**1. Introduction**

Archaeological sites provide a unique source for recognizing human–environment interactions and the ecosystems a ffected by di ffering degrees of human impact at a wide variety of temporal and spatial scales and thus both reconstruct past environmental conditions and reveal human behavior [1]. As early as 4000 BC, the ancient humans from the Banpo site had primary thoughts of utilizing the environment according to their residential locations along Weihe River [2]. With the increase of urban sprawl worldwide, archaeological sites are facing more threats from the shifting relationship between culture, climate, and land use changes [3], meaning that we need to re-recognize the various archaeological information and their cultural connotation under the human–environment relationship. Through the study of site selection pattern of archaeological sites, we can learn more about ancient environmental view and behavior patterns, which will help us with further archaeological research and in the conservation of cultural heritage.

Ancients' environmental preference was to select a site following basic principles, of which geomancy was one of the most classical theories in ancient life [4,5]. Geomancy claims to utilize nature to harmonize individuals or their communities at scales ranging from vast landscapes to houses and is indicated as the method and practice employed to guide site selection of a settlement, city, or tomb according to the surrounding natural environment [6–10]. It is necessary for us to learn this kind of site selection pattern in archaeological study. In the 1950s, this theory began to formally enter into the perspective of modern science. As a milestone event, Subai introduced geomancy into an archaeological study of the tombs from the Song Dynasty in the northern Paisha Town [11]. Thereafter, analysis based on geomancy rules gained extreme attention from modern scholars for its hidden scientific value [12–15]. Magli [16] discussed the cultural links between topography and traditional geomancy for studying royal mausoleums. Moreover, Gui [17] studied cultural landscape changes and site selection under geomantic principles in the Dongcun settlement of the Taihu Basin. Yang [18] further emphasized the function of geomantic evaluation in landscape and human–land relations. The significance of landscape ecology analysis was to explain the ancient environmental preference from geomantic theory, and it has facilitated some valuable research [19–23]. This was also the point of view we focused on in our study.

The ancient view on the geomantic environment is essentially related to the ecological landscape. Typical ecological characteristics are shown in the edge effect in ecological transition zones and the interaction of ecological units in an ecosystem. The edge effect plays a decisive role in the formation of ecological structure and dynamics of ecological patches. In a variety of applications, it has been used in the fields of microclimate, plant diversity, and bird communities [24–26]. In particular, in forest ecosystems, the impact of the edge effect has received universal attention [27]. Zhou et al. [28] focused on the boundary sensitivity, which is associated with forest ecotone types and exhibits remarkable habit heterogeneity at different scales. Yuan [29] and Chen [30] studied the coupling network of forest landscape patches and analyzed the characteristics of the edge effect in major forestry landscapes mixed with the arborous layer, shrub layer, and herbaceous layer. Although the edge effect and its scale problem have rarely been considered in spatial archaeology studies, it cannot be ignored in practice. In our study of edge effect, we did not focus on biocenosis specifically but put emphasis on the syntagmatic relations of ecological patch to look for site selection patterns. Therefore, the calculation of edge effect was different from previous studies. However, we still considered the factors of shape and size of patches and the quantified form in the established model, as mentioned in some other studies [31,32].

For the other ecological characteristics, the internal relations of di fferent units in the ecological network was proposed in cultural landscape analysis in this study. The ecological network provides an operational methodology in the practices of studying ecological structure that relies on the concept of connectivity included in landscape graphs [33,34]. In recent years, there has been a rapid increase of studies that advocate network analysis to ecologically manage landscapes that su ffer from fragmentation and loss of connection [35]. The focus is on the aspects of establishment and evaluation of networks. For example, habitat networks were constructed to assess how climatic variability influences potential connectivity for water organisms in the Murray Darling Basin of Australia by Robbi et al. [36]. Andrea et al. [37] proposed one methodology to build a network with the patch's area as the weight index of nodes for landscape planning in the peri-urban and urban areas of the town of Nuoro (Italy). When evaluating the network, connectivity metrics of landscape graphs have usually been applied to reflect a basic form of interaction between species and their environment [38–40]. Beyond ecological processes and social issues, network analysis is also meaningful in terms of cultural landscape perspectives. For instance, Agnieszka et al. [41] studied sustainability in terms of a cultural landscape service located in Malopolska Province (Poland) by quantitative assessment and qualitative categorization based on landscape diversity and connectivity. In fact, successful cases like this are rarely presented, except for the most qualitative discussions about network characteristics which are di fficult to explain the internal mechanism of ecological structure [42–44]. In this study, we tried to

build ecological networks to quantify the site selection pattern based on the connections between the ecological landscape and geomantic environment.

By both the edge e ffect and network structure in the ecosystem, the ancient site selection pattern and the scientific basis could be discussed according to the discovered archaeological sites in the study area. Multi-source remote sensing data in two times phases were used to reflect the changes of ecological structure. It could help us to speculate the major ecological structure and reveal the human–environment relationship in ancient times.
