Small-scale societies, economic change and market integration

Most in situ economic transitions from self-sufficiency to cash-based market integration among small-scale societies are driven by a set of factors that have remained largely consistent across place and time since the colonial era [41]. These factors can be summarized into three areas 1) exchange of information and goods with outsiders that engender desires for acculturation and technology acquisition, 2) encroachment pressures via deception, threat or violence, from outsiders seeking land, labor and natural resources and 3) increased population size that strains territorial carrying capacity by depleting the natural resources needed for a self-sufficient lifestyle [42–45]. Together, these factors help to determine new patterns of intergenerational wealth transmission, sometimes resulting in an increasingly more unequal distribution of opportunities via existing and emergent inequalities [46]. For this reason, determining how access to opportunity is affected by market integration and its specific contextual drivers in the short- and long-term is of particular concern in developing settings. Globally, increasing levels of inequality threaten to induce political instability and imperil the well-being of many resource-insecure populations [7,14,47].

Social incongruity

Research on social incongruity has provided novel evidence for the multidimensional nature of social status by capturing the individual consequences of simultaneously occupying conflicted social positions [7,21]. For example, Dengah [48] found that social incongruity among Brazilian Pentecostals due to conflict between religious and secular lifestyles explained more variance in individual well-being than even SES. Similarly, Sorensen et al. [38] discovered that among Yakut reindeer herders undergoing economic change in Siberia, SES in excess of material resources resulted in higher levels of stress for respondents with market- and subsistence-based lifestyles. When considered in conjunction with Gravlee and colleague’s [49] work on color incongruity in Puerto Rico, these studies highlight the context specificity of social incongruities and the potential for culture to mediate inequalities in a population.

The frequent occurrence of social incongruity at different societal scales and levels of wealth also represents a puzzle as to its origins and purpose in human populations. One possible explanation comes from theories of social identity, social dominance and system justification [49]. All three contend that inequalities due to determinants that are at least partially outside of an individual’s control (like social incongruity) are rooted in the maintenance of the status quo by the elite and powerful in a society. In particular, high-status individuals (and those aspiring to high status) actively police social boundaries as a way of preserving the system that empowers them. Several common social phenomena have been empirically linked to this type of social policing including dominance behaviors, ethnocentrism, homophily, ingroup bias, intergroup conflict, outgroup antipathy, resistance and self-interest [49]. The evidence supporting these theories resonates with studies on social incongruity that support its possible role in policing social boundaries for the maintenance of norms and existing hierarchies. What is not clear, is whether the role of social incongruity is the same in non-market and market economies and therefore if social incongruity has the same consequences on social status and wealth inequality in both settings [48,50,51].

Tsimane’ social status and social hierarchy

Prior to the 1970s–80s, and like many other Amerindian groups, religious belief and health maintenance were inextricably connected in Tsimane’ ontology (i.e. worldview), making the medicinal and spiritual guidance of the local cocojsi (shaman) powerful in the Tsimane’ social order [53]. Tsimane’ life was substantively altered when, during these years, the confluence of several significant events began to break these past ontological ties. Their symbolic climax was the loss of the last cocojsi, which signaled a deterioration of the strong connections between religious beliefs, traditional knowledge of healing and social order. These changes, however, were long foreshadowed by the increasing influence of Christian missionaries, western medicine, ranchers and logging companies in the area [53]. Part of what precipitated this process was a new road built through the territory–usable mostly during the dry season–that brought increasing numbers of new settlers. Unlike many nearby indigenous groups, the majority of Tsimane’ had resisted religious conversion up until this period by escaping into the forest upon confrontation or conflict with outsiders. With increasing acceptance of Evangelical and other missionaries came increased access to western education and biomedicine as well as more permanent settlements in which schools could be established [53]. Missionaries and logging companies also aided the Tsimane’ in the establishment, recognition and support of the Gran Consejo Tsimane’, which now serves as the official autonomous polity of the group in all interactions with the Bolivian federal government and with those seeking access to territorial resources.

Change in social order and exposure to outsiders brought new opportunities for political, religious, skills-based and monetary wealth status. Members of the Gran Consejo and corregidores (elected community magistrates) gained novel political positions, new school teachers gained novel skills and regular incomes, households selling surplus crops gained economic clout, and Christian lay pastors took up the mantel of spiritual leadership. By the mid-2000s federally funded benefits for mothers, children and the elderly became available–contingent on officially registering with the government and obtaining approved identification–further disrupting social hierarchies by introducing supplemental resources for subgroups who had previously been some of the most vulnerable and dependent [56]. Despite these changes, continuity remains in much of Tsimane’ life, including patterns of intergenerational wealth transfer and egalitarianism. For example, Hooper and colleagues [57] found that Tsimane’ parents, grandparents and siblings continue to provide significant downward transfers of food across generations but that the extent of provisioning depends on household circumstances such as productivity and the demographic composition of families. They also determined that participation in culturally meaningful social activities, like drinking shocdye’ (beer brewed with locally available grains) among men tended to result in larger social networks (i.e. relational wealth) and increased social status.

Reyes-García and colleagues are the only previous researchers to explicitly investigate a consensus measure of lifestyle among the Tsimane’ [58,59] but their research faced a unique measurement challenge that may have affected the accuracy of results. This occurred previous to construction of lifestyle rating questions and consensus analysis, during the elicitation phase for lifestyle items. In particular, researchers asked study participants to free list components of ‘a good Tsimane’ lifestyle’ rather than ‘a good life in the community.’ What is typically the gold standard for this type of exercise (i.e. explicitly specifying the cultural group of interest), in the case of the Tsimane’ may have misled participants. This is because among the Tsimane’ the use of “Tsimane’” as a qualifier, including the common translation of chätidye’ (a Tsimane’ word with a meaning similar to relative or relation in English), is often conflated with a description of living in an outdated way. Therefore, it is unclear if their lifestyle model documents a modern, outdated or mixed form of lifestyle [19]. Keeping this in mind, Reyes-García and colleagues reported the following five items as most important to lifestyle (in descending order): to visit kin, to spend time with close family, to have good food, to have a good agricultural plot and to be visited. Three of these items are concerned with relational wealth and two with producing embodied wealth. Among an additional 791 respondents, researchers then tested the cultural model of lifestyle against an existing dataset of behavioral measures to estimate cultural consonance. Cultural consonance measures how well a person approximates the prototype of lifestyle in their own beliefs and behaviors. It has been associated with improved health and well-being across several populations [60] but this expectation was not reflected in Reyes-García and colleague’s study. Instead, there found only mixed evidence for salubrious effects from lifestyle consonance. More research on modern Tsimane’ lifestyle is needed to verify and clarify the existing research.

Social order and social hierarchy among Tsimane’ men

For Tsimane’ men, knowledge of new ways to make a living and higher rates of interaction with outsiders have begun to spur a livelihood transition [61]. Traditional livelihoods are now commonly supplemented by non-traditional means, the majority of which are physically demanding (e.g. farming crops for surplus, logging on contract, or wage labor like clearing jungle for ranchers) and involve substantial risks (e.g. crop-loss, physical injury, legal peril or employers who refuse to pay) [55]. As compared to women, who are the primary caretakers of children, men’s status is tightly tied to their ability to provide for their household’s needs and to support broader familial and social networks to maintain relational wealth. This has made Tsimane’ men an obvious target for investigations of social status.

Among the two principal groups of researchers who have studied social status explicitly among the Tsimane’, both have principally focused on men. The first group is led by von Rueden and the other by Reyes-García [18,62–65]. von Rueden and colleagues [18] first studied relative social status among 57 Tsimane’ men as ascribed by third-party male evaluators using four measures of perceived and functional status (i.e. respect, community influence, gets way in a group and wins dyadic fights) drawn from published theory and research. They applied their method in a relatively market-integrated Tsimane’ community, Ton’tumsi, and found that men with more traditional skills were conferred a qualitatively different kind of social status, respect, than those with non-traditional skills who instead possessed higher community influence. They also determined that social support best predicted getting one’s way in a group. These results indicate that increased community market integration may influence the kind and function of status acquired via different kinds of wealth. In another study [65], Tsimane’ communities closest to a market town were found to have higher population densities and higher levels of inequality in informal political influence between residents, suggesting a role for market integration in increasing the severity of relational wealth inequality between men. Finally, in a sample of 199 Tsimane’ men from four communities [66], higher incomes were associated with higher levels of objectively-measured stress, while peer-evaluated status was associated with less stress. This suggests that there may be trade-offs for pursing monetary and material wealth over relational wealth among Tsimane’ men. That is, in these four communities higher monetary wealth was correlated with relatively less embodied wealth (more stress), while increased relational wealth was correlated with relatively more embodied wealth (less stress).

Reyes-García and colleagues also studied social status among Tsimane’ men. They conducted a study on social rank [62] using third-party nominations of ‘important people’ in the community to estimate community social rank for 289 Tsimane’ men from the same community. Social rank was found to correlate with three separate measures of nutrition, indicating another possible link between relational wealth and embodied wealth among Tsimane’ men.

Social status and social hierarchy among Tsimane’ women

Most of what we know about the determinants of social status and social hierarchy among Tsimane’ women is based on ethnographic accounts. For example, Tsimane’ women spend the majority of their time either working at home, taking care of children or foraging and farming near the household (often accompanied by children) [54]. Within the community, women exercise their power as sources of social information and cultural knowledge [67] but rarely hold modern political positions, such as corregidor, or occupy new high status skills-based roles such as school teacher. As is common in many societies, gender in Tsimane’ society plays a determinant role in social status including disadvantages for women reflected in high rates as reported victims of intimate partner violence and a lack of equality in opportunity for new income-generating strategies [64,68]. Common barriers to women’s market participation include social expectations that discourage work with outsiders and male peers, difficulty in taking time away from family obligations for short-term migrations, lower levels of education and fewer Spanish language skills. These barriers represent conflicts that originate from the traditional role of women in Tsimane’ society and when combined with discrimination against Tsimane’ by other ethnic groups, creates a double burden of gender- and ethnic-based inequity that limits access to, and acculturation into, Bolivian society [69,70]. These conclusions are supported by a new comparative study by von Rueden et al. [71] that found women at a disadvantage in community-level influence linked to body size, education, and social networking. The authors propose that each of these links act as status determinants for both genders but that women are at a disadvantage in acting on them. For example, the male advantage in body size is mostly pre-determined, while cultural expectations for in-home labor limits girl’s access to schooling and a limited number of prescribed social strategies are open to women that would facilitate maintaining a large social network. Expanding our knowledge of status determinants among Tsimane’ women will help to clarify the influence of economic transition on all of Tsimane’ society.

Ethical considerations

Research design and all informed-consent and data collection procedures were approved by the Institutional Review Board 02, University of Florida (IRB-02-U-341-2011) and the autonomous authority of the Great Tsimane’ Council. Due to high rates of illiteracy among study participants, a waiver to written consent was granted for the study and verbal consent was obtained from all participants.

Study design

Below, I present detailed descriptions of sampling and measurement methods for each of the phases of research preceded first by a brief summary of the general study design. Prior to the start of this research, I completed extensive participant observation in each of the two study communities over a period of one year [72]. The research described herein began with two iterative phases of data collection and analysis to build and test a cultural consensus model of lifestyle [73] including free listing activities and ratings exercises. The analysis of free listing results was used to develop the ratings exercises, which were then piloted and administered in a subsequent phase. Once consensus analysis of the ratings was complete, I performed two rounds of regression analyses on the loadings for the first and second factors of the lifestyle model to 1) identify possible subgroups within either factor of lifestyle and 2) assess the presence of social incongruity between aggregate subgroup preferences and status determinants (i.e. age, gender, years of education and relative material wealth). Next, to aid in interpretation of the consensus and regression results, I plotted deviation scores between subgroups’ aggregate loadings on lifestyle model items and the ‘ideal’ consensus answer key ratings as calculated in consensus analysis of the entire study sample. Lastly, using a truncated set of lifestyle items (for ease of interpretation), I assigned themes to plots of subgroup item deviation scores based on previous ethnographic data, published literature and input from Tsimane’ translators.

Sampling and variables

Aided by two Spanish-Tsimane’ translators (Ignacio Huasna and Orlando Durbano Tayo), I collected all of the study data in two Tsimane’ communities (Fig 2) located roughly two-hours (Ají) and two-days (Serrucho) travel time, respectively, from the nearest market town. All interviews occurred during 2012–2013 and were given in the local language (Tsimane’) with a translator present to ensure accuracy and ease of communication. The two study communities were selected to complement one another based on 1) average travel time from the nearest market town as a proxy for level of market integration 2) sufficient size to facilitate the purposive life stage sampling method and 3) ethnic makeup (>95% Tsimane’ in each community) to help ensure an unbiased elicitation of the cultural lifestyle model without needing to qualify lifestyle as ‘Tsimane’ lifestyle.’ I chose travel time to the market town of San Borja as the study proxy for level of market integration, as it has proved reliable in past studies [56]. From Ají, travel one-way to San Borja on clear roads takes 50 minutes and costs $2–5 USD. For residents of Serrucho, travelers must first canoe (8–16 hours motorized, $5–10 USD) or hike (two days) to reach the highway near Ají before continuing on to San Borja.

To maximize variability in potentially important status determinants related to market integration, I used life stage purposive sampling based on age groups, gender and relative material wealth (Table 1) [74]. I limited the range of participants in the free list exercises to Tsimane’ adults aged 16–70 years in order to maximize comprehension of the activity and reliability of the responses. I expanded this range to 15–75 years for the ratings exercises, which are usually easier for participants to comprehend. I conducted free list exercises and ratings surveys with 32 and 69 adult respondents, respectively. The ratings subsamples represented 30.4% and 21.6% of the eligible populations in Ají (n = 115) and Serrucho (n = 157), respectively, based on contemporaneous community censuses [72]. Age, education, gender and ethnicity were assessed via self-report and verified against government identification cards (when available) and with community records kept by local officials. Due to highly variable incomes and commonly shared resources, I used a dichotomous measure of relative household material wealth, rather than recent income, to assess financial wherewithal. Material wealth was ascribed as more or less wealthy than average according to the mean of three third-party assessments by senior community members who ranked household wealth based on “the money this household has, the things they own and the things they can afford to buy, like food, clothing, toys, tools, and weapons.”

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Purposive life stage samples^a.

https://doi.org/10.1371/journal.pone.0220432.t001

Free list and rating exercises

With each respondent, I used the same free list elicitation, which translates as “List all of the people, items and activities a person needs to live a good life in your community.” When, on occasion, respondents did not readily name at least 20 items, I used follow-up prompts for additional ‘people,’ ‘items’ and ‘activities’ most needed for a good life. A typical free list interview lasted 20–40 minutes and I recorded all items in a notebook. Later, I combined free lists into a database and Tsimane’ translators aided in confirming and condensing similar responses into composite elements (study data is available at OPEN ICPSR http://doi.org/10.3886/E104423V3).

Of note, in this study, a pile sort activity–typically used to elicit how knowledge is structured within a cultural domain–was excluded from development of the rating exercises due to a poor level of observed comprehension among respondents during pilot exercises. This was likely due to the exclusive use of pictures to represent complex lifestyle items for piles sorts, without accompanying written descriptions–an approach necessary due to widespread illiteracy.

Ratings questions were based on 38 items (37 from the free list results and isätri, a charm used when hunting, drawn from participant observation). The ratings question template translates as, “How much does someone need _____ to live a good life in your community?” on a scale from 1 (“They don’t need them/those at all”) to 5 (“They need them/those very much”). Exercises included the 25 most-mentioned items (listed by >50%), six medium-frequency items (listed by 50–25%), six low-frequency items (listed by <25%) and one unmentioned but observed item, isätri. I chose to include isätri because it was frequently a topic of discussions about hunting in Serrucho, but not Ají. The inclusion of high-, medium- and low-frequency free list items in ratings exercises has been shown to improve domain measurement when there is potential for little variation in ratings across high frequency items [75]. Pilot ratings and the final results of this study bear this out, with the average rating of the 20 top-rated lifestyle items ranging between 4.5–5.0/5.0. I also used reversed ratings for half of the questions after detecting automatic acquiescent response bias during pilot ratings exercises (i.e. “yea-saying” or respondents giving all high ratings often due to boredom, predictability of answers or social desirability bias) [76]. Contrary to previous research on Tsimane’ lifestyle [59,77], I did not quality lifestyle as ‘a good Tsimane’ lifestyle’ in the final ratings exercises because during pilot tests participants often misunderstood a ‘good Tsimane’ lifestyle’ to mean “in the time of the grandparents,” as one respondent stated–an outdated way of life. Instead, I asked about “a good life in the community,” to emphasize modern cultural perspectives within communities populated by >95% self-identified Tsimane’. Ratings exercise were conducted among a fresh sample of participants separate from respondents who completed free list exercises.

Consensus and regression analysis

I conducted consensus analysis of the ratings exercises using Analytic Technologies’ UCINET 6 software [78], which provides both formal and informal versions of consensus based on different analytic assumptions and data types. The informal model uses reliability analysis and has the advantage of allowing the use of interval and ranked response data while the formal model relies on a mathematical model. The validity of results from these two approaches has been demonstrated to vary little when analytic expectations such as number of items, congruity of content, sample size and sample diversity are met [73]. Consensus analysis is similar to a factor analysis except inverted to examine participants’ lifestyle knowledge scores, rather than individual lifestyle items. It produces four key outcomes, including 1) individual cultural knowledge scores (0–1.0), 2) model goodness-of-fit, in the form of an eigenvalue ratio between loadings on the first and second factors (>3:1 is the minimum ratio recommended for a determination of consensus), 3) a cultural consensus answer key that estimates the ‘ideal’ or prototypical rating of each item/characteristic in the model based on the combined, knowledge-weighted item ratings from all respondents and 4) residual agreement coefficients (±) which load onto the second factor of the lifestyle model and can indicate membership in a subgroup that diverges in one direction or another from the overall cultural consensus.

Following consensus analysis, I used the loadings on the first (i.e. knowledge scores) and second (i.e. residual) factors of lifestyle to test for associations with subgroups using linear regressions [25]. For each subgroup found to be associated with loadings on a factor, I then calculated item-specific subgroup deviation scores for every item. Item deviation scores are an extension of a core concept in consensus analysis: that the ratings answer key represents the ideal emic ratings for the cultural domain examined. Deviation scores build on this by allowing comparisons at the subgroup level, which is done by calculating the relative difference (±) between the consensus answer key and the average item ratings among all members of a subgroup. Based on the results of regression analyses, I calculated deviation scores for gender subgroups associated with the first factor loadings and community/market integration and wealth subgroups associated with the second factor loadings, following steps outlined by Dressler [25].

Finally, to further explore the meaning behind lifestyle heterogeneity in the second factor, I plotted subgroup deviation scores for the items that varied most as determined by natural breaks in the data (supplementary analysis and the cutoff plots used to determine item inclusion in these sets are available at OPEN ICPSR http://doi.org/10.3886/E104423V3). Briefly, one-dimensional plots of absolute item deviation scores were used to establish natural cutoffs when irregular gaps occurred in the plots. For the gender subgroups, the items within the cut-off represent the highest 15% (n = 5 of 38) of absolute deviations from the consensus answer key, while for community subgroups and wealth subgroups the items within the cut-offs represent the highest 35% and 30%, respectively, of lifestyle items (n = 11 and n = 13). To ensure the representativeness of these subsets for gender-, integration- and wealth-based subgroups, I reanalyzed each data set to look for substantive alterations in meaning after both adding and subtracting ~10% of total lifestyle items. This step produced no meaningful differences from the original analyses. Lastly, I analyzed and categorized the resulting plots for common themes based on insights drawn from ethnography and consulted Tsimane’ research assistants to ensure emic validity.

Market integration and cultural lifestyle

The consensus analysis of lifestyle items demonstrates a level of sharing within the first factor sufficient to conclude moderate consensus (Table 4; Eigenvalue ratio 4:1) [79]. The first factor accounts for 72.7% of sample variance and the average knowledge of raters was 0.63 (± 0.12). Recommended cutoffs for the standard deviation statistic produced by consensus analysis using the informal model have not been published. However, the deviation statistic did meet the cutoff for a formal analysis model, (≤0.14 for this sample size and number of questions) as proposed by Hruschka and Maupin [80–83]. Table 4 lists consensus results in each community and combined.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Cultural consensus in the first factor (‘market lifestyle’) in Aji, Serrucho and combined.

https://doi.org/10.1371/journal.pone.0220432.t004

In answer to the first research question, I find that Tsimane’ respondents maintained consensus in a cultural model of lifestyle despite the presence of some heterogeneity. Based on the model contents and ratings for the first factor, as well as the results of residual analyses described below, I label the cultural consensus model with its generally higher ratings for market-based items, market lifestyle. By contrast, I label residual agreement in the second factor syncretic lifestyle because of the nature of its comparative ratings, which are closer to parity for market-based and traditional items implying more of an amalgamation of lifestyles. The finding of subcultural heterogeneity is consistent with many published results from consensus analyses that regularly include residual agreement in the second factor, especially in contested or changing cultural domains [25]. Fig 3 is a plot of the loadings on the first and second factors of lifestyle and shows a clear demarcation between respondents from the study communities along the y-axis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Scatterplot of lifestyle knowledge (factor 1) and lifestyle (factor 2).

Markers sized by age; community trend lines match label colors; dotted line indicates zero on the x-axis.

https://doi.org/10.1371/journal.pone.0220432.g003

Lifestyle variation across subgroups and market integration

Table 5 displays the results of regression analyses of market lifestyle and syncretic lifestyle on explanatory variables. In response to the second research question, I find that women (β = .342) tend to be more knowledgeable of the market lifestyle prototype than men while no other explanatory variables are even weakly associated. Fig 4 demonstrates this difference among respondents in a non-parametric LOWESS (Locally Weighted Scatterplot Smoothing) of market lifestyle knowledge against age and stratified by gender. There is a higher level of lifestyle knowledge among women across stages of the life course and the plot also suggests that younger women have the keenest knowledge of market lifestyle, though this is not supported by the regression results. Fig 5 includes the top quartile of items for which market lifestyle ratings of men and women most diverged. Men varied most in their higher ratings for items that contribute to relational wealth, including non-Tsimane’ friends, large family size, non-family helps, social tobacco smoking, and gathering to listen to flute music. The only exception to this pattern was men’s higher ratings for gas stoves, an emically conspicuous material item that is rarely used in communities since there are no accessible sources for the fuel needed to use them. In comparison to men, women’s rating averages only strayed substantially from the market lifestyle prototype with lower ratings for having non-Tsimane’ friends. Other deviations were much smaller.I also find that more relative material wealth (β = –.331) and residence in the less market integrated community of Serrucho (β = .456) are associated with a syncretic lifestyle, supporting the existence of lifestyle subgroups. Education and age did not correlate with loadings on either lifestyle factor.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Nonparametric LOWESS of lifestyle knowledge by respondent gender and age.

https://doi.org/10.1371/journal.pone.0220432.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Largest ratings deviations (upper quartile) from market lifestyle by gender.

https://doi.org/10.1371/journal.pone.0220432.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Regression of market lifestyle and syncretic lifestyle coefficients on explanatory variables (standardized regression coefficients).

https://doi.org/10.1371/journal.pone.0220432.t005

Market lifestyle

Among the top-rated items in Tsimane’ market lifestyle (>4 on a scale of 1–5), over 40% were explicitly market-related (e.g. manufactured items and tools, market-acquired livestock or education). The remaining top-rated items consisted of high-quality versions of basic necessities that are at least sometimes market-related (e.g. good clothes, a good house and good food) and various kinds of relational wealth (e.g. care for children, having family around and help from family). Taken together, the contribution of market items to the lifestyle model appears substantial while still overlapping with some traditional approaches to fulfilling basic necessities and maintaining social customs. When traditional and non-traditional market alternatives both appeared, market items were typically rated higher with the exception of some food items. For example, electronics for entertainment were valued more highly than Tsimane’ flute music or dancing, hunting with a firearm was valued over bow hunting (and over the use of Tsimane’ isätri charms for success in hunting), biomedicine was valued over forest medicine or alternative treatment modalities like the qurpa (see below) and education was valued over living in one’s own village.

Syncretic lifestyle and level of market integration.

Results of residual agreement analysis indicate a more syncretic lifestyle is preferred among respondents from the less market-integrated community of Serrucho. This result complicates interpretation of associations between market integration and lifestyle, since overall both communities agree on a market lifestyle model, but respondents in Serrucho also maintain a set of more traditionalist preferences. Fig 6 plots directional deviation scores (±) for a truncated set of the most divergent items in this subgroup. Twelve of 13 items are rated higher in less-integrated Serrucho while just one is rated higher in Ají. Table 2 also denotes (^†) each of these 13 items along with their ideal market lifestyle rating and adjusted syncretic rating based on subgroup deviation scores. Below, I organize these 13 items into themes and contextualize each according to respondents’ descriptions from the free listing exercises.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Items and characteristics that deviate most (±) from the market lifestyle prototype among respondents at two levels of market integration.

Higher positive deviation scores on the Y-axis and within the grey gradient indicate higher ratings from respondents in less integrated Serrucho. Higher positive deviation scores on the X-axis and within the brown gradient indicate higher ratings from respondents in more integrated Ají.

https://doi.org/10.1371/journal.pone.0220432.g006

First is the category of traditional social and subsistence customs valued primarily for their utility and symbolic power as part of a self-sufficient, small-scale forager-farming lifestyle. This set of eight items was exclusively preferred in less-integrated Serrucho: 1) handmaking and hunting with bows and arrows, a socially-acquired skill that requires extensive training and practice, 2) residing outside of one’s birth village, in order to establish new, or maintain old, social ties with blood relatives and new affiliative family, 3) use of a subcutaneous isätri pebble charm, believed to increase success in the hunt by providing a supernatural advantage, 4) gathering to do ritual Tsimane’ dances, which are believed to build good will among the community, 5) avoiding a small family to demonstrate success, fortify social ties and support a subsistence lifestyle, 6) exchanging help between family members, to support a subsistence lifestyle and accrue social capital/relational wealth, 7) weaving traditional cotton carrying bags, or sáray, used to transport everything from crops to small children for foraging and field work, 8) having a firearm to hunt game, which are still easily hunted in more remote and isolated Tsimane’ communities like Serrucho.

Second, four items that deviate from market prototypes based on level of community market integration can be categorized as traditional spiritual or health customs. That is, each is believed to be protective or curative of the spirit and/or health thereby producing embodied wealth. Three of four items in this category were preferred in the less market-integrated community of Serrucho 1) practicing the annual Väij fruit tree ritual (peach palm, the tree used for bow making and to ferment a seasonal type of homebrew, buri), among household members at first fruiting to protect from natural and supernatural mishaps and reinforce relational wealth, 2) a qurpa (an alum/double sulfate salt), acquired through trade with Aymaran highlanders for divining and curing common illnesses such as cough and fever [84] and 3) use and knowledge of locally-foraged forest medicine, which includes common items such as certain types of tree bark that are ingested to relieve pain and decrease inflammation. The fourth item, planting a shịshi red chili pepper bush outside one’s home, was preferred in more-integrated Ají. Shịshi and their bright red color are thought to offer powerful protection to families from certain kinds of sent sickness and other natural and supernatural dangers. Their exclusive preference in Ají is notable since, on average, residents rated traditional items lower there but prioritized this item to promote embodied wealth.

Finally, the third category is new customs, which indicates items that are novel for the Tsimane’. In this case, having a zinc-plated aluminum roof is a new custom preferred in Serrucho where roofs are traditionally woven from harvested palm leaves. These cajtafa roof panels, however, are no longer a practical choice for personal use in Serrucho after a decade of the community serving as a hub for commercial buyers of roof panels has made cajtafa scarce locally.

Syncretic lifestyle and material wealth.

To better understand variation in subcultural preferences for a syncretic lifestyle by relative household material wealth Fig 7 plots the 11 items within the cut-off for the largest absolute deviation scores from the market lifestyle prototype. Ten of 11 items are favored by participants with less wealth, while just ‘non-Tsimane’ friends’ is favored by participants with more wealth. Table 2 also denotes (^††) these 11 items alongside their consensus answer key ratings and adjusted syncretic answer key rating based on subgroup deviation scores. Below, I organize the items into three themes and contextualize each based on respondents’ explanations from the free listing exercises.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Items and characteristics that deviate most (±) from the market lifestyle among respondents in dichotomized wealth groups.

Higher positive deviation scores on the Y-axis and within the grey gradient indicate higher ratings from those with more wealth. Higher positive numbers on the X-axis and within the brown gradient indicate higher ratings from those with less wealth.

https://doi.org/10.1371/journal.pone.0220432.g007

First, is the category of traditional social and subsistence customs (see description above), which includes six items that were rated more highly by respondents with less material wealth including, 1) hunting with handmade bows and arrows, 2) gathering to do ritual Tsimane’ dances, 3) use of an isätri pebble to bring success in the hunt, 4) having dogs that help deter thieves from the homestead and aid in hunting, 5) exchanging help between family members and 6) weaving sáray carrying bags. The second category is traditional spiritual and health customs. All three of the items in this category were rated higher than the market prototype among respondents with less wealth including 1) practicing the seasonal Väij fruit tree ritual, 2) use and knowledge of forest medicine and 3) qurpas to aid health. Finally, the third category new social customs describes lifestyle characteristics respondents consider helpful in their attempts to increase household income or find work outside of the community. Non-Tsimane’ friends is the only item in this category and it was rated more highly by respondents with greater wealth who considered these connections a form of relational wealth that supplies the information needed to find work and increase material wealth.

In answer to the third research question, I find that the patterning of lifestyle by gender (differences in market lifestyle) and material wealth (residual agreement in syncretic lifestyle) indicates that the domain is contested due to conflicts between traditional priorities and market priorities. I also find that the patterning of syncretic lifestyle by level of market integration is similar to the patterning of lifestyle by subgroups of more- and less-wealthy respondents, including the two central themes in the analysis of deviations, traditional social and subsistence customs and traditional spiritual and health customs. Taken together, this evidence supports the tentative conclusion that market integration has differentially impacted the modern Tsimane’ prototype of lifestyle. The first factor of lifestyle, in particular, explains the vast majority of variance in participant lifestyle knowledge and can be understood to largely favor market alternatives. That said, gender was the only status determinant associated with this factor of lifestyle knowledge. For the second factor of lifestyle, both relative wealth and market integration were associated with factor loadings but overall these loadings accounted for less than a fifth of the total consensus analysis model variance. Future research is needed to determine definitively whether the conflict between men and women regarding lifestyle knowledge, or the existence of wealth- and integration-based subgroups indicated by residual agreement analysis, represent social incongruities that can help to predict wealth inequality and well-being among the Tsimane’.

A market lifestyle among the Tsimane’

The cultural lifestyle model I describe in the results is a mix of traditional and market components with clear preferences for non-traditional market items in most circumstances where a market alternative exists. Comparison of prototypical ratings for items in the market lifestyle model and adjusted ratings for the syncretic subcultural model (see Table 3) help to demonstrate this difference. For example, in preferences for firearm hunting versus bow hunting, biomedicine versus forest medicine or market supplies/tools versus weaving sáray cotton bags. Among less market-integrated residents of Serrucho, who were more likely to adhere to the syncretic model than those from Ají, the average rating for use of the qurpa in healing was 0.66 higher (4.58/5.0) when adjusted for ratings deviations, fully 96% of the score for the use of biomedicine (4.78). Also, the difference in ratings between modern firearm hunting and traditional bow hunting is smaller among respondents from Serrucho (bow hunting is rated 3.94 and firearm hunting is rated 4.86) while ratings for all hunting are higher compared to the market lifestyle model. Only a few market lifestyle items rated in the final 38-item modelling did not have a traditional alternative. The existence of so many market alternatives is itself noteworthy as a possible indicator of change that has already occurred and supports the use of lifestyle as appropriate for investigating market influence among the Tsimane’.

The market lifestyle model reported builds on work Reyes-García and colleague’s published on Tsimane’ lifestyle [58,59] that elicited items for a ‘good Tsimane’ lifestyle’ to assess consensus in the domain. In comparing the top ten highest rated lifestyle items between this study and that, there are clear differences in market and traditional items. Here, more market-related items are in the top-ten including livestock, market supplies/tools and education, while the Reyes-García et al. model contains just one market item, ‘to acquire commercial goods.’ There are a couple of ways to interpret this difference. First, it could mean that the Reyes-García et al. study reflects a different and less market-integrated period of time (or sample of respondents). Depending on the speed of change, this is possible. Or, it could be the result of a measurement difference either due to qualitative choices made to include/exclude lifestyle ratings for elicitation exercises or, as described above, from bias introduced during elicitation of lifestyle items that altered some participant responses from modern to traditional lifestyle.

Social incongruity in gender and knowledge of market lifestyle

On average, women had 13% more accurate knowledge of market lifestyle prototypes compared to men. To explore heterogeneity in market lifestyle by gender, Fig 5 charts the relative deviation scores for the nine items that varied most between men and women. These deviations are principally driven by men’s higher ratings for relational wealth items including the social use of homegrown tobacco, social gatherings based on listening to traditional flute music and items concerned with the maintenance of social ties to family, non-family and outsiders [88,89]. The results reflect a similar pattern to von Rueden and colleague’s findings on relational wealth among Tsimane’ men in which prosociality proved important because it tended to result in more social support in the community [18]. Independently, they also found that acculturation predicted social support in the community, a potential motivator for men’s higher ratings here for items related to relational wealth. The findings may also be indicative of a social incongruity between gender and relational wealth. In fact, other researchers have already documented the asymmetric costs to women of gendered social hierarchy in Tsimane’ communities [90]. In this case, the social incongruity between gender and lifestyle items related to relational wealth might pose a threat to women’s future social mobility. In particular, the devaluing of market-related kinds of relational wealth during the early-stages of market integration, could exacerbate existing disparities in women’s opportunities to participate in the market economy on top of existing barriers to education access, travel outside of the community and language skills. During the ethnographic phase prior to the current study, participants also reported that strict gender roles were being policed via the belief systems proselytized by Evangelical missionaries in the area. More research is needed to explain why women in this study displayed more knowledge of market-oriented lifestyle, whether it is related to a desire for greater autonomy or social mobility and what types of behaviors best predict their desired outcomes [91]

An important caveat to the interpretation of these gendered study results is that knowledge is not the same as achievement, and achievement of lifestyle was not measured explicitly in this study. Achievement of lifestyle was measured, however, in the two studies by Reyes-García and colleagues mentioned above. In their studies, only two small gendered differences were found in consonance measures of lifestyle, with associations between more consonance in men and larger body size, and more consonance in men and higher body mass index [59]. Both of these associations are indicative of a relationship between the capacity to adhere to cultural lifestyle and embodied wealth. However, depending on the local interpretation of Reyes-García et al.’s elicitation, these results may reflect modest embodied wealth benefits for consonance with a modern, mixed, or traditional form of Tsimane’ lifestyle.

Future research should test how the knowledge advantage among women detected in this study compares to their achievement, or consonance, in lifestyle. Adding to our knowledge of lifestyle incongruities among Tsimane’ women will help to clarify the influence of economic transition on the whole of Tsimane’ society, eliminate gender as a possible source of omitted variable bias in assessments of social status and help determine whether gendered incongruities drive wealth inequalities.

The patterning of syncretic lifestyle

The multilevel and contextual variability of most social determinants makes accurate measures of social status and social incongruity particularly challenging to capture in diverse populations. For example, in Gravlee and colleague’s research [33], color incongruity was found to only affect participants who met the local racial criteria for being negro in Puerto Rico, a circumstance largely beyond personal control, yet personally and contextually specific to the setting. Similarly, Sorensen et al.’s research on the Yakut in Siberia [38] found that only those who lacked material wealth but strived for higher SES suffered from high rates of psychosocial stress. In this study, while health outcomes were not evaluated, it is apparent that respondents from the less market integrated community, Serrucho, and respondents with less material wealth, preferred a syncretic lifestyle with a more balanced set of traditional and market priorities. Further evidence of this variation is apparent in Table 4, which displays results for intra-community consensus analyses of market lifestyle. Compared to the aggregate lifestyle consensus these intra-community models show much higher agreement for market lifestyle in Ají compared to Serrucho, a possible effect of higher market exposure in Ají.

Still, identifying possible motives for market integration is difficult. For example, wealth among the Tsimane’ has been estimated to exceed that of non-Tsimane’ residents of rural lowland Bolivia when non-monetary wealth is included in estimates [92]. Combined with resource distribution expectations that result in significant downward transfers of food across generations in families [57], the evidence implies a generally high quality of life that is not likely to be a strong driver of change in livelihood practices [41]. Instead, what is more likely is that many households have been enticed into change little-by-little via exposure to modern time- and effort-saving technologies that complement traditional livelihood practices and patterns of wealth distribution in situ. An example of this process is the rapid adoption of peke-pekes. These efficient and affordable outboard canoe motors (documented here as part of the market lifestyle item ‘market supplies and tools’) have been rapidly transforming river transportation in Tsimane’ territory during the last decade. In combination with chainsaws and other modern tools, they have also greatly reduced the effort it takes for a household to expand from subsistence farming into surplus farming for profit since they facilitate easy access to far-off fields. The adoption of peke-pekes, then, represents an elective transformation with mixed motivations but whose collective effects may go well beyond immediate household goals of increasing monetary income and material wealth.

Within this context of change during the last two decades, the Tsimane’ territorial population has been increasing rapidly, mostly due to a combination of improved childhood survival rates and continued low rates of out-migration despite tightly-bounded land on which to settle [93]. Together, these patterns threaten to create a novel strain on Tsimane’ household resources, but especially among those who wish to maintain a semblance of a traditional subsistence livelihood while facing increasing difficulties in securing the space and cropland needed. It is within this context of overlapping changes that the subcultural model of syncretic lifestyle must be understood. While smaller in its overall explanatory power compared to market lifestyle (18.2% model variation explained versus 72.7%), it represents systematic variation in the lifestyle domain that parallels traditional Tsimane’ beliefs. In this way, it can help to pinpoint divisions between subgroups at different levels of market integration and those with relatively more and less wealth. Here, I explore the meaning of this patterning in light of the published research on Tsimane’ social hierarchy and social status.

Traditional livelihood and wealth preferences

The study findings help to advance our understanding of livelihoods and wealth at the early-stages of market integration. Specifically, the market priorities of modern Tsimane’ lifestyle indicate a departure among study participants from a traditional lifestyle that favored livelihoods which were more focused on relational and embodied wealth than material wealth. This conclusion is based on the residual agreement analyses above that demonstrate a minority preference for a more traditionally-oriented lifestyle of subsistence livelihoods and relational forms of wealth. It is also reflected in the gendered social incongruity between market lifestyle and knowledge of specific relational wealth items. The devaluing of relational wealth among women in a seeming trade-off between relational wealth and material or embodied wealth, may signal increasing wealth inequalities between Tsimane’ men and women. Future research should test this hypothesis with longitudinal data.

Limitations

The research described has several limitations. First, rigorous hypothesis testing of change requires a longitudinal design such as a cohort or panel study. In lieu of such a design, cross-sectional results must be interpreted with caution as they may reflect unanticipated confounds and cannot determine cause and effect owing to a lack of multiple measurements. Additionally, although the sample sizes I use for the chosen research design meet agreed upon standards, they are limited in geographic- and population-scope having been collected in just two communities. Another specific concern is the borderline significance (p < 0.056) for the regression model that tested loadings on the first factor of the lifestyle model, likely due to a small sample size. Also of note, is the limited number of variables (five) tested to examine status determinants against lifestyle. Combined with the limited number of highly-rated domain items included in consensus analysis (25) due to the limitations of the method, the consensus analysis may have suffered from some amount of omitted variable bias. Finally, as discussed in the methods section, I was not able to use respondent pile sorts to help organize lifestyle items into emic themes due to high illiteracy rates. I instead based themes on insights from months of ethnographic participant observation. Such post-hoc categorization of consensus data runs the risk of overlooking meaningful distinctions due to gaps in the ethnography. To minimize this risk, I also reviewed and modified themes with long-time Tsimane’ research assistants to confirm their emic face validity.