To facilitate data comparison, the mean values of the index weights in each period are used as the final weights of the evaluation system (Table 3). Then, the TOPSIS method, a commonly used and effective method in multiobjective decision-making analysis51, was adopted to calculate the carrying capacity level of EPLLs and the comprehensive landscape of traditional villages. This method has been widely applied in carrying capacity evaluation for a small number of indicators52,53. The formula is as follows.

$$Z=({Z}_{{ij}})m\times n,{Z}_{{ij}}={w}_{j}{x}_{{ij}}(i=1,2,3,...,{m;j}=1,2,3,...,n)$$
(4)
$${D}_{i}^{+}=\sqrt{\mathop{\sum }\limits_{j=1}^{n}{({Z}_{{ij}}-{Z}_{j}^{+})}^{2}}$$
(5)
$${D}_{i}^{-}=\sqrt{\mathop{\sum }\limits_{j=1}^{n}{({Z}_{{ij}}-{Z}_{j}^{-})}^{2}}$$
(6)
$${C}_{i}={D}_{i}^{-}/{(D}_{i}^{-}+{D}_{i}^{+})(i=1,2,3,...m)$$
(7)

where \(Z\) is the weighted normalized decision matrix; \({Z}_{{ij}}\) is the matrix element; \({w}_{j}\) is the weight of the jth indicator; \({x}_{{ij}}\) is the standardized value of the indicator; \({Z}_{j}^{+}\) and \({Z}_{j}^{-}\) are the positive (maximum) and negative (minimum) ideal solutions of the jth indicator; \({D}_{i}^{+}\) and \({D}_{i}^{-}\) are the Euclidean distances between the evaluation object and \({Z}_{j}^{+}\) and \({Z}_{j}^{-}\), respectively; and \({C}_{i}\) indicates the closeness between the carrying capacity of the ith year and the optimal carrying capacity, 0 ≤\({C}_{i}\) ≤ 1; the higher the score is, the better the bearing capacity level, and vice versa.

The mean values of closeness were divided into grades I, II, III, IV, and V at equal intervals from [0,1], corresponding to the carrying capacity level from low to high. Finally, the inverse distance weighting interpolation method was used to analyze the spatiotemporal evolutionary characteristics of the TVLCC and carrying capacity level of EPLLs at the regional scale. Additionally, Sankey diagrams were employed to analyze the transfer of the number of traditional villages among different numerical intervals of carrying capacity level of EPLLs from 2017 to 2022.

Analysis method of influencing factors on EPLLs carrying capacity

First, considering the impacts of natural environmental and socioeconomic factors in mountainous areas on the landscapes of traditional villages, indicators such as mean annual temperature (MAT), mean annual precipitation (MAP), relief amplitude (RA), road network density (RND), annual total tourist arrivals (ATTA), annual total tourism revenue (ATTR), and number of honorary titles (NHT) were selected. The data of MAP (unit: mm) and MAT (unit: °C) were sourced from the National Tibetan Plateau Scientific Data Center platform (https://data.tpdc.ac.cn/).The RA (unit: m) was calculated based on DEM data. The RND (unit: km/km²) was represented by the road network length per unit area of the village, and the data was sourced from the OpenStreetMap (https://www.openstreetmap.org). The data of ATTA (unit: 10,000 person) and ATTR (unit: 100 million yuan) of each county and city were sourced from the Xiangxi Statistical Yearbook from 2017 to 2022 and the statistical bulletins on the national economic and social development of each county and city. The NHT of each village indirectly reflected the investment intensity of the village, and its statistical list was sourced from the official websites of relevant national and provincial government departments.

Then, the random forest algorithm in machine learning was used to analyze the influencing factors of the EPLLs carrying capacity. This method obtained the feature importance and contribution degree of each indicator by calculating the reduction of Gini Impurity. Considering the distribution characteristics of the indicator data and the sample balance requirements of the classification model, the carrying capacity of ecological landscapes (0.2 and 0.5 quantiles) and the carrying capacity of production and living landscapes (0.2 and 0.4 quantiles) were divided into three categories to construct the data labels. The influencing factor indexes were used as independent variables, and the classification results of ecological, living, and production carrying capacities were used as dependent variables respectively. The six-year data was divided into a training set and a test set at a ratio of 7:3 to calculate the contribution degree of each factor. Finally, the accuracy, weighted precision, weighted recall, and weighted F1-score were used to conduct a model accuracy test. The higher the values, the higher the model accuracy.

Results

Distribution of TVLCC grades

As shown in Table 4, the majority of woodland coverage of traditional villages is at a medium-high level, while the overall coverages of cultivated land and building are at a low level. The fragmentation degrees of woodland and cultivated land are mainly at the medium level, while the ethnic minority population density is low.

Table 4 Range of mean values from 2017 to 2022 and the number of villages for each indicator
Full size table

The overall level of TVLCC is relatively low to medium. The annual values are distributed in grades I, II, and III (Table 5). Shawan Village in Baojing County (Fig. 3a), the only grade I traditional village with mean closeness value of 0.156, is adjacent to the Youshui River. It is surrounded by water on three sides and steep hillsides on one side, making land transportation very inconvenient. The mean values of WC, WF, CLC, CLF, and BC are 0.266, 6.507, 0.012, 7.841, and 0.111, respectively. This indicates that the coverage of woodland, cultivated land and buildings in Shawan village is low, and the degree of fragmentation is large. In addition, the average population density of ethnic minorities, mainly Tujia, is 0.539. The very low landscape carrying capacity of Shawan village is primarily due to the limited and uneven distribution of ecological and agricultural resources, which restricts ecological functions and reduces agrarian productivity.

Fig. 3: Photos of representative traditional village landscapes.
figure 3

a Shawan Village; b Shibadong Village; c Hangsha Village; d Tizi Village; e Azhahe Village; f Dehang Village; g Zhonghuang Village. (LS Longshan County, YS Yongshun County, BJ Baojing County, GZ Guzhang County, HY Huayuan County, FH Fenghuang County, LX Luxi County, JS Jishou City)

Full size image
Table 5 Statistical table of grade and number of traditional villages
Full size table

The number of grade II traditional villages is the greatest. From 2017 to 2022, most of the villages were grade II, such as Shibadong Village in Huayuan County (Fig. 3b), Hangsha Village (Fig. 3c), Tizi Village (Fig. 3d), Azahe Village in Baojing County (Fig. 3e), and Laosicheng Village in Yongshun County, whose mean closeness values are 0.33, 0.335, 0.346, 0.331, and 0.358, respectively. Among them, Shibadong Village, the pioneer of China’s targeted poverty alleviation, is mainly inhabited by the Miao ethnic group and features mountainous and gorge terrain. The mean values of WC, WF, CLC, CLF, BC, and EMPD are 0.680, 5.290, 0.053, 5.983, 0.267, and 0.809, respectively. This shows that the high woodland coverage, small-scale and scattered cultivated land limit agricultural production, and the low building coverage limits the development of social activities.

The number of grade III traditional villages is relatively small, reaching a maximum of 18 in 2019 and showing a significant decline in 2020. After that, there was a recovery in 2022. Dehang Village (Fig. 3g) in Jishou city, which is located in the core area of the Dehang Scenic Spot (national level), had a closeness value of 0.415 in 2022. It has favorable conditions for tourism development. Surrounded by mountains and traversed by rivers, the village is abundant in animal and plant resources. The mean values of WC, WF, CLC, CLF, BC, and EMPD are 0.762, 3.095, 0.096, 2.854, 0.143, and 1.196, respectively. This indicates that the woodland coverage is high and the fragmentation is low, forming a good ecological landscape. Due to its low coverage, the cultivated land is not suitable for large-scale agricultural production. The low building coverage restricts the development of tourist activities. The population density of ethnic minorities is at a medium level, and the Miao population is relatively concentrated. In addition, there are Zhonghuang Village (Fig. 3f) in Jishou City, Pachuke Village in Yongshun County, Modaoyan Village in Guzhang County, and so on.

Spatiotemporal evolutionary characteristics of TVLCC

Figure 4 shows that: 1) In 2017, areas with a closeness value between 0.35 and 0.4 extended through the eastern, western, central, and northern parts of the whole region. The village with the highest value was located in Luxi County in the southeast part of Xiangxi Prefecture. 2) From 2018 to 2020, the coherence of areas with a closeness value between 0.35 and 0.4 in the eastern and western parts weakened, and a zonal distribution pattern was formed in the northern and southern parts. The overall carrying capacity level decreased in the north but increased in the south. 3) In 2021, the overall carrying capacity level of the central region decreased, and the closeness value dropped from between 0.35 and 0.4 to between 0.3 and 0.35. 4) Compared with 2021, in 2022, the carrying capacity level of the whole region increased to a certain extent. However, areas with a closeness value between 0.35 and 0.4 still lacked spatial continuity in both the eastern-western parts and northern-southern parts. In addition, an obvious agglomeration area formed in the southeastern part, with a closeness value between 0.4 and 0.5.

Fig. 4: Spatiotemporal evolution of TVLCC.
figure 4

Distribution maps of the grades of traditional villages and landscape carrying capacity levels in Xiangxi Prefecture from 2017 to 2022.

Full size image

Spatiotemporal evolutionary characteristics of the carrying capacity level of EPLLs

Figure 5a shows that: 1) From 2017 to 2022, the ecological landscape carrying capacity of traditional villages was relatively high compared with that of living and production landscapes. 2) The carrying capacity showed an increasing trend each year from 2017 to 2019 and formed a north-south axis pattern. The distribution range of high values greater than or equal to 0.7 gradually increased. 3) In 2020, the distribution range of medium-high values greater than or equal to 0.3 decreased sharply. However, from 2021 to 2022, the overall carrying capacity level was significantly improved again.

Fig. 5: Spatiotemporal distribution of the EPLLs carrying capacity levels of traditional villages from 2017 to 2022.
figure 5

a Ecological landscape carrying capacity level; b Production landscape carrying capacity level; c Living landscape carrying capacity level.

Full size image

Figure 5b shows that: 1) The carrying capacity of traditional village production landscapes was at a low level from 2017 to 2022, but it was relatively higher in 2017. 2) From 2018 to 2022, a continuous low-value distribution area was formed in the north-south direction across the whole region. 3) Areas with medium-high values of 0.3 or above were rare and were mainly distributed at the border of Xiangxi Prefecture.

Figure 5c indicates that: 1) From 2017 to 2022, the carrying capacity level of traditional village living landscapes was medium to low but showed an increasing trend year by year. 2) The carrying capacity changed slightly from 2018 to 2021 but increased significantly in 2022. 3) The carrying capacity levels of the living landscapes of traditional villages in the central-western and southern parts of Xiangxi Prefecture were relatively good and at a medium level.

By analyzing the annual transfer of the number of traditional villages in different numerical ranges of the carrying capacity of EPLLs from 2017 to 2022, the following can be obtained: 1) Regarding the ecological landscape carrying capacity, the number of traditional villages with increasing numerical ranges yearly is 29, 14, 1, 111, 17 in sequence, and those with decreasing ranges is 21, 19, 120, 2, 15, respectively (Fig. 6a). Notably, the ecological landscape carrying capacity of traditional villages declined most sharply in 2020, by 69.8%, and rose most significantly in 2021, by 64.5%. 2) For the production landscape carrying capacity, the number of villages with ascending numerical ranges annually is 15, 9, 24, 20, 15, while those with descending ranges is 42, 16, 12, 9, 31 (Fig. 6b). Overall, it shows a slow decline. 3) Concerning the living landscape carrying capacity, the number of villages with increasing numerical ranges year on year is 24, 20, 15, 18, 34, and those with decreasing ranges is 9, 9, 13, 7, 3 (Fig. 6c). Overall, it demonstrates a significant upward trend.

Fig. 6: Sankey diagrams and the contribution degree of influencing factor indicators.
figure 6

a Annual transfer in ecological landscape carrying capacity level; b Annual transfer in production landscape carrying capacity level; c Annual transfer in living landscape carrying capacity level; d Contribution degree of influencing factor indicators.

Full size image

Main influencing factors of the EPLLs carrying capacity

Regarding the changes in influencing factors from 2017 to 2022, the values of RA and MAT remained relatively stable. Meanwhile, RND and NHT values generally trended upward, yet the trends of ATTA and ATTR values differed among various counties and cities. Notably, the MAP value decreased from 2017 to 2019. The highest MAP value dropped from 1646.7 mm to 1352.3 mm, and the lowest decreased from 1317.1 mm to 1180.1 mm. In 2020, the MAT value was significantly higher than in other years. It peaked at 1726 mm, an increase of 373.7 mm compared to 2019 and 196.5 mm compared to 2021. From 2021 to 2022, the MAT value continued its downward trend.

By calculating the accuracy, weighted precision, weighted recall, and weighted F1-score of the test set of the random forest (Table 6), it can be seen that the overall accuracy evaluation results of this model are moderately above average. Considering that there is no extremely high requirement for precision in this study and it is restricted by the relatively small number of village samples, such accuracy evaluation results are acceptable.

Table 6 Accuracy evaluation results of random forest
Full size table

According to the factor contribution degree results obtained from the random forest (Fig. 6d), the following insights are derived: 1) In terms of the ecological landscape carrying capacity, RA demonstrates the highest contribution degree, followed by MAP and MAT, while the significance of RND, ATTA, ATTR, and NHT is comparatively low. 2) Regarding the production landscape carrying capacity, RA’s contribution degree is significantly higher than that of other factors, with RND following. Additionally, the contribution degrees of ATTA, MAP, and MAT all exceed 10%, while those of ATTR and NHT are the lowest. 3) As for the living landscape carrying capacity, RA, MAT, and RND exhibit relatively high contribution degrees, and the contribution degrees of ATTA and MAP both surpass 10%, whereas the contribution degrees of ATTR and NHT are relatively low.

Discussion

Among the influencing factors of carrying capacity, mountainous terrain is the most important factor affecting the carrying capacity of EPLLs in traditional villages, and traffic conditions significantly impact the carrying capacity of production and living landscapes. This is specifically reflected in the following aspects:

  1. 1.

    The carrying capacity of ecological landscapes in traditional villages of Xiangxi Prefecture is predominantly dominated by natural factors such as topography and climate, and is less affected by socioeconomic factors. This indicates that current human activities, such as tourism development, have not severely damaged the ecological environment yet. In 2020, the significant increase in the MAP value compared with other years affected the stability of the natural environment and was the main cause of the sharp decline in the carrying capacity of ecological landscapes.

  2. 2.

    High relief geomorphic features and low connectivity traffic conditions play a dominant role in shaping the carrying capacity of production landscapes by influencing agricultural production, transportation, etc. Although in recent years, rural vitalization efforts have increased the construction of transportation road networks, the improvement in the carrying capacity of production landscapes has not been significant. Moreover, rural tourism activities and the impact of precipitation and temperature on the production process also come into play.

  3. 3.

    The carrying capacity of living landscapes is mainly determined by mountainous landforms, temperature, and traffic conditions. Among them, the year by year increase in its carrying capacity benefits from the improvement of rural traffic conditions. Tourism activities and precipitation conditions also have some influence on villagers’ lives.

  4. 4.

    Tourism revenue and honorary titles have the least impact on the carrying capacity of EPLLs. Perhaps it is because tourism revenue has not been effectively invested in village landscape construction, and honorary titles have not been effectively transformed into substantial improvements in village landscapes.

In this study, the annual influencing factors were not analyzed due to the small sample size of villages, which led to low - accuracy analysis results. Additionally, the analysis of influencing factors for comprehensive landscape carrying capacity was not included because of its insignificant spatiotemporal changes.

The evolution of the carrying capacity of EPLLs reflects its correlation with rural tourism development and the policies for the protection and utilization of traditional villages. This is specifically manifested as follows:

  1. 1.

    The evolution of the ecological landscape carrying capacity of traditional villages is similar to the overall development trend of the tourism industry in Xiangxi Prefecture. The carrying capacity reached its optimal state in 2019. In the same year, the “Regulations on the Protection of Traditional Villages in Xiangxi Prefecture” were officially implemented, which put forward protection requirements for the ecological environment and overall style.

  2. 2.

    The evolution of the production landscape carrying capacity in Xiangxi Prefecture is opposite to the overall tourism development trend. In November 2020, the “Implementation Plan for the Protection and Utilization of Traditional Villages in Xiangxi Prefecture” was issued. The plan proposed to focus on protecting basic farmland and constructing irrigation facilities. Under its influence, the production landscape carrying capacity slightly increased in 2021, but the short term improvement was not obvious.

  3. 3.

    The evolution of the living landscape carrying capacity shows a slow upward trend. This trend coincides with the implementation period of the “Three-year Action Plan for the Protection and Renovation of Characteristic Residential Buildings and Traditional Scenery (2018–2020)” issued by Xiangxi Prefecture. This plan gives priority to the protection and renovation of traditional dwellings to prevent excessive demolition and reconstruction. It also guides and regulates new residential buildings and municipal public infrastructure projects in terms of style, color, and details to match the traditional style.

This study can provide guidance for the formulation of policies on the protection and utilization of traditional village landscapes in mountainous areas, as well as for tourism development planning at both the regional and village levels. At the regional scale, targeted landscape protection plans can be formulated based on the distribution of the TVLCC, as well as the spatiotemporal evolutionary characteristics and main influencing factors of the carrying capacity of EPLLs. Special attention should be paid to the impact of natural disasters such as floods, landslides, and debris flows caused by continuous heavy rainfall on the ecological landscape carrying capacity. Moreover, tourism revenue and government investment need to be better utilized for the protection and maintenance of traditional village landscapes. At the village scale, Grade III villages can serve as key villages for rural tourism, emphasizing both landscape protection and tourism development to drive the development of surrounding villages. In Grade II villages, landscape protection needs to be strengthened before developing tourism. Moreover, tourism routes should be built to strengthen the linkages between villages. Grade I Shawan Village is vulnerable to damage from tourism. Therefore, tourism development can be carried out only after the landscape carrying capacity level has significantly improved, and strict landscape protection strategies should be formulated. Furthermore, the synergy between the landscape protection and tourism development of traditional villages can be promoted through digital technologies and community participation methods, such as intelligently monitoring the dynamic changes of the landscape54,55, optimizing the forms of tourist participation experience56, and leveraging the strength of the community to carry out landscape protection actions57,58.

The farming radius is a commonly used method in the study of village spatial scope and has been applied in the regrouping45 and spatial optimization46 of villages in flat terrain areas, as well as in the study of spatial distribution59, operations scale of business entities60, and farmland marginalization61 of villages located in hilly and mountainous areas. However, there are still limitations in the research on landscape boundary delineation for a large number of villages in mountainous areas. Due to the complex terrain, the landscape boundaries of each village in mountainous areas are unique, which is not conducive to the quantitative evaluation of landscapes at the regional scale. The method used in this paper can simultaneously quantify the landscape boundaries of villages. However, due to the large number of traditional villages, it is difficult to obtain an accurate average K value. Considering that setting the K value to 1 will overestimate the actual cultivated area and narrow the actual landscape boundary. During the research process, we have confirmed by checking the remote-sensing images that the EPLLs of traditional villages are basically within the landscape boundary range, which does not affect the analysis of carrying capacity. This may be because the highly undulating mountains limit the actual landscape boundary range to a relatively small level. Therefore, the landscape boundary delineation method in this study is applicable to Xiangxi Prefecture and similar mountainous areas. However, it is worth noting that traditional village landscape boundaries are dynamic and influenced by population changes and resistance factors.

Existing studies on the landscape carrying capacity of human settlements have mainly focused on the urban scale36,37,62,63,64,65. When it comes to the study of village carrying capacity, it encompasses social38,66, human67, and ecological aspects68. Among these, land use is the primary factor influencing village carrying capacity36,62,64. Considering the land use composition, along with the ease of data collection and quantification, this study selected TVLCC indicators related to landscape area proportion, shape, and population density. Given the small scale and simple composition of traditional village landscapes, landscape functions were not considered during the indicator selection process. In the next step of work, for specific traditional village cases, more detailed field investigations are needed on land use and village population data. These surveys aim to clarify the scope of various types of land use and determine more accurate K values, enabling more precise adjustment of the landscape boundaries of each village and the acquisition of more accurate evaluation results of landscape carrying capacity.