Introduction

Through globalization, ethnic groups throughout the world are coming in contact with new people who bring with them new ideas, cultural information, and practices. Yunnan Province, known in mainland China as the “wild frontier” due its high concentration of ethnic minorities, intertwined with its incredibly diverse natural habitats, provides a perfect location to study cultural exchange. Northwest Yunnan holds the status as a biodiversity hotspot, due to extreme richness of flora and fauna as well as high risk of danger to this diversity (Myers et al. 2000). Approximately 25 of the 55 ethnic minorities in China reside in Yunnan. Some of these ethnic minorities live in villages consisting predominantly of members of their own ethnic group, while other minorities have frequent contact with other nearby ethnic minorities and/or the Han majority.

Many studies focusing on cultural ecological knowledge exchange due to contact with outside groups have been conducted world-wide. For example, adoption of other cultures’ ethnobotanical knowledge and practices was shown among the Piaroa of Venezuela, who increasingly came in contact with other ethnic groups, missionaries, and majority ethnicities (Zent 2001). Similarly, Iquito speakers in Peru have incorporated uses of medicinal plants from other cultures into their own pharmacopoeia (Jernigan 2012). A study conducted in the Doñana region in Spain shows that an increase in participation with a market economy has impacted traditional agricultural knowledge (Gomez-Baggethun et al. 2010). Additionally, loss of traditional ecological knowledge was also evident in Shoshone of the Great Basin due to loss of language and general change in diet and medicinal use (Fowler 2000). Weckerle et al.’s (2009) research in Shaxi Valley, Yunnan, suggested that the Han and Bai have spent many years living in close contact and, as a consequence, the Bai people have adopted some of the local Han majority’s use of medicinal plants described in Traditional Chinese Medicine (TCM) practices and have little separate medicinal knowledge.

Increased contact, however, does not always lead to exchange of cultural knowledge. For example, research conducted by Pieroni et al. (2011) in southwest Serbia compares the ethnobiological knowledge of Serbians and Albanians who have been in contact with each other for approximately 300 years. The Serbians and Albanians in this area have maintained separate, unshared knowledge (Pieroni et al. 2011). Likewise, research with Senegalese immigrants in northwest Italy showed that those interviewed did not vary their medicinal plant uses despite living in a different country (Ellena et al. 2012). Zarger and Stepp (2004) illustrated, through a comparative study with Tzeltal Mayan children conducted over a span of 30 years, that despite many changes to the community, children have retained consistent knowledge about plants. In a study conducted with the Tsimane’ of Bolivia, respondents from villages of varying distances from a market town share a high cultural consensus of ethnobotanical knowledge, despite having a higher income and increased contact with outsiders (Reyes-Garcia et al. 2005). In Yunnan Province, research by Huai et al. (2011) found that homegarden area, types of species, and number of species among eight different ethnic groups were significantly different between ethnic groups. The authors stated, “In China different ethnic groups have lived many generations in close proximity to each other while still retaining their own cultures and traditions” (Huai et al. 2011).

Our study expands on previous work conducted on knowledge exchange between ethnic groups by focusing on seven ethnic minorities as well as the Han majority in Yunnan Province, China. The cultural groups studied here are the Bai, Dulong, Han (majority), Lisu, Naxi, Nu, Tibetan, and Yi. We examined different factors that may affect sharing of knowledge. Groups sharing similar languages (Bradley 1997) and/or who live in close contact with other ethnicities are the Bai and Han (live in heterogeneous settlements), and the Dulong, Lisu, and Nu (share similar languages and live in heterogeneous settlements). Groups with individuals that are generally mono- or bilingual and have ethnically homogeneous settlements are the Naxi, Tibetan, and Yi. Groups that have factors that suggest they are likely to share cultural knowledge may have a weak cultural identity and risk a change in cultural knowledge.

To test cultural exchange we conducted surveys to determine the ethnobotanical knowledge surrounding Rhododendron (Ericaceae) and the response heterogeneity within and among ethnic groups. Rhododendron spp. is a useful tool to determine response heterogeneity because worldwide evidence suggests that Rhododendron is important ethnobotanically to various ethnic groups (Koca and Koca 2007; Kunwar et al. 2006; Pei 1989; Skinner 1903). Rhododendron is a large genus of approximately 1,000 species of evergreen or deciduous shrubs to trees with showy flowers, and over 500 species are native to China (Fang et al. 2005).

An ethnobotanical survey of uses of Rhododendron in Yunnan Province, China, was used to test the contrasting hypotheses that ethnic groups either extensively exchange cultural practices (Weckerle et al. 2009) or that they retain separate cultures (Huai et al. 2011). The objectives of this study were to 1) test the contrasting hypotheses laid out by Weckerle et al. 2009 and Huai et al. 2011, and 2) determine which groups have heterogeneous responses to interview questions suggesting that they are less culturally cohesive than groups with homogeneous responses.

Methods

Study Sites

Interviews were conducted by E.G. in Dali Prefecture (Dali City, Eryuan County, Midu County), Diqing Prefecture (Deqen County, Shangri-La County, Weixi Lisu Autonomous County), Lijiang Prefecture (Gucheng district, Ninglan Yi Autonomous County, Yulong Naxi Autonomous County), and Nujiang Prefecture (Fugong County, Gongshan Dulong, Nu Autonomous County) in Yunnan Province’s northwest. Fig. 1 highlights the Prefectures where these cultural groups are mainly concentrated, but exact interview locations (including village names) are withheld to protect the anonymity of our interviewees. Bai, Han, Naxi, Tibetan, and Yi villages were easily reached by roads (sometimes paved), while Dulong, Lisu, and Nu villages often were reached on foot. Bai and Han interviews were conducted in the same area so that they would be comparable. Bai and Han interviews were mainly conducted in village markets as the high tourism pressures create a strong divide between locals and foreigners, making it challenging to conduct interviews in homes (E.G. field notes 2011 and 2012). Dulong, Lisu, Naxi, Nu, Tibetan, and Yi interviews were often conducted in the home or farm plot of the interviewee. The homes visited of members in the latter six ethnic groups had electric lights, but no indoor heating and typically no indoor plumbing. Fire was used for cooking. Water is piped to houses from nearby rivers through a series of hoses. The vast majority of homes visited had a functioning television, and members of the household had cell phones with reception.

Fig. 1
figure 1

Map of Yunnan highlighting the regions where interviews were conducted.

Full size image

The descriptions of these seven ethnic minorities and the Han majority are based on observations by E.G. during fieldwork (May 2010–August 2010, January 2011–August 2011, February 2012–June 2012) in particular villages. They are thus not necessarily representative throughout Yunnan. For example, the Yi ethnic group is widespread throughout Yunnan, and their language and dress vary depending on the location (Harrell 2001). However, the information presented is valid for the specific areas where this research was conducted.

Fieldwork

An interview pilot study was conducted in northwest Yunnan in June–August 2010 and February 2011 to test the validity of the questions asked and the relevance of the study. Interviews used for this research were conducted in March–July 2011 and February–June 2012. A total of 252 interviews were conducted during these two field seasons (30 Bai, 29 Dulong, 24 Han, 45 Lisu, 32 Naxi, 32 Nu, 29 Tibetan, 31 Yi). All interviews were conducted in Mandarin Chinese or a minority language by E.G. and a trained local field assistant, which allowed for a comfortable atmosphere for the interviewee (trained by E.G. following the University of Wisconsin-Madison’s Social and Behavioral Science [SBS] Internal Review Board [IRB] guidelines; this interview protocol was qualified as exempt as of June 10, 2010). E.G. speaks the Yunnan dialect of Mandarin Chinese, and the field assistants usually spoke a minority language (Dulong/Nu, Naxi, Tibetan, Yi) and Mandarin Chinese. Throughout the course of this fieldwork eight different field assistants, who are kept anonymous, aided in this research.

To avoid confusion due to different words for “rhododendron” the Mandarin Chinese word (杜鹃花)) was avoided, and instead interviewees were shown a contact sheet of a compilation of photographs of numerous Rhododendron species (different contact sheets were shown for different areas with local species pictured). If the interviewee recognized the flowers (typically by describing the habitat), the interviewee was included in the survey. The interviews were semistructured and conducted with adults (18 years or older) of the eight different groups. Here we report on the structured interviews only, which were limited to ten Yes or No questions (Table 1). The interview questions were printed on separate sheets and were filled out for each individual either during or directly after the interview. Snowball sampling was conducted with no more than five people per snowball to prevent similarity of answers based on familiarity of individuals.

Table 1. Initial Yes and No Questions Asked During the Interviews.
Full size table

Cluster Analysis

Using a spreadsheet, the ten initial Yes/No interview questions were coded as numbers (1 = Yes; 0 = No; 0.5 = Maybe/Ambiguous; 3 = No response or I don’t know). Interviewees were given the option to not answer questions if they 1) felt uncomfortable responding to a question, 2) did not understand the question, 3) did not know the answer, or 4) did not have enough time to complete the interview (all coded as 3). Only a few interviewees chose to decline to answer some questions.

The program Primer 6 (Clarke and Gorley 2006) was used to conduct cluster analyses, to visualize similarities within and among cultural groups. The cluster analysis was square root transformed and resembled using a Euclidean distance distribution. The data were transformed in order to increase the statistical power of the analysis. A Euclidean distance resemblance is a multidimensional Pythagorean Theorem that calculates the differences in data. The specific cluster analysis conducted was a group average (UPGMA). Three analyses were conducted to allow for comparison of cultural groups. Analysis one allows for a comparison between the Dulong, Lisu, and Nu ethnic groups who live in close contact with one another (Fig. 2). Analysis two allows for a comparison between the Naxi, Tibetan, and Yi ethnic groups who live in homogeneous settlements of their own ethnic group (Fig. 3). Analysis three allows for a comparison between the Bai minority and Han majority within Dali Prefecture (Figs. 4).

Fig. 2
figure 2

Cluster analysis of Dulong, Lisu, and Nu interview responses.

Full size image
Fig. 3
figure 3

Cluster analysis of Naxi, Tibetan, and Yi interview responses.

Full size image
Fig. 4
figure 4

Cluster analysis of Bai and Han interview responses.

Full size image

Cultural Cohesion

For a finer analysis of the within-ethnic-group variation of answers (compared to the cluster analysis), a new analysis method was designed, evaluated, and utilized. This calculation was conducted individually for each ethnic group.

  1. Step 1:

    Count the number of Yes responses as 1 unit and Maybe/Ambiguous responses as 0.5 units. Yes and Maybe responses were counted rather than the No responses because we wanted to focus on the known uses for rhododendrons. Counting of Yes answers is similar to McMillen’s (2012) Local Ecological Knowledge (LEK) scores, yet different because we also counted Maybe/Ambiguous responses. Answers scored as 0.5 included responses such as, “We think that rhododendrons will be important for tourism in the future, however they are not now” or, “I know people who use rhododendrons for medicine but do not know how to use them myself.” Another difference from McMillen (2012) was that these numbers were not counted as 0 = No Knowledge and 1 = Knowledge, because with our data 0 = No could mean either no knowledge of that kind of use or no use of rhododendron for that question.

  2. Step 2:

    Divide the sum of Yes and Maybe responses by the total number of interviewees for each ethnic group because the total number of interviews completed for each ethnic group was not equal. The division of Yes answers by the total number of interviewees is similar to the calculation of use values in Shen et al. (2010b), although they did not turn their final numbers into percentages.

  3. Step 3:

    Turn the product from divisions in Step 2 to percentages. Percentages allow for easy comparison of each ethnic group’s shared knowledge. A high percentage of Yes responses indicates that interviewees share similar knowledge about rhododendrons within their ethnic group. A very low percentage of Yes responses also indicates that interviewees share similar knowledge about rhododendrons in that they do not have uses for rhododendrons in a particular use category.

  4. Step 4:

    Include the percentages from Step 3 in a table and highlight the cells for percentages of 10% and below and 70% and above (see Table 2). Middle percentages (11% to 69%) would be inappropriate to include in this cultural cohesion analysis because an ethnic group with 50% Yes responses to a question would show that the group does not share knowledge for that question. If an ethnic group had all responses with middle percentages, it would suggest that the ethnic group does not share knowledge among their members and may not be culturally cohesive. We tested several different threshold percentages (20% and below and 70% and above, 50% and above, 10% and below and 50% and above, etc.), but these percentages did not prove informative. For example, 20% and below and 70% and above was not informative because counting 20% and below increases the number of cultural cohesion units drastically, especially in groups that have high heterogeneity of answers, thus showing false homogeneity.

    Table 2 Cultural Cohesion.
    Full size table
  5. Step 5:

    For these data, a group was said to be culturally cohesive with a number of five because that is just over the average of our cultural cohesion numbers (4.9). Five is also half of the maximum score of ten cultural cohesion units that is theoretically possible because ten questions were asked of each ethnic group. A score of ten cultural cohesion units would represent a percentage of 70% or above or 10% and below for all answers. Numbers below five showed low cultural cohesion.

This measurement was used in place of Romney et al. (1986) Cultural Consensus Analysis because we aimed to calculate the “cultural cohesion” of each ethnic group. In this study cultural cohesion is the knowledge of uses of rhododendron shared within an ethnic group suggesting a strong cultural tradition. The Cultural Consensus Analysis (Romney et al. 1986) is used to look at patterns of heterogeneous responses to questions within an ethnic group, whereas we are looking at patterns of heterogeneous responses among multiple ethnic groups.

Furthermore, calculating the culturally correct answer (using the Cultural Consensus Analysis) may limit the data by creating conflicts with the answers of the “experts” we interviewed (e.g., religious specialists or traditional medicine doctors; Zent 2001). This does not happen with calculating the cultural cohesion because total counts of answers are combined for every question. A conflict with specialist vs. general knowledge would only occur when interviews with specialists equal those of generalists in the community, or if a very small number of interviews were conducted. This analysis would not be appropriate for those data unless generalist and specialist knowledge were calculated separately. This analysis is appropriate for categorical data that can be summed and where approximately 30 (or more) interviews have been conducted in each culture.

Results

Cluster Analysis

Three different cluster analyses from the structured interview data were conducted. Figs. 2, 3, and 4 show results of cluster analyses of the interview responses, which were used to test the extent to which these cultures’ uses have remained distinct (Huai et al. 2011) or have been extensively exchanged through contact (Weckerle et al. 2009). A cluster analysis was conducted of the Dulong, Lisu, and Nu ethnic minorities, which live in close proximity to each other in Nujiang Prefecture (Fig. 2). Ethnic groups living in distinct groups (Naxi, Tibetan, and Yi) were included in a separate cluster analysis (Fig. 3). Fig. 4 shows the results of a cluster analysis of the interview responses of the Bai minority and Han majority. In the three cluster analysis figures, the tips connected with a line indicate respondents answering questions with all the same responses to all ten questions.

Cultural Cohesion

No interviewed ethnic group scored a maximum cultural cohesion number (ten units) or a minimum cohesion number (zero units). Four ethnic groups scored cultural cohesion numbers of five units and above (Dulong: 7 units; Tibetan: 7 units; Naxi: 6 units; Yi: 5 units), and four cultural groups scored cultural cohesion numbers of four units and below (Bai: 4 units; Nu: 4 units; Han: 3 units; Lisu: 3 units; Table 2). This analysis allowed for a comparison of the questions for which each ethnic group had homogeneous or heterogeneous interview responses.

Knowledge of Uses

Ethnic minorities often have vast knowledge about the plants that grow around them (Campas and Ehringhaus 2003), including the ever-present Rhododendron. The interviews conducted for this research identified uses for rhododendrons by cultural groups in Yunnan, some of the uses previously undescribed, and others that expand on previously described uses from China and other parts of the world. Uses of Rhododendron that were stated by the interviewees are summarized in Table 3, and selected uses will be described in more detail in a forthcoming manuscript.

Table 3 Summary of Knowledge of Uses of Rhododendron by the Eight Ethnic Groups Interviewed.
Full size table

Discussion

Intermingling of Cultural Groups

The Dulong, Lisu, and Nu ethnic minorities live in close contact with one another, frequently intermarry, can often speak each other’s minority languages, and live in the poorest prefecture (Nujiang Prefecture; Shen et al. 2010b). Fig. 2 illustrates the effect close contact can have on these ethnic minorities’ responses to interview questions about rhododendrons. Dulong, Lisu, and Nu have intermixed representation among the branches as well as cases of different ethnic minorities answering questions with the exact same responses (see Fig. 2 *1 and *2). When interviewees responded to all questions the same way, indications are that these groups share similar knowledge of uses despite being different cultural groups. In contrast, the Naxi, Tibetan, and Yi ethnic minorities do not live in close contact with one another or with other cultural groups. In many cases, these ethnic groups answer questions similarly to others within their same ethnic group (Fig. 3 *1 and *2). This suggests that knowledge is shared among members within an ethnic group but not between these ethnic groups.

The responses that unite the Bai cluster to the bottom in Fig. 4 (denoted by an *) are due to interviewees responding Yes to a few questions (mostly eating and tourism) but had No Response (coded as 3 in the spreadsheet) to the other questions. In Dali, Prefecture Bai and Han are often in contact with one another at markets and in the tourist destination of Dali Old Town, and have a long history of contact (Weckerle et al. 2009). It is possible that this ever-increasing contact due to rapid development of the area is causing exchange of knowledge of uses between the Bai and Han (e.g., Han people eating Rhododendron corollas). Because of the challenges interviewing Bai and Han individuals at home, we acknowledge that the Bai and Han results may be biased because people frequently exchange knowledge in markets (Reyes-Garcia 2001; 197).

The evidence provided from Figs. 2, 3, and 4 refute hypotheses, such as those described by Huai et al. (2011), which suggest that ethnic groups that frequently intermingle do not share ethnobotanical knowledge, at least for this area of Yunnan. The data collected from the ethnic groups interviewed for these analyses suggest that living in close contact with other ethnic groups encourages sharing of knowledge, as Weckerle et al (2009) suggested (Figs. 2 and 4), while isolation of ethnic groups may prevent sharing of knowledge among groups (Fig. 3). Ethnic groups experience cultural exchange not only due to contact with a majority culture, but with other minority groups through adoption of other, outside knowledge, ideas, and beliefs.

Cultural Cohesion

The cultural cohesion analysis was used to test which cultural groups answer questions related to rhododendrons more similarly within or among the group. Results of the cultural cohesion analysis generally show that ethnic groups that share languages and live close together have heterogeneous knowledge of uses within groups, thus a low cultural cohesion number (Table 2; Bai, Han, Lisu, and Nu). The Dulong, Naxi, Tibetan, and Yi answer questions similarly to those within their ethnic group and thus have higher cultural cohesion numbers. Additionally, the Naxi, Tibetan, and Yi are the only ethnic groups that had cultural cohesion percentages above the 70% threshold. The Naxi, Tibetan, and Yi live in villages comprised of other members of their own ethnic group and frequently are monolingual or bilingual, speaking only their own ethnic language and potentially also Mandarin Chinese. Dulong, Lisu, and Nu are often multilingual, speaking their own language, languages of other nearby ethnic groups, and Mandarin Chinese (Goodman 2008; E.G. field notes). For example, Dulong people can speak Dulong and often Nu and/or Lisu languages as the Dulong and Nu languages are very closely related (Bradley 1997). Speaking other languages and living in ethnically heterogeneous settlements appears to create heterogeneous knowledge of uses about Rhododendron. Nujiang Prefecture is the poorest in Yunnan Province and is where Dulong, Lisu, and Nu people are concentrated (Shen et al. 2010b). This lack of wealth may also cause correlation of results where Dulong, Lisu, and Nu have little knowledge of uses of Rhododendron.

The high cultural cohesion score for the Dulong may be misleading, because all of the cultural cohesion percentages show that the Dulong do not have many uses for Rhododendron (Table 2 and 3). This may be due to a lack of traditional uses of the plant or severe erosion of traditional environmental knowledge of Rhododendron uses (or a combination of both). Despite being located in a rural, mountainous area on the border of Myanmar and Yunnan, it may be that Dulong ethnobotanical knowledge has been shrinking over time (Luo 1995; Shen et al. 2010a). Preliminary discussions with Dulong people suggest a rapid and intense change of their culture during the past 50 years (E.G. field notes 2012). An example of this may be the extinction of the practice of face tattooing of women of the Dulong ethnic group (Luo 1995). This practice was common before the Cultural Revolution but has completely stopped, rather than gradually losing fashion (E.G. field notes 2012).

Conclusions

Ethnic groups living in the cultural and biodiversity hotspot of northwest Yunnan commonly exchange knowledge of uses of rhododendrons. Ethnic groups that have heterogeneous knowledge often live in settlements with multiple ethnic groups, are multilingual, and some live in the poorest prefecture in Yunnan. Ethnic groups that have homogeneous knowledge of uses of rhododendrons may be mono- or bilingual and generally live in settlements of members of their own ethnic group. There are many factors that affect cultural exchange, though in northwest Yunnan it appears that living in close proximity and being multilingual are the largest influences for cultural exchange.