Introduction

Agricultural heritage systems, remnants of sustainable agricultural practices from the era of agricultural civilization, continue to offer food security, livelihood resources, and services to local communities [1]. They exemplify ecological resilience and sustainability [2]. In 2002, FAO initiated a project to manage and protect Globally Important Agricultural Heritage Systems (GIAHS). In order to ensure the protection and sustainable management of unique traditional agricultural systems and the landscapes, knowledge systems, cultures, and biodiversity that they are connected with, this effort seeks to obtain worldwide recognition for these systems [3]. China, being one of the first nations to respond to the FAO’s call, received approval for its initial GIAHS in 2005 and instituted an evaluation and protection system for CNIAHS [4]. As of April 2024, China boasts 22 GIAHS sites, the highest number globally, and has released seven batches totaling 188 CNIAHS. The identification and preservation of agricultural heritage systems have ensured the utilization and inheritance of sustainable agricultural ecological wisdom, reinforced China’s poverty alleviation successes, and facilitated the enduring progress of rural revitalization. These endeavors have also diversified the methods of GIAHS protection and offered valuable references for the identification and preservation of GIAHS globally.

Agricultural heritage sites encapsulate the agricultural history and ecological wisdom characterized by distinct regional attributes. As an important part of national land, CNIAHS witnesses the development and succession of agricultural history and carries the regional context and connotation. It’s a spatial system composed of natural environment, historical process, cultural accumulation and national/local cognition, which has the attributes of all-region, all-time, holistic and comprehensive. Based on relevant FAO documents and the attributes of CNIAHS types [5], this study classifies CNIAHS into 11 categories: (1) high-value crop and spice systems, encompassing the cultivation of tea, fruit trees, vegetables, and spices, such as the Fuding white tea culture system in Fujian, the ancient pear orchard in Zhao County, Hebei, and the traditional aquatic vegetable cultivation system in Wuzhong, Jiangsu; (2) multiple cropping/polyculture farming systems, including various agro-forestry-fishery-animal husbandry mixed systems and crop rotation systems, exemplified by the rice-fish co-culture system in Qingtian, Zhejiang, and the paddy-upland rotation system in Nanxiong, Guangdong; (3) food crop systems, covering common food crop production systems in both paddy fields and dry farming systems, such as the dryland farming system in Fumeng, Liaoning; (4) understory farming systems, which fully utilize the space under forest canopies for agricultural development, such as the Panax notoginseng cultivation system in Wenshan, Yunnan; (5) forestry systems, featuring various characteristic seedling, flower, and economic forest production systems, such as the ancient osmanthus tree group in Xianning, Hubei; (6) nomadic and semi-nomadic pastoral systems, represented by heritage sites such as the alpine nomadic system in Damxung, Tibet; (7) below sea level systems, including delta or wetland cultivated areas characterized by soil and water management techniques, and freshwater fishery systems, such as the mountain spring stream fish farming system in Xiuning, Anhui, and the dike-pond agricultural system in Wujiang, Jiangsu; (8) animal husbandry systems, referring to agricultural heritage systems involving the artificial breeding of livestock, poultry, and silkworms, excluding nomadic systems, such as the Rongchang pig breeding system in Chongqing and the sericulture and mulberry system in Tongxiang, Zhejiang; (9) ancient irrigation, soil, and water management systems, pertaining to areas where agricultural production is challenged by drought or other factors, but effective agricultural production systems have been developed through water conservancy facilities. Examples include the dryland stone terrace system in Shexian, Hebei, and the Karez agricultural system in Turpan, Xinjiang; (10) mountain rice terrace agroeco systems, which are comprehensive agricultural systems based on rice cultivation that exploit mountainous terrains, such as the Hani rice terraces system in Honghe Yunnan; and (11) complex multi-layered home gardens, which are intricate multi-layered family gardens with courtyards as the basic production unit, such as the traditional urban vineyard in Xuanhua, Hebei.

Since the launch of the GIAHS project in 2002, research on agricultural heritage has progressively expanded, spanning disciplines such as sociology, ecology, economics, and agriculture. The research can be categorized into three principal domains: (1) investigations into dynamic conservation strategies for agricultural heritage, including studies on conservation measures [6] and policies adopted by various countries [7, 8]; (2) analyses of sustainable development models for agricultural heritage sites [9], encompassing research on utilization patterns [10], tourism development [11, 12], and livelihoods of local residents [13, 14]; and (3) evaluations of heritage value, focusing on ecosystem services [15] and landscape significance of heritage sites [16, 17]. Initially, the research concentrated on specific heritage sites at a micro level [18], but it has evolved to encompass regional and national scales [19, 20]. Methodologies such as bibliometric analysis [21] and geospatial analysis [19] have been employed.

The past one-site based protect method neglects the numerous atypical or unrecognized historical and cultural relics, and the heritage surroundings, resulting in the isolation of the cultural landscape and the fragmentation of heritage space. However, those ignoring parts play an irreplaceable role in preserving heritage characteristics, and maintaining the continuity and integrity of regional cultural experience. Nowadays, with the deepening of the research on the construction of heritage protection systems, the main body of heritage protection has gradually changed from a single site to a regional heritage corridor [22], a holistic heritage space, or a regional heritage protection area [23, 24]. In 2021, the Chinese government issued the “The advice on Facilitating Cultural and Historical Protection and Inheritance in Urban–rural Construction”[25], which explicitly proposed the construction of an all-in-one historical and cultural protection and inheritance system, promoting the construction of China's historical and cultural heritage system effectively. As an essential part of Chinese historical and cultural heritage system, the construction of CNIAHS system has attracted increasing academic attention. The research perspective has shifted from individual protection to regional and holistic protection, which not only enhances the connectivity of heritage resources in spatial dimensions, but also maintains the inherent cultural connections and inheritance of heritage resources in temporal dimensions.

Based on the background above, current research on agricultural heritage systems has shifted from individual heritage sites or specific regional sites to a broader, macro-level examination of national or even global contexts. However, existing macro-level studies mainly address synchronic spatial distribution and its underlying factors, with a notable deficiency in diachronic research on spatio-temporal pattern changes, which directly leads to the lack of research on the time dimension and cultural correlation of agricultural cultural heritage protection, and causes a negative impact on the following excavation of agricultural cultural heritage and the construction of regional heritage protection system. Furthermore, spatial entropy—a metric for measuring system randomness, complexity, and disorder—is widely used in fields like biology and meteorology [26, 27] but remains underutilized in heritage spatial studies. For CNIAHS has varied types and a complicated formation process, studying the law of entropy variation of CNIAHS spatio-temporal pattern, can help to explore the stage and regional development trends at the multiple dimensions of type, pattern and time. By revealing the spatio-temporal pattern characteristics of identified heritage sites, the excavation of CNIAHS can be further explored and enriched, and the construction of CNIAHS protection system can be solidly founded. Hence, this study centers on China, a key cradle of agricultural civilization, to explore the evolution of spatio-temporal distribution patterns of heritage sites, hoping that will provide a reference for the future construction of heritage protection system [28]. Employing methods including the spatial information entropy, hotspot analysis, kernel density estimation, and nearest neighbor index (NNI), this paper aims at: (1) analyzing the spatio-temporal patterns of CNIAHS systems; (2) uncover the developmental trajectory of these systems; and (3) provide valuable insights for the exploration, spatial pattern optimization, and preservation of agricultural heritage systems both in China and globally[20].

Materials and methodologies

Data sources

Data pertaining to agricultural heritage were obtained and consolidated from the official website of the China Agricultural Museum (www.ciae.com.cn), the Ministry of Agriculture and Rural Affairs of the People’s Republic of China (http://www.moa.gov.cn/), and relevant governmental information disclosures. Additionally, the watershed division and water system data utilized in this research were obtained from the Chinese Academy of Sciences' Resource and Environment Science and Data Center. (https://www.resdc.cn/).

Methodologies

Spatial information entropy

Spatial information entropy reflects the overarching characteristics of multi-element layout differentiation and serves as an instrument for measuring system complexity, disorder, and randomness [26]. It quantifies the level of imbalance in the structure of geographical features and characterizes the prevailing spatial homogeneity or heterogeneity [29]. For the spatial pattern of CNIAHS, based on the principles of information compression in discrete spatial fields, spatial information entropy can be employed to gauge the complexity of heritage site spatial patterns and their variations across different historical periods.

$$\begin{aligned} {\text{A}}\left( {\text{x}} \right) = - \sum\nolimits_{{i = 0}}^{n} P \left( {x_{i} } \right)log_{2} \left[ {{\text{P}}\left( {x_{i} } \right)} \right] & \\ P\left( {x_{i} } \right) = \frac{{k_{i} }}{N}{\text{,}}\,{\mkern 1mu} {\text{i}} = 1,2, \ldots ,{\text{N}} \\ \end{aligned}$$
(1)

where A(x) is the spatial information entropy of CNIAHS, P(xi) is the probability value of increase for each type of cultural space, ki is the frequency of increase for a certain type of heritage site during a specific period, and N is the total number of CNIAHS during that period. Utilizing Shannon’s spatial information entropy theory, this study measures the spatial information entropy of agricultural heritage sites across various periods. The complexity of different types of heritage sites is employed to characterize changes in spatial patterns. A higher entropy value signifies a more complex spatial pattern of heritage sites during that period. An increase in entropy denotes enrichment of heritage types in the space, while a decrease in entropy signifies no increase in heritage types.

Nearest neighbor index (NNI)

NNI can be used to uncover the spatial distribution pattern of point features [30],

$$\text{NNI}=\overline{{r }_{i}}/\overline{{r }_{e}}$$
(2)
$$\overline{{r }_{e}}=\frac{1}{\sqrt[2]{D}}=\frac{1}{\sqrt[2]{m/A}}$$
(3)

where NNI is the nearest neighbour index, ri is the average observed nearest neighbour distance, and re is the theoretical nearest neighbour distance. A is the area of the study region, m is the number of CNIAHS, and D is the point density. When the NNI = 1, it suggests that the point features tend to be randomly distributed, indicating a random pattern. A NNI < 1 indicates a tendency towards clustered distribution, representing a clustered pattern. Conversely, a NNI value > 1 suggests a tendency towards uniform distribution, indicating a uniform pattern.

Spatial hotspot analysis

The cold and hot spots analysis is capable of identifying areas where similar values cluster, facilitating the examination of structural characteristics in spatial distribution. The results of the analysis categorize the spatial pattern into hot spots, which are areas of high-value concentration, and cold spots, which are areas of low-value concentration. This method enables the identification of regions where agricultural heritage sites are concentrated.

$${\text{Gi}} = {{\sum\limits_{{i = 1}}^{n} {W_{{ij}} } a_{j} } \mathord{\left/ {\vphantom {{\sum\limits_{{i = 1}}^{n} {W_{{ij}} } a_{j} } {\sum\limits_{{i = 1}}^{n} {a_{i} } }}} \right. \kern-\nulldelimiterspace} {\sum\limits_{{i = 1}}^{n} {a_{i} } }}$$
(4)
$${a}_{i}={S}_{ir}/{\overline{S} }_{ir}\times 100\%$$
(5)
$$K=\sum_{j=1}^{n}\left({K}_{j}\times {W}_{j}\right)$$
(6)

where ai and aj are the indices of CNIAHS distribution in regions i and j, respectively; Wij is the spatial weight matrix; Gi is the score of the hot and cold spot index; S ir is the mean value of the actual number of CNIAHS in region i; K is the total score of the evaluation unit; Kj is the total score of the jth factor; and Wj is the weight value of the jth factor.

Kernel density estimation

The kernel density estimation can reflect the distribution density and clustering state of agricultural heritage sites across the country.

$$\lambda \left( x \right){ = }\frac{1}{mh}\mathop \sum \limits_{i = 1}^{n} k\left( {\frac{{x - x_{i} }}{h}} \right)$$
(7)

where k is the kernel function; x is the location of the target point of the agricultural heritage site; xi is the location of the ith important agricultural heritage within the spatial range centered at x with a radius of h; and m is the number of important agricultural heritage sites. The value of h in the formula affects the smoothness of the kernel density estimation results.

Results

Temporal distribution characteristics of CNIAHS

Considering the historical evolution of agricultural technology in China [31], the temporal distribution of agricultural heritage sites can be segmented into six distinct stages (Table 1). The table reveals that a substantial number of heritage sites originated during the stage of the Qin and Han dynasties, and the Yuan dynasty, attributed to their prolonged duration. Similarly, the Ming and Qing dynasties, being more recent in historical context, have also contributed a significant number of preserved heritage sites. In contrast, the Stone Age, extending up to the Spring and Autumn and Warring States periods, has resulted in a comparatively smaller number of preserved agricultural heritage sites.

Table 1 The NNI of CNIAHS in Different Stages
Full size table

Spatial pattern characteristics of CNIAHS in different periods

The emergence and development of CNIHAS are intrinsically linked to the historical progression of agricultural technology, regional demographic shifts, and climatic conditions. Broadly, the identified agricultural heritage sites display a markedly clustered distribution nationwide, with prominent hotspots in the Pearl River and Yangtze River Delta regions (Fig. 1a). This pattern radiates from these core areas to adjacent regions, forming a spatial configuration centered on the Yangtze River and Pearl River Delta, with the Beijing-Tianjin-Hebei region and Hunan-Chongqing area serving as secondary hubs (Fig. 1b). Analyzing the temporal evolution of this spatial pattern (Table 1), the hotspots have transitioned from the southwestern region through the central area to eastern China, establishing a distinct hotspot in the east before extending towards northern China (Fig. 2). Among the six developmental stages, the periods from the Qin-Han to Northern and Southern Dynasties and from the Sui-Tang to Song-Yuan Dynasties not only feature a high number of heritage sites but also exhibit pronounced clustering in their spatial distribution. This clustering is especially evident during the Qin-Han to Northern and Southern Dynasties. The other four stages present a more uniform distribution of heritage sites (Table 1).

Fig. 1
figure 1

Overall Spatial analysis of CNIHAS. a Getis-Ord Gi of CNIHAS. b Kernel density of CNIHAS

Full size image
Fig. 2
figure 2

Analysis of spatial–temporal patterns of CNIAHS. a Paleolithic Age and Neolithic Age. b Xia, Shang and Zhou. c Spring and Autumn Period and Warring States Period. d Qin and Han to the Southern and Northern Dynasties. e Sui, Tang, Song and Yuan Dynasties. f Ming and Qing Dynasties

Full size image

Regions with plentiful water resources generally foster better agricultural development. Hence, this study further investigates the spatio-temporal pattern changes of agricultural heritage sites across China’s nine major river basins (Fig. 3). During the Stone Age, CNIHAS were predominantly located in the Yangtze River basin, Huai River basin, and Songhua-Liao River basin (Fig. 3a). China’s largest by water volume, the Yangtze River basin, maintained its heritage site density advantage until the Ming and Qing dynasties. The Yellow River and Haihe River basins demonstrated a more significant spatial distribution advantage in the Spring and Autumn and Warring States Period (Fig. 3b). The Southeast River basin, near the Yangtze River estuary, also exhibited certain spatial layout advantages due to its proximity. Overall, exorheic basins demonstrated a clear advantage in the spatial distribution of CNIHAS sites. Over time, the spatial distribution of heritage sites shifted from central and eastern basin areas towards the southeastern basin regions.

Fig. 3
figure 3

Analysis of CNIAHS’ basin spatial–temporal patterns. a Paleolithic Age and Neolithic Age. b Xia, Shang and Zhou. c Spring and Autumn Period and Warring States Period. d Qin and Han to the Southern and Northern Dynasties. e Sui, Tang, Song and Yuan Dynasties. f Ming and Qing Dynasties

Full size image

Analysis of spatial pattern changes in CNIAHS across different periods

A notable increase in the number of CNIAHSs was observed during the Qin-Han and Song-Yuan periods (Table 1). The substantial quantity of CNIAHSs established during these two eras laid the foundation for the current spatial pattern, wherein heritage sites are predominantly concentrated in the southeastern region of the country.

Agricultural heritage sites from the Stone Age are primarily concentrated in the Yangtze River Basin, encompassing provinces such as Yunnan, Guizhou, Sichuan, Hunan, and Jiangxi (Fig.3a). Additionally, scattered sites can be found in Shandong Province within the Yellow River Basin, as well as in regions north of the Great Wall, including Inner Mongolia, Liaoning, and Heilongjiang. These areas are recognized as key regions where primitive agriculture originated in China.

During the Xia, Shang, and Zhou dynasties, the middle and lower portions of the Yellow River were the center of China's political, economic, and cultural activity. The loess soil in this region was loose and easily cultivated, which facilitated the spatial shift of agricultural heritage sites towards the Yellow River Basin. These sites were predominantly distributed in the upper and middle parts of the Yellow River, specifically in Shaanxi, Henan, and Shanxi provinces (Fig. 3b). Notably, these provinces also encompassed the areas where the capital cities of the Xia, Shang, and Zhou Dynasties were located.

During the Spring and Autumn and Warring States periods, the distribution of agricultural heritage sites shifted towards the middle and lower regions of the Yellow River and Yangtze River regions, including areas such as Henan, Hebei, Anhui, and Jiangsu, as well as the Pearl River Delta (Fig. 3c). In this stage, the population in the Yellow River Basin increased rapidly. Additionally, some capital cities of feudal states in the Yangtze River basin, such as Shouchun (present-day Shouxian in Anhui) of the Chu State and Wu (present-day Suzhou) of the Wu State, also became prosperous areas with concentrated populations. The population growth in these regions further promoted the development of agriculture in these areas.

From the Qin and Han dynasties to the Northern and Southern dynasties, the spatial distribution advantage of the Yellow River Basin became more pronounced, with a significant increase in areas of high distribution density (Fig. 3d). Numerous heritage sites were distributed in provinces such as Gansu, Ningxia, and Henan within the Yellow River Basin. Additionally, provinces south of the Yangtze River, including Zhejiang and Jiangxi, also exhibited relatively dense distributions. Following the unification of the six states by Qin, the Yellow River Basin became the national economic center. Economic and production measures, such as reducing taxes and levies, improving water conservancy, encouraging agricultural labor, and cultivating frontier lands, led to China’s first historical peak in agricultural production. The southward migration of populations during the Eastern Han and Western Jin dynasties resulted in a substantial increase in population density south of the Yangtze River. The population density in southern Zhejiang and Fujian also increased, leading to growth in food demand and agricultural development. Apart from these regions, most of the southern areas during this period remained sparsely populated.

During the Sui-Tang, Song, and Yuan dynasties, the concentrated distribution areas of agricultural heritage sites shifted more prominently towards the southeast compared to the previous period. The spatial agglomeration advantage became more evident in the Yangtze River Delta area, comprising Anhui, Jiangsu, and Zhejiang, as well as Fujian and Jiangxi provinces (Fig. 3e). Following the An Lushan Rebellion, the national economic center gravitated towards the south of the Yangtze River, resulting in a higher population density in this area. Gradually, a demographic pattern emerged with a higher population in the south and a lower population in the north of China. Consequently, land use conflicts in the southern regions became increasingly prominent.

In the Ming and Qing dynasties, the distribution of agricultural heritage sites was relatively dispersed (Fig. 3f). In the north, the concentrated distribution areas were primarily in Beijing and Tianjin within the Haihe River Basin, as well as Jilin and Liaoning in the Songhua-Liaohe River Basin. In the southern regions, the concentrated distribution areas shifted towards the Pearl River Basin. After the Ming Dynasty, as the population continued to increase, the development of the Yangtze River Basin, with its favorable natural conditions, had already been completed. Consequently, agricultural development expanded towards more remote areas in the north and south, resulting in a more scattered spatial pattern of agricultural heritage sites.

Distribution characteristics of CNIAHS categories

Spatial pattern characteristics of different types of CNIAHS

The spatial patterns of CNIAHS types highlight the tendencies of various agricultural systems and illustrate the relationship between heritage site types and their geographical regions (Table 2). Among the 11 heritage types, high-value crop and spice systems, predominantly involving tea and fruit cultivation, constitute the largest proportion and exhibit a clustered spatial pattern. The NNI for both multiple cropping/polyculture farming systems and food crop systems is less than 1, with P-values greater than 0.1, indicating a random spatial distribution pattern. Additionally, except for nomadic and semi-nomadic pastoral systems, which also exhibit a random distribution pattern, the other six types do not display significant spatial autocorrelation. Complex multi-layered home gardens, represented by only one example in China, are not analyzed separately.

Table 2 The NNI of Different Types of CNIAHS
Full size table

Analysis of spatial patterns for different types of CNIAHS

High-value crop and spice systems are predominantly found in southeastern China, forming a tri-core, extensive distribution pattern with dense areas in Beijing-Tianjin-Hebei, Jiangsu-Zhejiang-Shanghai, and Fujian Province (Fig. 4a). Within this category of CNIAHS, fruit and tea production represent 38 and 21 items, respectively. Further scrutiny of these two major heritage types indicates that fruit-related systems are primarily located in the northeastern regions of China, along the Yellow River, Huai River, and Haihe River basins, including Beijing, Tianjin, Hebei, Shaanxi, and Henan (Figs. 4b, 5b). Tea-related heritage sites are concentrated in the southeastern Yangtze River Basin provinces such as Zhejiang, Fujian, and Hunan, as well as in the southwestern provinces of Yunnan, Guizhou, and Sichuan, recognized as the origin of tea cultivation (Figs. 4c, 5c). Comparatively, tea heritage sites in the southeastern region have a more recent origin, and are better preserved, resulting in a more concentrated distribution. Multiple cropping/polyculture farming systems show a relatively dense distribution in the Yangtze River Delta region, an area that experienced rapid agricultural development following southward population migration (Fig. 5d). The core areas for food crop systems are situated in the Pearl River and Yangtze River Basins of Guangxi and Guizhou, as well as in the Songhua-Liaohe River Basins of Liaoning and Inner Mongolia (Figs. 4e, 5e). Within this category, dryland farming and rice cultivation systems exhibit distinct spatial separation, with dryland farming types predominantly found in the northern and western border regions, where natural conditions are less favorable and more suitable for dryland agriculture. The core distribution area of forestry systems is located in the Huai River and Yangtze River Basins, including provinces such as Hebei, Shandong, and Zhejiang (Figs. 4f, 5f). Understory farming systems are concentrated in Zhejiang Province and its surrounding areas within the southeastern river basins, primarily comprising medicinal herb cultivation systems (Figs. 4g, 5g). Nomadic and semi-nomadic pastoral systems display a more dispersed pattern, primarily distributed in Inner Mongolia and the Qinghai-Tibet Plateau Region, representing the only heritage type with relatively high density in the inland river basin (Figs. 4h, 5h). Below sea level systems, though few in number, are located mainly in the Yangtze River Delta and Pearl River Delta regions (Fig. 5i). Animal husbandry systems are mainly distributed in the Yangtze River Basin, with core areas in Sichuan and Chongqing (Figs. 4j, 5j). Mountain rice terrace agroeco systems are primarily found in the Yangtze River Basin, the region where terraced fields first emerged (Fig. 5k). Ancient irrigation, soil, and water management systems are located in the inland river and Yellow River Basins, such as Xinjiang and Ningxia, where natural conditions are less favorable and drought is more severe (Figs. 4l, 5l). The complex multi-layered home gardens category is represented by a single example: the traditional urban vineyard in Xuanhua, Hebei Province.

Fig. 4
figure 4

Analysis of spatial patterns for different types of CNIAHS. a High-value crop and spice systems. b High-value crop and spice systems-Fruit. c High-value crop and spice systems-Tea. d Multiple cropping/polyculture farming systems. e Food crop system. f Forestry system. g Understory farming systems. h Nomadic and semi-nomadic pastoral systems. i Below sea level systems. j Animal husbandry system. k Mountain rice terrace agroeco systems. l Ancient irrigation, soil and water management systems

Full size image
Fig. 5
figure 5

Analysis of basin spatial patterns for different types of CNIAHS. a High-value crop and spice systems. b High-value crop and spice systems-Fruit. c High-value crop and spice systems-Tea. d Multiple cropping/polyculture farming systems. e Food crop system. f. Forestry system. g Understory farming systems. h. Nomadic and semi-nomadic pastoral systems. i Below sea level systems. j Animal husbandry system. k Mountain rice terrace agroeco systems. l Ancient irrigation, soil and water management systems

Full size image

Entropy variation distribution pattern of CNIAHS

Spatial information entropy variation patterns of CNIAHS

From a comprehensive viewpoint, the excavated CNIAHS reveals a process of increasing diversification in types and growing complexity in spatial patterns. The entropy variation values of CNIAHS capture the stage-wise fluctuations and overall upward trend of agricultural heritage sites throughout agricultural development, population movements, and technological advancements in agriculture (Fig. 6). In the Stone Age, with a limited number of heritage sites, a multi-polar entropy increase pattern emerged (Fig. 6a). During the Xia, Shang, and Zhou dynasties, most regions experienced entropy reduction, with only Tibet, Shaanxi, Shanxi, Hubei, Chongqing, and Zhejiang exhibiting entropy increase trends, as the entropy-increasing areas shifted eastward (Fig. 6b). The Spring and Autumn and Warring States Period had the fewest heritage sites, resulting in more extensive entropy reduction areas across the country, while entropy-increasing regions were concentrated along the eastern coast (Fig. 6c). From the Qin and Han to the Wei, Jin, Southern, and Northern Dynasties, China saw significant entropy variations in agricultural heritage sites, with multi-polar growth nationwide (Fig. 6d). During the Sui-Tang, Song, and Yuan dynasties, substantial entropy variations continued nationwide. However, entropy reduction areas increased compared to the previous period (Fig. 6e). A clear spatial differentiation emerged between entropy-increasing and entropy-reducing regions, with the southeastern regions showing significantly higher entropy variations than the northwestern areas, indicating an eastward development trend of agricultural heritage sites. In the Ming and Qing dynasties, despite a decrease in heritage sites, the entropy reduction trend intensified, primarily concentrated in the southeastern regions (Fig. 6f). The northwestern regions exhibited an entropy increase pattern. By this period, following earlier developmental stages, the eastern regions had stabilized in terms of heritage site types and numbers, while technological advancements in agriculture further boosted development in the northwestern regions.

Fig. 6
figure 6

Spatial information entropy variation of different epoch. a Paleolithic Age and Neolithic Age. b Xia, Shang and Zhou. c Spring and Autumn Period and Warring States Period. d Qin and Han to the Southern and Northern Dynasties. e Sui, Tang, Song and Yuan Dynasties. f Ming and Qing Dynasties. The direction of the long semi-axis of the ellipse represents the extension direction of the entropy variation hot spot area, and length of the short semi-axis reflects the aggregation degree of the entropy variation hot spot area

Full size image

Subsequently, a Spearman correlation analysis was performed on the number of heritage sites and entropy variations across different historical periods. Regarding the number of heritage sites (Fig. 7a), starting from the Xia, Shang, and Zhou dynasties, the spatial correlation between different periods showed a strengthening trend, peaking during the Qin, Han, Wei, Jin, and Northern and Southern dynasties. This correlation slightly decreased during the Sui-Tang, Song, and Yuan dynasties, and later increased again in the Ming and Qing dynasties. Notably, the stages of Qin, Han to Northern and Southern dynasties with the Xia, Shang, and Zhou dynasties, the Autumnand Warring States Period, the Sui-Tang, Song, and Yuan dynasties, and the Ming and Qing dynasties exhibiteds tronger correlations, as well as the Autumn and Warring States Period with the Ming and Qing dynasties. Regarding the entropy variation correlation of heritage sites (Fig. 7b), weak positive correlations were observed between the Qin-Han to Southern and Northern Dynasties with the Stone Age, and the Ming-Qing period; as well as the Sui-Tang-Song-Yuan period with the Xia-Shang-Zhou and the Spring and Autumn-Warring States periods. The spatial entropy variation correlations for other periods were negative, indicating that the spatial pattern of heritage sites exhibits distinct transitional characteristics. Collectively, these findings suggest that while the number of heritage sites demonstrates cumulative effects with mutual influences, their spatial patterns display a leap-like transitional properties.

Fig. 7
figure 7

Spearman correlation analysis on CNIAHS’s quantitative and variation. a Correlation analysis of quantitative. b Correlation analysis of entropy variation

Full size image

Spatial information entropy pattern characteristics of CNIAHS

In terms of entropy variation hotspots, CNIAHS reveals distinct periodic spatial differentiation characteristics. The entropy variation distribution shows a reciprocating transition pattern, starting in the central Yellow River Basin, moving to the eastern Yangtze River Basin, returning to the central region, and finally shifting eastward again. This pattern underscores the pronounced periodic nature of China’s agricultural development, with regional cycles contributing to the overarching national spatial pattern of western pastoral and eastern farming, as well as southern rice and northern wheat cultivation. Examination of spatial entropy distribution across provinces reveals significant entropy value fluctuations (Fig. 8). High-entropy regions predominantly appear in eastern areas during different periods, demonstrating periodic reciprocating characteristics. The entropy distribution transitioned from a multi-core pattern in the Stone Age (Fig. 9a) to a central concentration during the Xia, Shang, and Zhou dynasties (Fig. 9b). It then shifted eastward during the Spring and Autumn and Warring States Period (Fig. 9c), reverted to a multi-core distribution from the Qin dynasty to the Northern and Southern dynasties (Fig. 9d), concentrated again in the eastern region during the fifth stage (Fig. 9e), and finally presented a dual-core pattern focused primarily in the east (Fig. 9f). Throughout this developmental process, eastern and coastal regions such as Jiangsu, Zhejiang, and Fujian experienced multiple instances of high entropy values. In contrast, western provinces, except for Xinjiang, Inner Mongolia, and Tibet, which each experienced a single instance of high entropy, generally remained low-value clustering areas. Overall, regions with higher frequencies of high entropy values are predominantly located in the east. Entropy values across all regions of the country show an upward trend, indicating that over time, the diversity of agricultural heritage types in various locations has generally increased. However, the spatial pattern also reveals a concentration of certain advantageous types based on local natural conditions.

Fig. 8
figure 8

Change of spatial information entropy in each provinces

Full size image
Fig. 9
figure 9

Spatial information entropy of different epoch. a Paleolithic Age and Neolithic Age. b Xia, Shang and Zhou. c Spring and Autumn Period and Warring States Period. d Qin and Han to the Southern and Northern Dynasties. e Sui, Tang, Song and Yuan Dynasties. f Ming and Qing Dynasties. The direction of the long semi-axis of the ellipse represents the extension direction of the entropy hot spot area, and length of the short semi-axis reflects the aggregation degree of the entropy hot spot area

Full size image

Discussion

The law of CNIAHS’ spatio-temporal pattern transformation

The spatial patterns and evolution of CNIAHS highlight the transformation processes and changes in China’s agricultural spatial structure, reflecting a diachronic human-geographical configuration shaped by human ecological wisdom in harnessing nature. While most contemporary studies based on spatio-temporal changes and driving factors disclose the motivations behind spatial formation, entropy variation analysis of CNIAHS provides a broader view of the formation processes and patterns of the CNIAHS protection system. There are two divergent perspectives on historical development processes: linear historical determinism [32] and Marxist historical materialism [33]. Entropy variation analysis has revealed that the CNIAHS spatial system shows a weak inter-annual correlation in its pattern. This suggests that development does not adhere to a simple linear sequence based on cause-and-effect but rather demonstrates leap-like spatial transformation. The spatial pattern development and changes of heritage sites in each period result from new rounds of agricultural spatial optimization and balance formed after disrupting the current spatial equilibrium, accumulating from various historical and cultural factors of the preceding period. This leads to cumulative layering in quantity and leap-like transitions in spatial patterns over time. Therefore, the CNIAHS spatial formation mechanism aligns with Marxist historical materialism, emphasizing disruption and leaps in causal chains. This is evident in the agricultural development process as the unity of internal and external factors, contradictions and continuity, discontinuity and continuity, and accumulation and process [34, 35].

Currently, various academic perspectives explain the formation of humanistic spatial patterns, including social ontology [36], geographical determinism [37] etc.. Examining the development and transformation process of CNIAHS, its spatial hotspots show a trend from dispersed to concentrated, shifting from northwest to southeast, with distinct regional typological characteristics. From the dispersed pattern of primitive agriculture in the Stone Age to the Xia, Shang, and Zhou dynasties, the development of slave society advanced agricultural progress. Heritage spaces gradually concentrated in the middle and upper regions of the Yellow River, forming a single-core stage. Subsequently, they distribute mainly in the eastern Yangtze River Delta and Pearl River Delta regions, forming another single-core spatial pattern. During the Yuan, Ming, and Qing dynasties, a dispersed pattern re-emerged. In the formation of the CNIAHS spatial system, social instability and wars prompted population migration, forming new population agglomeration areas, leading to the transition of agricultural spatial patterns. Conversely, stable social environments promoted population growth and agricultural technological advancements, solidifying regional agglomeration. Although this process involved site selection based on geographical conditions and human-land conflicts due to population growth, the fundamental cause of spatial transitions in heritage sites was social factor-induced changes. The regional heritage site type classification after macro-spatial pattern changes was more determined by geographical environmental factors. Therefore, unlike conventional theories that consider the geographical environment the main influencing factor in agricultural spatial formation, entropy variation analysis from a macro-temporal scale reveals that the primary factor driving the formation of agricultural heritage spatial systems is social ontological change. The influence of geographical environmental factors occurs after this transition, as humans adapt to new territorial environments and make agricultural-type choices.

The implications for the construction of CNIAHS protection system

The temporal and spatial pattern characteristics of the excavated heritage can roughly reflect the agricultural development in Chinese history and the dominant factors behind the leap-forward changes. At the same time, some information, that will contribute to the excavation and protection of CNIAHS and the construction of CNIAHS protection system in the future, is revealed.

Firstly, it is a common consensus that the later a heritage emerges, the more likely it is to be preserved up to the present. However, in light of the period of origin of the excavated heritage to date, the heritage from the most recent Ming and Qing Dynasties does not hold an absolute preponderance in terms of quantity (Table 1). Although the entropy change during this period has manifested a distinct shift towards the southwest and northeast, the quantity increase of CNIAHS is far less than that of the previous two periods. Viewed from the regional distribution of heritage sites at different stages, many northern and western provinces retaining the oldest Stone Age heritage, such as Yunnan and Heilongjiang, do not have an advantage in the number of excavated sites. All the above information reflects the deficiency in the exploration of CNIAHS in the northeastern and southwestern regions. Secondly, agriculture is the foundation for the survival of our ancestors. The existence of traditional settlements can, to some extent, indicate that the region has a demand for traditional agricultural production. Existing studies have demonstrated that different types of historical and cultural heritage exhibit certain correlations in the spatial pattern [38, 39]. However, the current CNIAHS presents a monopolar pattern centered in southeastern Zhejiang, which contrasts sharply with the three-core pattern of traditional villages centered in southwestern Yunnan-Guizhou (ranking first in the number of traditional settlements nationwide), southeastern Zhejiang, and northwestern Shanxi. Thirdly, although the southeastern region enjoys a distinct advantage in the CNIAHS pattern, the excavated heritage in this region is mainly inland or freshwater agriculture, lacking the exploration of marine agriculture in the southeastern area, which remains a void in CNIAHS. Fourthly, CNIAHS is inseparable from non-material elements such as agricultural history, agricultural technology, agricultural product varieties, and agricultural folk customs. The exploration of heritage sites can enhance and complement the non-material heritage related to agriculture, and the research on non-material heritage can further facilitate the exploration and protection of heritage sites.

In light of the findings above, this research puts forward following instructions for the construction of the CNIAHS protection system: (1) In terms of spatial pattern, the subsequent excavation of CNIAHS should fan out from point to area to promote the exploration of various types of CNIAHS in northeast and southwest China. It is necessary to enhance the exploration of more types of heritage sites in current hotspot regions, such as Zhejiang and Jiangsu; excavate more heritage sites in potential areas, like Yunnan and Guizhou; and promote the construction and improvement of the CNIAHS protection system in different regions. (2) Regarding the temporal pattern, it is essential to strengthen the research on agricultural history and further explore CNIAHS in different periods, especially paying greater attention to those periods with obvious exploration potential. (3) From the aspect of protection level, the construction of a protection system centered on key heritages cannot be separated from an authentic regional historical environment, which means the protection of some heritage with historical significance but no outstanding characteristics is important as well. (4) In terms of heritage types, the exploration, research, and protection of material and intangible agricultural cultural heritages are mutually reinforcing and indispensable.

Conclusion

Agricultural heritage systems represent the crystallization of the productive activities and survival wisdom of working people over the long term. As a legacy vital to humanity’s future, they encompass rich multidimensional values in social, economic, cultural, and ecological domains, as well as significant ecosystem service functions. As one of the world’s three major agricultural origins and an early responder to the FAO’s initiative, China has been at the forefront of CNIAHS identification and conservation. However, compared to other heritage types, there remains considerable room for development in constructing spatial systems and implementing systematic protection for agricultural heritage. Existing research has predominantly focused on the spatial patterns of agricultural heritage, but studies on the historical development of these patterns are relatively scarce. This has led to a lack of comprehensive strategies for regional agricultural heritage spatial construction in current identification and conservation efforts. To construct a complete CNIAHS protection system, it is necessary to broaden the scope and content of protection. This should not be limited to the conservation of tangible heritage but also include the identification and protection of intangible agricultural heritage, such as agricultural literature and techniques, to foster regional traditional agricultural development and promote traditional agricultural technologies. Through an analysis of entropy variation of heritage sites across different periods, this study has identified leapfrog-like spatial pattern transitions in the development process of CNIAHSs. By integrating this analysis with the history of Chinese agricultural development, the influencing factors and evolution patterns have been traced. This provides a reliable basis for constructing a comprehensive spatial perspective on agricultural heritage. From a broader perspective, it offers valuable references for characterizing the regional features of agricultural heritage systems.