A comparison of isoglosses of folk nomenclature between local trees and rice landraces revealed clear differences (Figs. 3, 4). Results from the V-measure and factor detector analyses suggested that the differentiation in the divergence processes of the isoglosses between the two plant resources may be attributed to different geographical and sociological variables (Table 1, Fig. 5). The divergence of local tree names showed a stronger association with geographical variables such as altitude, precipitation, temperature, and daylight hours (Table 1). Our results suggested a spatial association between the folk nomenclature group of trees and areas with slash-and-burn cultivation extending from the eastern to the central inland parts of the prefecture (Table 1, Fig. 1, Additional file 1: Fig. S1). This included the inland mountainous region in Shikoku where slash-and-burn cultivation was once widely practiced [25]. These findings may correspond to the shared knowledge of trees serving as fertilizer sources for the slash-and-burn cultivation in this area.

Unlike trees, rice is an annual plant, and rice landraces with desirable traits have been frequently introduced from region to region. Locally well-adapted varieties were selected until the mid-twentieth century [13]. During the Shikoku Henro Pilgrimage, in which pilgrims circumambulate the 88 temples on Shikoku Island mainly in a clockwise direction [37], rice landraces were scrutinized by pilgrim farmers in autumn, who later brought back panicles to their homelands [38]. In fact, rice landraces with a prefix or suffix related to the pilgrimage (e.g., “Henro” or “Hendo” or with the name “Henro” itself) were recorded in our source reference [16]. Moreover, landrace names associated with other pilgrimages to Izumo Shrine and Ise Shrine [39, 40], such as ‘Izumo’, ‘Izumo-modori’, ‘Izumo-wase’, ‘Ise’, ‘Ise-shiro’, and ‘Ise-naka’, indicated domestic seed transfers during farmers' travels. As shown in Fig. 4, indicator varieties included place names such as “Ise” shiro, Kuro “izumo”, Uwajima, and “Oozu” nishiki, indicating both pilgrimages and other seed transfer opportunities from outside, making the landrace composition unique in each region. Rice cultivation is closely associated with Japanese local rituals [41], which accords well with the V-measure results, which show better spatial association between the isogloss maps of rice landraces and folklore types as compared to those of local tree names.

Dialects are known to dynamically change with human mobility and be influenced by various factors, for example, infrastructure (e.g., roads) [42]. The presence of a specific group of rice landrace names at the road intersection areas of Shufu and Kuwamura, both recorded and depicted in the late nineteenth century, can be considered as an example of this dynamism.

Amid Japan's economic stagnation in recent decades, there has been a push for local government mergers. Remote municipalities, especially those in mountainous areas, face challenges such as aging populations, declining population numbers, and budget constraints. Consequently, implementing concrete measures to address national environmental issues, such as the utilization of biomass energy, has proven difficult at the local municipality level [43]. Therefore, effective surveys are required to conserve agricultural biodiversity in such areas, including indigenous knowledge such as local plant names and traditional crops, while considering the resource constraints of local governments. From such a perspective, our approach is similar in concept to that proposed by Jeszenszky et al. [5], in that it utilizes clustering results of existing linguistic data. In the concept of biocultural diversity, it is thought that the diversity of life comprises of biological, cultural, and linguistic diversity in a complex socio-ecological adaptive system [6]. Under this assumption, our approach focusing on plant folk nomenclature has been shown to be effective for the rapid selection of a reduced number of survey sites to revisit for the conservation of plant resources as biological diversity assets and of local knowledge as cultural and linguistic diversity assets. According to our V-measure analysis (Table 1), high homogeneity (at K = 2–5) indicates that the variation in folk nomenclature of local trees and rice landraces within municipalities was not large. Conversely, the low completeness and the group formation in the maps (Figs. 3, 4) suggest that the same folk nomenclatures of local trees and rice landraces are distributed across adjacent several municipalities. Under limited human and budget resources, these results suggest how inter-municipal collaboration can be spatially designed to preserve not only linguistic diversity, but also indigenous knowledge related to those plant resources and traditional rice varieties themselves.

The results of this study provide important insights into the selection of survey areas for indigenous knowledge and traditional crops in regions without existing plant folk nomenclature data. A quantitative analysis of dialect in Finland [15] suggested a process in which multiple complex factors influence the isolation of settlements, leading to dialect divergence. The results of the V-measure and factor detector analyses in this study did not reveal any unique thematic maps strongly spatially associated with the isogloss maps of local trees and rice landraces. However, a couple of thematic maps had moderately stronger spatial associations with the isogloss maps of the two plant resources in different ways. This result may reflect a similar community isolation process governed by multiple factors [15]. If such a process prevails in other regions, it would be effective to overlay multiple available maps, such as folklore and environmental variables showing higher V-measure scores and q-statistics in this study, using the Geographic Information System and to identify possible local areas that may have a different culture of plant resource usage before conducting field surveys.

In this study, hierarchical clustering by Ward linkage was applied to create isogloss maps for local tree names using the novel distance index and rice landrace names using Jaccard distance. The clustering results at K = 2–5 showed that all municipalities in each group were in adjacent relationships, making it easy to interpret as natural group formation. The method may have some inaccuracies as compared to dialect comparison studies that target specific balanced lexical sets. However, many folk nomenclature studies were conducted individually in specific regions according to their respective purposes, making it difficult to prepare balanced lexical sets by using these different information sources. Therefore, proposing a new dissimilarity index that enables inter-regional comparisons using such unbalanced lexical sets from various data sources is considered beneficial. Evaluation of word formative structure [44], semantics [9,10,11], and etymology [12] of plant names will also provide information about human contact history relative to plant resources. Moreover, reflecting cognancy and similar wording in the dissimilarity index is another lexico-statistical challenge, but the identification of cognates requires a considerable amount of time [45]. The determination of how to weigh similar vocabulary and formative structure is also a significant challenge when dealing with a vast vocabulary set. From these perspectives, the distance metric we propose in this study is versatile because it can mechanically identify groups using similar plant folk nomenclature immediately.

Much of the folk nomenclature for trees referenced in this study was recorded locally several decades ago, and rice variety names were recorded over a century ago. Consequently, much of the biocultural diversity associated with this folk nomenclature may have already been lost. Therefore, the biocultural diversity remaining today is invaluable, and efforts to collect it are urgently needed. However, there is a wealth of descriptive data of plant folk nomenclature, and representative references were compiled in [46]. Using such available information, it is desirable to test our approach in different or wider regions and verify its validity in additional studies. Fletcher et al. [47] discuss the importance of local and global economic and political efforts to expand protected areas with local linguistic and cultural connections. Our analytical approach will contribute to such efforts, particularly the use of the proposed novel dissimilarity index on plant folk nomenclature utilization for the selection of survey sites. Moreover, this study highlighted the dynamic nature of plant folk nomenclature divergence and the importance of inter-municipal collaboration in conserving indigenous knowledge and resources.