1. Epidemiology and Global Health
  2. Evolutionary Biology
Download icon

Productivity loss associated with functional disability in a contemporary small-scale subsistence population

  1. Jonathan Stieglitz  Is a corresponding author
  2. Paul L Hooper
  3. Benjamin C Trumble
  4. Hillard Kaplan
  5. Michael D Gurven
  1. Université Toulouse 1 Capitole, France
  2. Institute for Advanced Study in Toulouse, France
  3. Economic Science Institute, Chapman University, 1 University Drive, United States
  4. Center for Evolution and Medicine, Life Sciences C, Arizona State University, United States
  5. School of Human Evolution and Social Change, Arizona State University, United States
  6. Department of Anthropology, University of California, Santa Barbara, United States
Research Article
  • Cited 0
  • Views 270
  • Annotations
Cite this article as: eLife 2020;9:e62883 doi: 10.7554/eLife.62883

Abstract

In comparative cross-species perspective, humans experience unique physical impairments with potentially large consequences. Quantifying the burden of impairment in subsistence populations is critical for understanding selection pressures underlying strategies that minimize risk of production deficits. We examine among forager-horticulturalists whether compromised bone strength (indicated by fracture and lower bone mineral density, BMD) is associated with subsistence task cessation. We also estimate the magnitude of productivity losses associated with compromised bone strength. Fracture is associated with cessation of hunting, tree chopping, and walking long distances, but not tool manufacture. Age-specific productivity losses from hunting cessation associated with fracture and lower BMD are substantial: ~397 lost kcals/day, with expected future losses of up to 1.9 million kcals (22% of expected production). Productivity loss is thus substantial for high strength and endurance tasks. Determining the extent to which impairment obstructs productivity in contemporary subsistence populations improves our ability to infer past consequences of impairment.

Introduction

As the longest living and slowest growing primate, humans experience unique physical impairments with potentially large economic and social consequences. Human hunter-gatherers have lower adult mortality than chimpanzees and a significant post-reproductive lifespan (Gurven and Kaplan, 2007; Hill et al., 2007; Kaplan et al., 2010; Kaplan et al., 2000; Wood et al., 2017), which generates ample opportunity for incident degenerative disease and dysfunction (Finch, 2010). Evidence of greater dietary reliance by hominins on difficult-to-acquire resources including hunted game at least two mya (Aiello and Wheeler, 1995; Antón et al., 2014; Thompson et al., 2019; Ungar, 2012) suggests delayed peak skill acquisition in adulthood (Gurven et al., 2006; Kaplan et al., 2000; Koster et al., 2020; Walker et al., 2002), possibly when senescence was already underway. Functional disability of degenerative origin may hinder hominin foraging in costly ways. These costs amplify if disability hinders resource transfers or provisioning of other assistance, given their potential fitness impacts (Gurven et al., 2012; Hawkes, 2003; Hill and Hurtado, 2009; Hooper et al., 2015; Marlowe, 2003; Schniter et al., 2018; Wood and Marlowe, 2013). The ubiquity among hunter-gatherers to form social groups of clusters of multi-generational resource-pooling units (Kaplan et al., 2009; Migliano et al., 2017), and in base camps at least ~400 kya (Kuhn and Stiner, 2019), may reflect a species-typical strategy of complex cooperation to minimize daily risks, including production shortfalls. Quantifying the cost of disability in terms of productivity loss in extant small-scale subsistence populations is thus critical for understanding the strength of selection pressures for cooperative strategies that minimize risks associated with disability.

Skeletal evidence suggests that functional disability of degenerative origin may not have been that uncommon throughout hominin evolution, and that participation in activities essential for survival and reproduction (e.g. bipedal locomotion, food acquisition, and load carrying) may have been constrained by disability. For example, marked thoracic kyphosis and vertebral disc degeneration are evident in Australopithecus afarensis (AL-288) from ~3.2 mya (Cook et al., 1983), as is spondylolisthesis in middle Pleistocene humans (SH-1) from ~430 kya (Bonmatí et al., 2010; also see Trinkaus, 2018). These ailments regularly cause lower back pain, difficulty walking, and carrying or lifting objects among modern clinical patients (Francis et al., 2008), who may provide indirect insights into the consequences of disability in past hominin populations. Given the variability across hominin species in terms of environmental exposures, diet, life history traits (e.g. lifespan), morphology (e.g. body and brain size) and locomotion from ~4 mya until 40 kya (Wood and Boyle, 2016), whether and how these impairments hindered survival and reproduction is unclear, and not readily discernable from any single hominin species.

Joint behavioral and epidemiological study of contemporary small-scale subsistence populations suggests that diverse impairments of degenerative or other etiology hinder resource production and transfers. Among Yora forager-horticulturalists of Peru, men are unable to forage due to illness or injury on 11% of all days (Sugiyama and Chacon, 2000). Among Shiwiar forager-horticulturalists of Ecuador, >60% of individuals experience an impairment (e.g. chronic pain, fracture, laceration, infection, animal bite or sting, and burn) severe enough to interfere with subsistence work for ≥1 month; bone fractures – a focus of the current study – may be more likely than other impairments to cause prolonged disability (Sugiyama, 2004). Among Tsimane forager-horticulturalists of Bolivia, the population studied here, 75% of adults report being bedridden due to illness or injury at least once in the 3 months prior to survey, and Tsimane are incapacitated on about 10% of all days (Gurven et al., 2012). The fact that these studies include both younger and older adults (aged 20+ years) indicates that prolonged disability is not conditional on reaching advanced ages.

While providing evidence that disability may have important consequences, the observational studies of contemporary small-scale subsistence populations mentioned earlier are limited: they rely mainly on self-reports and thus cannot identify objective, specific somatic impairments associated with diminished subsistence effort; they utilize crude measures of subsistence involvement (e.g. able to work or not) and thus cannot quantify the magnitude of productivity loss or ascertain the range of subsistence behaviors associated with disability; they often rely on small sample sizes of adults, including adults in their peak productive years; they do not consider potential for bidirectional associations between somatic condition and economic production; and they are not designed to isolate specific mechanisms linking somatic condition and productivity. Precise assessment of the economic cost of disability and consideration of potential compensatory strategies to mitigate disability-related productivity loss is important in light of the comparatively low mortality rates characteristic of human life histories (Gurven and Kaplan, 2007). Studies of contemporary small-scale subsistence populations can thus complement bio-archaeological and paleoanthropological studies for understanding factors affecting foraging behavior and resilience in past hominin populations. In addition, in populations lacking formal institutional mechanisms that minimize risks of economic insecurity (e.g. workers’ compensation and disability insurance) studies of the economic cost of disability are important for understanding relationships between security, sociality, and well-being. As subsistence populations like the Tsimane increasingly participate in local market economies over time, studies of the economic cost of disability may also help inform policies designed to provide social security benefits.

Hominin social organization in multi-generational, multi-layered cooperative units are believed to be a necessary precursor for lowering mortality rates (Bird et al., 2019; Hawkes et al., 1998; Hill et al., 2007; Isaac, 1978; Kaplan et al., 2000; Washburn and Lancaster, 1968). The fact that many subsistence skills require cumulative knowledge that is often transmitted from older to younger generations implies that functionally disabled adults can maintain ‘fitness value’, despite perhaps compromised ability to produce and transfer resources (Gurven et al., 2020b). How functional disability may have influenced behavior is unclear. Some anthropologists have interpreted human skeletal remains showing potential signs of impairment as evidence of compassion and morality of group members, who presumably would have had to support the disabled (see Dettwyler, 1991 and references therein; Tilley, 2015). This interpretation is contentious but supported by observations of modern hunter-gatherers frequently sharing food with impaired adults (e.g. Gurven et al., 2000). Ethnographic reports of death-hastening behaviors among frail elders (e.g. abandonment), who themselves may advocate such behaviors (e.g. Balikci, 1970), suggest that decisions to support the disabled may be complex and influenced in part by their expected future productivity. Quantifying the economic cost of disability is therefore critical for understanding the extent of cooperation that must have co-evolved with hominin life history traits.

While there are diverse factors affecting functional ability, here we examine among Tsimane adults whether compromised bone strength is associated with diminished participation in routine subsistence tasks. We utilize thoracic computed tomography (CT) to measure two primary indicators of bone strength in thoracic vertebral bodies: bone mineral density (BMD), which accounts for ~70% of the variance in bone strength (NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy, 2001), and fracture presence and severity. We focus on thoracic vertebrae for several reasons. Thoracic vertebral deformity is regularly observed in diverse hominin skeletal samples (Chapman, 1972; Cook et al., 1983; Dequeker et al., 1997; Foldes et al., 1995; Gresky et al., 2016; Lieverse et al., 2007; Mays, 1996; Trinkaus, 1985). Thoracic vertebral fractures are also among the most common fragility fractures in humans living in post-industrialized populations (Sambrook and Cooper, 2006), and are more common among Tsimane compared to age- and sex-matched US (Los Angeles) controls (Stieglitz et al., 2019). While other great apes experience degenerative disease (Jurmain, 2000; Lovell, 1990), spontaneous vertebral fractures have not been observed in other apes, even in individuals with severe osteopenia (Gunji et al., 2003). Bipedal hominins thus appear to be especially susceptible to fragility fractures of the spine (Cotter et al., 2011) and perhaps other anatomical regions. Human thoracic vertebrae track mechanics of both lower and upper limbs, and thoracic vertebral body fracture can directly impede mobility. Clinical vertebral deformities are commonly associated with pain and impaired quality of life and can have serious long-lasting economic consequences (Francis et al., 2008). Declining thoracic vertebral body BMD is a manifestation of senescence more generally, and rather than directly inhibiting specific functional abilities per se, a co-occurrence of reduced BMD and disability could indicate more general degenerative processes that do not necessarily involve vertebral fracture (e.g. vertebral disc degeneration, osteophytes, and nerve damage). Regardless of the specific mechanisms, determining whether and how compromised thoracic vertebral body strength is associated with diminished productivity in an extant small-scale subsistence population whose behavior can be directly studied improves the ability to infer more generally potential economic costs of disability for an obligate biped reliant on extractive foraging and food sharing.

In this paper we identify economic correlates of compromised bone strength (i.e. thoracic vertebral body fracture and lower thoracic vertebral body BMD) by examining involvement in four common tasks in foraging economies: hunting, tree chopping, tool manufacture (i.e. weaving), and walking long distances. These four tasks were selected because of their importance in daily subsistence and their variation in strength and skill requirements. Hunting has likely been an important source of hominin protein and fat production since early Homo (Aiello and Wheeler, 1995; Antón et al., 2014; Wrangham and Carmody, 2010), and perhaps earlier, although the dietary availability of meat and other animal products (e.g. marrow) in the Plio-Pleistocene was variable across space and time. Tree chopping is essential for constructing shelters and footbridges for water crossings, hunting arboreal or burrowing prey that hide in tree trunks, manufacturing certain tools, gathering firewood, and horticultural production. Tool manufacture and repair have likely long improved extractive foraging efficiency for omnivorous, manually dexterous hominins, as suggested by the fact that Oldowan tool-using hominins absorbed energetic costs of transporting stones for processing carcasses over long distances (>10 km) (Braun et al., 2008). Weaving of items used for carrying diverse objects (e.g. woven bags that store animal carcasses) and for resting (e.g. ground mats) is routinely performed by women in numerous small-scale subsistence populations, including the Tsimane (Figure 1). Lastly, the ability to efficiently travel long distances is evident for bipedal hominins relative to quadrupedal apes by at least the mid-Pliocene (Pontzer et al., 2009); long-distance travel is crucial for participating in diverse hominin foraging tasks (e.g. hunting and fishing) and in social activities (e.g. visiting) within and across residential settlements.

Tsimane women weaving bags used for carrying diverse objects (left panel; photo credit: Jonathan Stieglitz) and ground mats used for resting (right panel; photo credit: Arnulfo Cary).

Finished woven products are also shown in each panel.

We first test whether probability of completely ceasing to perform a given subsistence task increases with compromised bone strength. There are several possible explanations for how compromised bone strength may be associated with reduced economic productivity, and here we focus on the mediating role of functional disability. One might expect considerable productivity loss associated with compromised human thoracic vertebral body strength, as most activities of daily living (e.g. sitting, walking, running, lifting, and even breathing) generate vertebral loads (Myers and Wilson, 1997; Polga et al., 2004; Rohlmannt et al., 2001; Stewart and Hall, 2006). For many common activities (e.g. neutral standing, standing with weight, mild trunk flexion and extension, and lifting objects above the head) the greatest spinal compressive loads are generated in the thoracolumbar region (Bruno et al., 2017). Compromised thoracic vertebral body strength could especially curtail participation in high-strength and/or high-endurance tasks involving greater vertebral loads, such as hunting and tree chopping; participation in less physically demanding tasks involving reduced vertebral loads, such as weaving, may not be strongly curtailed. To the extent that men may more often participate in tasks requiring high levels of strength and/or endurance in subsistence economies like that of the Tsimane with marked sexual division of labor, compromised thoracic vertebral body strength could disproportionately hinder male subsistence activities.

We next estimate the magnitude of productivity losses (i.e. lost kcals/day) from hunting cessation associated with compromised bone strength. Productivity losses are estimated in two ways: (i) daily losses at age x; and (2) expected cumulative future losses from age x onward, where future losses are discounted by mortality. Substantial cumulative future productivity losses may be expected from compromised vertebral body strength because trabecular bone losses during adulthood are not fully regained (Mosekilde, 1989; Mosekilde et al., 1987). Once spinal fracture occurs it may impede daily subsistence activities for the remainder of life. We focus on hunting productivity loss given the centrality of animal-based foods to theoretical models of hominin behavioral and life history evolution (Bliege Bird et al., 2001; Hawkes et al., 2001; Hill and Hurtado, 2009; Isaac, 1978; Kaplan et al., 2000; Marlowe, 2000; Washburn and Lancaster, 1968). Since hominins as a taxonomic group show high plasticity in life history traits (e.g. expected adult lifespan) and diet (e.g. reliance on meat), it is possible that our estimates of productivity loss associated with compromised bone strength better approximate those for hominins exhibiting relatively modern life histories. In the present context, hunting productivity losses refer to lost hunting calories from hunting cessation associated with thoracic vertebral body fracture and/or lower thoracic vertebral BMD.

We do not assume ex ante that compromised bone strength per se is the only cause of disability or reduced productivity, or that these factors necessarily co-vary in some coordinated fashion – these are empirical questions that are partially addressed in this paper. We do expect that individuals with compromised bone strength who are no longer able to perform specific subsistence tasks will adjust their time allocation depending on various factors (e.g. own somatic condition and household demand). To gain insights into potential compensatory behavioral strategies of adults with compromised bone strength who can no longer hunt, we compare expected productivity losses from hunting to observed fishing production rates, as a means of ascertaining the degree of foraging modification required to offset disability-associated energetic deficits. This exercise provides insights into whether adults with compromised bone strength who cease hunting, through their own behavioral modifications, can easily retain their ability to provision themselves and kin with critical animal products in the diet, or whether provisioning by group members or other behavioral modifications may be necessary. Lastly, utilizing data on anthropometrics and self-reported reasons for ceasing to perform specific subsistence tasks, we examine whether compromised bone strength precedes task cessation, rather than the reverse.

Materials and methods

Study population

Request a detailed protocol

Tsimane are semi-sedentary forager-horticulturalists inhabiting >90 villages in lowland Bolivia. Villages are usually located along rivers or other water sources (e.g. lagoons) and are composed of dispersed clusters of kin-related households. The household is the basic economic unit for production and consumption, and most daily coordination is confined to the nuclear and extended family. Tsimane hunt with a diverse toolkit that can include any/all of the following: rifle or shotgun (but guns and ammunition are often unavailable), bow and arrow, machete, slingshot and trap, and sometimes assistance from tracking dogs (Gurven et al., 2006). On average, men hunt about once every week or two; hunting is more common during the wet season and in interior forest villages. The average hunt lasts 8.4 hr and covers 17.9 km (Trumble et al., 2014) and often involves carrying animal carcasses long distances. Tsimane women occasionally hunt, much less often than men, and use different strategies (e.g. women rarely use firearms) with lower return rates . Because of women’s relatively infrequent participation in hunting, we focus here only on men’s hunting behavior. Fishing typically occurs in rivers, streams, or lagoons and involves use of hook-and-line, bow and arrow, machete, plant poison, and/or net. Single-day fishing and hunting activities are mostly solitary or in pairs with siblings, sons, in-laws, or age-mates. Entire families sometimes go on multi-day fishing and hunting trips that can last from 2 days to several months. Food sharing is widespread within extended families, but also limited in scope across families. Kinship and relative need largely determine the magnitude and direction of food transfers (Gurven et al., 2012; Hooper et al., 2015).

Much of the protein and fat consumed by Tsimane come from hunting and fishing (Kraft et al., 2018), whereas most (62%) of the total calories in the diet come from cultigens grown in small swiddens. In order to clear space for cultigens, machetes are first used to clear smaller vegetation, and then metal axes (and more recently chainsaws when available) are used to cut larger trees before planting and burning (Trumble et al., 2013). While both sexes plant cultigens, help clear smaller vegetation, and harvest, the felling of larger trees is typically done by men. Women’s physical activity level (PAL; defined as the ratio of total daily energy expenditure to daily basal metabolism) is in the ‘moderate to active’ range (PAL = 1.7–1.9) and is constant throughout adulthood. Men’s PAL is ‘vigorously active’ (PAL = 2.0–2.2) and declines by 10–20% from the peak (in the late 20s) to older adulthood (age 60+ years) (Gurven et al., 2013).

Physical strength peaks in the mid-to-late 20s, then declines thereafter, reaching ~60% of maximum adult capacity by age 70 (Gurven et al., 2006). Impaired musculoskeletal function is not uncommon among Tsimane adults (see Figure 3H in Gurven et al., 2020a). BMD estimates from the axial and appendicular skeleton – particularly from sites rich in trabecular bone (i.e. thoracic vertebrae, calcaneus, and distal radius) – suggest that osteoporosis is not uncommon, especially for post-menopausal women (Stieglitz et al., 2015a; Stieglitz et al., 2016; Stieglitz et al., 2017; Stieglitz et al., 2019). Early and rapid reproduction may contribute to women’s skeletal fragility later in life. Compared to more sedentary Los Angeles matched controls, age-standardized prevalence of thoracic vertebral body fracture is two times higher for Tsimane women and 3.3 times higher for men using directly comparable CT methods (Stieglitz et al., 2019). For adults aged 50+, 38% experience some movement restriction (e.g. during flexion or extension), most commonly in the dorsolumbar, hip, knee, or shoulder regions. Nearly 70% of adults aged 50+ are also clinically diagnosed with some condition of the musculoskeletal system and connective tissue (i.e. codes M00–M99 using the International Classification of Diseases version 10) by Tsimane Health and Life History Project (THLHP) physicians. By age 70 most adults report at least some physical discomfort during activities of daily living (e.g. bathing).

Study design and participants

Request a detailed protocol

The Tsimane thoracic CT sample used to measure thoracic vertebral body BMD and fracture presence and severity includes all individuals who met the inclusion criteria of self-identifying as Tsimane and who were aged 40+ at study enrollment (mean ± SD age = 55.9 ± 9.9, age range: 41–91 years, 48% female, n = 493) (see Appendix for additional details and Appendix 1—table 1 for descriptives of study variables). This CT sample is thought to be representative of all Tsimane aged 40+ years . Younger adults (<40 years old) were excluded because CT scans were performed as part of a broader project on atherosclerosis, including arterial plaque formation, which is usually negligible at younger ages for physically active adults. No Tsimane were excluded based on any health condition that can affect BMD or fracture risk.

Institutional IRB approval was granted by UNM (HRRC # 07–157) and UCSB (# 3-16-0766), as was informed consent at three levels: (1) Tsimane government that oversees research projects, (2) village leadership, and (3) study participants.

Thoracic CT

Request a detailed protocol

Thoracic CT scans were performed at the Hospital Presidente German Busch in Trinidad, Bolivia using a 16-detector row scanner (GE Brightspeed, Milwaukee, WI, USA). A licensed radiology technician acquired a single non-contrast thoracic scan, which was supervised and reviewed by at least one of the HORUS study team cardiologists and radiologists with whom we collaborate to study cardiovascular functioning (Kaplan et al., 2017). Breath-hold instructions were given in the Tsimane language to minimize respiratory motion artifact and misregistration. CT settings were as follows: 250 ms exposure, 2.5 mm slice thickness, 0.5 s rotation speed, 120 kVp, and 40 mA (see Kaplan et al., 2017; Stieglitz et al., 2019 for additional details).

Thoracic vertebral body BMD

Request a detailed protocol

CT-derived vertebral body BMD estimates are highly accurate and strongly positively correlated (Pearson r’s > 0.9) with ash weight of dry bone (Ho et al., 1990; Reinbold et al., 1991). Thoracic vertebral body BMD was measured manually in each of three consecutive vertebrae (T7–T10 range) by a radiologist with 20+ years of experience (see Appendix for additional details). Lumbar vertebrae were not assessed as they did not appear in the CT image field of view. BMD measurement started at the level of the section that contained the left main coronary artery (LMCA) caudally (beginning at either T7 or T8, depending on the origin of the LMCA). The LMCA was set as the reference site to allow reproducible detection of a spinal level; because the LMCA is covered in 100% of images and the field of view can be completely reconstructed, it is an optimal reference point to locate the starting measurement level. The LMCA has previously been found to originate at T7 about as frequently as T8 (Budoff et al., 2010).

The region of interest was located around the center of the vertebral body, with the edges 2–3 mm from the vertebral shell. This distance ensured that vertebral body BMD measurements excluded the cortical bone of the vertebral shell. For each vertebra, the radiologist manually positioned a circular region of interest while demarcating cortical from trabecular bone based on visual inspection. Any area with large vessels, bone island fractures, and calcified herniated disks were excluded as much as possible from the region of interest with use of the manual free tracing protocol. Mean BMD for the three consecutive thoracic vertebrae was then calculated. This CT-derived thoracic vertebral body BMD measure is strongly positively correlated (Pearson r’s > 0.9) with CT-derived lumbar vertebral body BMD measures (Budoff et al., 2012). CT-derived BMD estimates can be obtained with and without calibration phantoms. Phantomless BMD estimates correlate strongly (Pearson r = 0.99) with standard phantom-based CT BMD estimates (Budoff et al., 2013). Hounsfield units were converted to BMD (mg/cm3) using a calibration phantom of known density, or a scanner-specific mean calibration factor for the T7–T10 vertebrae from scans performed without the phantom.

Thoracic vertebral body fracture

Request a detailed protocol

For each participant the radiologist also classified seven vertebrae (T6–T12) according to Genant’s semi-quantitative technique (GST) (Genant et al., 1993). While there is no consensus regarding the radiologic definition of vertebral fracture, the GST provides highly reproducible diagnosis of fractures and is currently the most widely used clinical technique for identifying and diagnosing fracture (Shepherd et al., 2015). Based on visual inspection, each vertebra is rated according to severity of loss of vertebral height and other qualitative features, including alterations in shape and configuration of the vertebra relative to adjacent vertebrae and expected normal appearances. Each vertebra is classified into one of five categories: normal (grade 0); mild fracture (grade 1; approximately 20–25% reduction in anterior, middle, and/or posterior vertebral height, and a 10–20% reduction in projected vertebral area); moderate fracture (grade 2; 25–40% reduction in any height and a 20–40% reduction in area); and severe fracture (grade 3; >40% reduction in any height and area). A grade 0.5 indicates borderline deformed vertebra (<20% reduction in any height) that is not considered to be a definitive fracture (see Appendix for additional details). Each participant is assigned one summary grade measure of all seven vertebrae. Participants with >1 vertebral deformity are classified according to their most severe deformity. Participants are considered to present fracture if any vertebral body is graded at least mildly deformed (i.e. grade ≥1); participants are considered to present no fracture if graded 0 or 0.5. Given lower observer agreement for mild fractures (the most common) relative to moderate and severe fractures (Lentle et al., 2018), we repeated some analyses using a more conservative fracture definition (i.e. grade ≥2). Unless otherwise noted we report results using the former definition.

Subsistence involvement and hunting productivity losses

Request a detailed protocol

A few days or weeks before traveling from their home villages to Trinidad for CT scanning, participants received in their villages a medical exam from a Bolivian physician and bilingual (Spanish–Tsimane) physician assistant from the THLHP. During the medical exam, participants were asked whether they continued or completely ceased involvement in routine subsistence tasks (e.g. ‘Do you still hunt? Have you hunted in the past few months, or have you stopped hunting altogether?’). Participants were queried about the following four tasks: hunting (men only), tree chopping (men only), weaving (women only), and ability to walk an entire day (both sexes) (also see Stieglitz et al., 2015b). Other subsistence tasks (e.g. fishing and collecting fruit) were omitted due to time constraints during data collection. Women were not queried about hunting or tree chopping (nor were men about weaving) because they rarely participate in these tasks (Gurven et al., 2009). We were able to minimize assignment of ‘false negatives’ (i.e. coding capable participant i as ceasing task j due to a temporary illness which only inhibited i’s involvement in j in the shorter but not longer term) by cross-checking participant response stability over time using repeated measures from medical exams performed in other project years (before and in a few cases after CT scanning; 98% of CT study participants received at least one medical exam in other [i.e. non-CT scanning] project years [mean ± SD number of medical exams/participant = 4 ± 2, min = 1, max = 8]). Responses were generally stable within individuals over time, resulting in very few false negatives for a given task. If a participant reported ceasing a given task in a given year, the physician then asked whether cessation resulted from specific functional limitations related to sensory perception (e.g. difficulty seeing or hearing), endurance (e.g. feeling tired or weak), and mobility limitations (e.g. hip problems).

Estimates of Tsimane age-specific daily caloric production (total and task-specific [e.g. hunting and fishing]) and time allocation have been published elsewhere (Gurven et al., 2012; Gurven et al., 2009; Hooper, 2011; Hooper et al., 2014; Hooper et al., 2015; Kaplan et al., 2010; Schniter et al., 2015; Stieglitz et al., 2014). Briefly, production was estimated by interviewing adults once or twice per week from January 2005 to December 2009 about time allocation and production for each co-resident individual over age 6 during the previous 2 days (n = 1245 individuals from 11 villages). Each family was interviewed an average of 46 times (SD = 20), yielding an average of 93 sample days per individual. Each co-resident individual contributes production data, regardless of their BMD or fracture status (which was unknown at the time of economic data collection); thus, from individual-level production data we cannot determine whether adults with fracture or lower BMD have lower production efficiency (e.g. kcals/hr) than adults without fracture or higher BMD.

To estimate men's hunting productivity losses associated with fracture and lower BMD, we multiply men’s average age-specific daily hunting production by the probability that men still hunt at a given age. This probability that men still hunt is modeled using binary logistic regression with the following covariates: age (years), fracture status (yes/no), and BMD (mg/cm3; also see Appendix). This probability thus serves as a multiplier adjusting men’s average daily hunting production for disability associated with fracture and lower BMD. To determine productivity loss we take the difference in estimated hunting kcals/day at each age for a given fracture and/or BMD value. For example, if a 40-year-old produces 2000 hunting kcals/day, and if the probability that a 40-year-old without fracture still hunts is 0.99 (vs. 0.85 for a 40-year-old with fracture), then the predicted hunting production is 1980 kcals/day for a 40-year-old without fracture (i.e. 2000 × 0.99), and 1700 kcals/day for a 40-year-old with fracture (i.e. 2000 × 0.85; productivity loss = 280 hunting kcals/day). Similarly, if a 50-year-old produces 1800 hunting kcals/day, and if the probability that a 50-year-old without fracture still hunts is 0.9 (vs. 0.75 for a 50-year-old with fracture), then the predicted hunting production is 1620 kcals/day for a 50-year-old without fracture and 1350 kcals/day for a 50-year-old with fracture (productivity loss = 270 hunting kcals/day; that is, 10 fewer daily lost kcals compared to age 40). We assume that hunting return rates (i.e. kcals/hr hunting) are identical for men regardless of fracture status or BMD. Since we do not assume reduced hunting efficiency for men with fracture or lower BMD, the productivity losses reported here may be conservative, and result entirely from complete hunting cessation. This hunting productivity loss is our estimate of the economic cost associated with compromised bone strength. We estimate this cost in two ways: (1) daily loss at age x, as in the prior example of a 40- and 50 year-old; and (2) expected cumulative future losses from age x onward, where future losses are discounted by mortality. Conceptually, this latter value is similar to Fisher’s reproductive value (expected future reproduction at each age) but substitutes age-specific fertility with age-specific caloric production from hunting, given Tsimane population age structure and men’s mortality rates (Gurven et al., 2012).

Socio-demographics and anthropometrics

Request a detailed protocol

Birth years were assigned based on a combination of methods described elsewhere (see Appendix for additional details). Height and weight were measured during THLHP medical exams using a Seca portable stadiometer and Tanita scale. The scale uses bioelectrical impedance to estimate body fat percentage using proprietary estimation equations. Using weight and percent body fat we calculated fat mass (weight × percent body fat) and fat-free mass (weight − fat mass).

Data analysis

Request a detailed protocol

In descriptive analyses we report fracture prevalence for each sex and test for effects of age and BMD on fracture risk using log-binomial generalized linear models. Chi-squared tests are used to compare the differences in subsistence participation (e.g. whether continuing or completely ceasing hunting or tree chopping) by fracture status. Binary logistic regression is used to model the probability of subsistence task cessation as a function of fracture status and BMD after adjusting for age. We use stepwise regression with fracture status included first, and then BMD, and we compare models based on Akaike information criterion (AIC). False discovery rate (FDR) q-values were computed using the R package ‘qvalue’ to account for the multiple testing burden across different subsistence tasks. Parameter estimates are reported as odds ratios (ORs) or predicted probabilities.

We apply several indirect methods for assessing reverse causality, whereby ceasing task performance leads to compromised bone strength (we cannot determine from the CT scan or any other THLHP data when fracture occurred). First, we use chi-squared tests to compare differences by fracture status in self-reported reasons for subsistence task cessation, including deficits in sensory perception, strength, endurance, and mobility. If fractures cause task cessation, then those with fractures should be more likely to report mobility limitations as a main reason for task cessation, rather than other reasons. Second, we use general linear models to compare BMD of adults who continue vs. cease task participation (adjusting for age and fat-free mass). This comparison helps determine whether adults who cease participation experience greater risk of skeletal fragility overall. Lastly, for reasons described below, we test for a mediating effect of anthropometric status in logistic regressions of task cessation on fracture status and BMD. For all analyses we use α = 0.05 as the cutoff for statistical significance (we report both p-values and FDR q-values where relevant). Participants with any missing values are removed from analyses. Data that support the findings of this study are available on Dryad (https://doi.org/10.5061/dryad.h44j0zphj).

Results

Descriptives: thoracic vertebral body fracture prevalence and covariation with age and thoracic vertebral body BMD

Men

Prevalence of any thoracic vertebral body fracture (i.e. grade ≥1; including mild, moderate, or severe) for men aged 40+ years is 36% (95% CI: 30–42) (Figure 2). Using a more conservative fracture definition (i.e. grade ≥2; including only moderate or severe), men’s prevalence is 11% (95% CI: 7–15). Neither men’s age nor thoracic vertebral body BMD is significantly associated with fracture risk (Figure 2), regardless of fracture grade.

Top: Proportion of Tsimane men with thoracic vertebral body fracture (grade ≥1) by age, estimated using the loess function.

The shaded region shows ±1 SE, and jittered data points represent fracture status. Bottom: scatterplot of thoracic vertebral body bone mineral density (BMD) by age and fracture status, including linear regression lines for each fracture status. N = 256 men.

Women

Prevalence of any grade 1 fracture for women aged 40+ years is 19% (95% CI: 14–24) (Figure 3). Women’s grade 2 fracture prevalence is 7% (95% CI: 4–10). Women’s age does not significantly predict fracture risk in univariate models, regardless of fracture grade. BMD is inversely associated with women’s fracture risk after adjusting for age (grade 1: adjusted relative riskBMD per SD increase=0.49, 95% CI: 0.34–0.72, p<0.001; grade 2: adjusted relative riskBMD per SD increase=0.25, 95% CI: 0.11–0.57, p=0.001; see Figure 3); in both of these models that include BMD as a covariate, age is inversely and significantly associated with fracture risk.

Top: Proportion of Tsimane women with thoracic vertebral body fracture (grade ≥1) by age, estimated using the loess function.

The shaded region shows ±1 SE, and jittered data points represent fracture status. Bottom: scatterplot of thoracic vertebral body bone mineral density (BMD) by age and fracture status, including linear regression lines for each fracture status. N = 237 women.

Hunting cessation is associated with thoracic vertebral body fracture and, albeit weakly, with lower BMD

Thirty-eight percent of men with fracture (grade ≥1) ceased hunting, compared to 13% with no fracture (Table 1). Results are similar using a more conservative fracture definition (i.e. grade ≥2; see Appendix 1—table 2; subsequently we only present results where fracture grade ≥1). After adjusting for age, odds of hunting cessation are 7.3 times greater (95% CI: 3.3–17.6, p<0.001, FDR q < 0.001, n = 256) for men with vs. without fracture (Appendix 1—table 3: Model 1). The association between odds of hunting cessation and fracture increases slightly (adjusted ORFracture = 7.4) after also including thoracic vertebral body BMD as a covariate (adjusted ORBMD per SD=0.62, 95% CI: 0.38–0.99, p=0.054, FDR q = 0.161, Appendix 1—table 3: Model 2).

Table 1
Percentage of men (95% CI) who completely ceased hunting by age and thoracic vertebral body fracture status.
Age category (years)% ceased hunting with fractureN% ceased hunting without fractureN
40–498^ (<1–19)260 (-----)55
50–5933*** (17–50)364 (<1–9)56
60–6950^ (24–76)1827 (12–42)37
70+100* (-----)1263 (36–89)16
Total38*** (28–48)9213 (8–19)164
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact Test [vs. no fracture]).

    Max age = 83 years (mean ± SD = 75 ± 4).

  2. Max age = 81 years (mean ± SD = 74 ± 3).

Hunting cessation associated with thoracic vertebral body fracture and lower BMD generates substantial productivity losses

Fracture

Figure 4 shows men’s age-specific hunting productivity by fracture status (see Appendix 1—table 4 for production values and additional details). Men with fracture under-produce to a lesser extent at younger vs. older ages because of high hunting participation rates by younger men regardless of fracture status (Table 1 and Appendix 1—table 2). As hunting cessation becomes increasingly likely with age, particularly for men with fracture, daily productivity losses associated with fracture (i.e. the hunting productivity differential for men with vs. without fracture) increase until age 61, when men with fracture produce 481 fewer hunting kcals/day than men without fracture (1007 vs. 526 kcals/day, respectively). This loss of 481 kcals/day – the equivalent of 120 g of protein (or 53 g of fat) – represents 19% of the mean daily per-capita energy intake for a Tsimane man that age (2592 kcals), and 52% of the mean intake from protein and fat (919 kcals). By comparison, this loss of 481 hunting kcals/day at age 61 exceeds daily fishing production at that age by 171 kcals (based on men’s observed age-specific daily fishing production [see Materials and methods and Figure 4]). Daily hunting productivity losses associated with fracture exceed fishing productivity for 14 years, from ages 55 to 68 (by 29–171 kcals/day). In this same age range, based on observed age-specific return rates and time allocation, to wholly offset hunting productivity losses a Tsimane man would have to fish an additional 51–76 min/day (e.g. from an observed 49 to an expected 125 fishing min/day at age 61 [an increase of 155%]; assuming identical fishing return rates regardless of fracture status and BMD; see Materials and methods). As men’s hunting participation rates decline with age regardless of fracture status, hunting productivity losses associated with fracture also decline, reaching zero by age 75 (when hunting returns reach zero; Figure 4).

Top: Age-specific productivity (kcals/day) for hunting by men’s fracture status, and for fishing (all men).

Bottom: expected cumulative future productivity (millions of kcals) from age x onward, discounted by mortality, for hunting by men’s fracture status, and for fishing (all men).

Figure 4 also shows expected cumulative future productivity over the rest of life from a specific age onward, discounted by mortality. Future productivity losses associated with fracture (i.e. the hunting productivity differential for men with vs. without fracture) peak at age 40, when men with fracture can expect to lose 1.15 million kcals over their remaining life; at that age, this loss represents 14% of men’s total future hunting production (8.49 million kcals). Future hunting productivity losses associated with fracture exceed future fishing productivity for 16 years, from ages 49 to 64 (by 2–32% [or 4772–96,419 expected cumulative future kcals]; Figure 4). In this same age range, based on the maximum observed fishing return rate (420 kcals/hr at age 49) and observed age-specific time allocation (e.g. 53 fishing min/day at age 49), to wholly offset future hunting productivity losses associated with fracture a Tsimane man would have to fish an additional 258–1780 hr (i.e. 324–2012 additional future fishing days). Here we applied the maximum observed return rate to derive a conservative estimate. If instead we apply the minimum rate (360 kcals/hr at age 64), then to wholly offset future hunting productivity losses associated with fracture a man would have to fish an additional 302–2076 hr (i.e. 378–2347 additional future fishing days).

Lower BMD

We now estimate additional productivity losses (again expressed as the hunting productivity differential) associated with having lower BMD. Losses are greatest for men with fracture and lower BMD vs. men without fracture and higher BMD (Figure 5; see Appendix 1—table 5 for production values and additional details). In this case, peak losses at a given age (age 60: 650 kcals/day – the equivalent of 163 g of protein [or 72 g of fat]) represent 25% of the mean daily per-capita energy intake for a Tsimane man that age and 70% of the mean intake from protein and fat. By comparison, this loss of 650 hunting kcals/day at age 60 exceeds daily fishing production at that age by 333 kcals (based on men’s observed age-specific daily fishing production [see Materials and methods]). Hunting productivity losses for men with fracture and lower BMD (vs. men without fracture and higher BMD) exceed fishing productivity for 21 years, from ages 50 to 70 (by 33–335 kcals/day). In this same age range, based on observed age-specific return rates and time allocation, to wholly offset hunting productivity losses a Tsimane man would have to fish an additional 54–102 min/day (e.g. from an observed 49 to an expected 151 fishing min/day at age 60 [an increase of 205%]; assuming identical fishing return rates regardless of fracture status and BMD; see Materials and methods). Hunting productivity losses are smallest for men without fracture and higher vs. lower BMD (max = 136 kcals/day at age 65; Figure 5).

Top: Age-specific productivity (kcals/day) for hunting by men’s fracture status and bone mineral density (BMD), and for fishing (all men).

For illustrative purposes we show daily hunting production for men with +1 SD and −1 SD of the BMD mean. Bottom: expected cumulative future productivity (millions of kcals) from age x onward, discounted by mortality, for hunting by men’s fracture status and BMD, and for fishing (all men).

Expected cumulative future losses are also greatest for men with fracture and lower BMD vs. men without fracture and higher BMD (Figure 5). At the peak (age 40) the former are expected to produce 1.88 million fewer future hunting kcals than the latter. At that age this loss represents 22% of men’s total future hunting production. Future hunting productivity losses for men with fracture and lower BMD vs. men without fracture and higher BMD exceed future fishing productivity for 28 years, from ages 40 to 67 (by 0.1–81% [or 1193–383,902 expected cumulative future kcals]). In this same age range, based on the maximum observed fishing return rate and observed age-specific time allocation, to wholly offset future hunting productivity losses a Tsimane man would have to fish an additional 160–4443 hr (i.e. 207–4849 additional future fishing days). Here, again, we applied the maximum observed return rate to derive a conservative estimate. If instead we apply the minimum rate, then to wholly offset future hunting productivity losses a man would have to fish an additional 202–5594 hr (i.e. 261–6105 additional future fishing days).

Tree chopping cessation is associated with thoracic vertebral body fracture but not lower BMD

Forty-six percent of men with fracture ceased chopping trees, compared to 21% with no fracture (Table 2). After adjusting for age, odds of tree chopping cessation are 6.9 times greater (95% CI: 3.1–16.6, p<0.001, FDR q < 0.001, n = 256) for men with vs. without fracture (Appendix 1—table 6: Model 1). The association between odds of tree chopping cessation and fracture weakens slightly (adjusted ORFracture = 6.8) after also including thoracic vertebral body BMD as a covariate, which is not associated with tree chopping cessation (adjusted ORBMD per SD=0.75, 95% CI: 0.47–1.16, p=0.204, FDR q = 0.306, Appendix 1—table 6: Model 2).

Table 2
Percentage of men (95% CI) who completely ceased tree chopping by age and thoracic vertebral body fracture status.
Age category (years)% ceased tree chopping with fractureN% ceased tree chopping without fractureN
40–4915** (1–30)260 (-----)55
50–5939*** (22–56)364 (<1–9)56
60–6967 (43–91)1854 (37–71)37
70+100 (-----)1281 (60–100)16
Total46*** (35–56)9221 (15–28)164
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact Test [vs. no fracture]).

    Max age = 83 years (mean ± SD = 75 ± 4).

  2. Max age = 81 years (mean ± SD = 74 ± 3).

Weaving cessation is not associated with thoracic vertebral body fracture but is associated, albeit weakly, with lower BMD

Twenty percent of women with fracture ceased weaving, compared to 13% with no fracture (Table 3). After adjusting for age, odds of weaving cessation are not significantly associated with fracture (adjusted ORFracture = 2.2, 95% CI: 0.8–6.4, p=0.129, FDR q = 0.172, n = 237, Appendix 1—table 7: Model 1). Weaving is the only task examined for which fracture is not associated with compromised participation. Adding BMD as a covariate further weakens the association between fracture and weaving cessation (adjusted ORFracture = 1.8; adjusted ORBMD per SD=0.51, 95% CI: 0.23–1.05, p=0.079, FDR q = 0.158; Appendix 1—table 7: Model 2).

Table 3
Percentage of women (95% CI) who completely ceased weaving by age and thoracic vertebral body fracture status.
Age category (years)% ceased weaving with fractureN% ceased weaving without fractureN
40–490 (-----)110 (-----)72
50–5914 (<1–29)225 (<1–11)60
60–6933 (<1–72)924 (10–39)37
70+75 (<1–100)455 (32–77)22
Total20 (8–31)4613 (8–17)191
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact Test [vs. no fracture]).

    Max age = 91 years (mean ± SD = 80 ± 9).

  2. Max age = 91 years (mean ± SD = 77 ± 6).

Regardless of fracture status or BMD, all women who stopped weaving reported experiencing problems in all bodily regions (i.e. hip, back, and hands) mentioned by the physician during medical exams, and nearly all of these women also reported visual impairment (Appendix 1—table 8).

Restricted day range is associated with thoracic vertebral body fracture and, albeit weakly, with lower BMD

Sixty-nine percent of adults with fracture are unable to walk a full day, compared to 38% with no fracture (Table 4). After adjusting for age and sex, odds of inability to walk all day are 8.2 times greater (95% CI: 4.8–14.5, p<0.001, FDR q < 0.001, n = 493) for adults with vs. without fracture (Appendix 1—table 9: Model 1). The association between odds of inability to walk all day and fracture weakens slightly (adjusted ORFracture = 7.8) after also including thoracic vertebral body BMD as a covariate (adjusted ORBMD per SD=0.77, 95% CI: 0.57–1.03, p=0.082, FDR q = 0.109, Appendix 1—table 9: Model 2).

Table 4
Percentage of adults (95% CI) who are unable to walk a full day by age and thoracic vertebral body fracture status.
Age category (years)% unable to walk all day with fractureN% unable to walk all day without fractureN
40–4949*** (32–66)3719 (12–26)127
50–5964*** (51–77)5827 (19–35)116
60–6989** (76–100)2764 (52–75)74
70+100^ (-----)1684 (72–96)38
Total69*** (61–77)13838 (33–43)355
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact Test [vs. no fracture]).

    Max age = 91 years (mean ± SD = 76 ± 6).

  2. Max age = 91 years (mean ± SD = 76 ± 5).

Suggestive evidence of fracture preceding task cessation, rather than the reverse

Men with fracture are more likely than men without fracture to report mobility limitations due to hip problems as a reason for hunting and tree chopping cessation (Appendix 1—tables 10 and 11). This is noteworthy because among men who ceased hunting and tree chopping, men with fracture are younger, on average, than men without fracture, and all else equal one expects younger men to be less likely than older men to report hip problems (for men ceasing hunting: mean age by fracture status = 63 vs. 68; for men ceasing tree chopping: mean age = 62 vs. 68). Consistent with their younger age, men with fracture are less likely than men without fracture, albeit not significantly, to report problems with endurance (i.e. feeling weak and tiring easily) or sensory processing (i.e. auditory and visual) as reasons for hunting cessation (Appendix 1—table 10). It is thus less likely that men with fractures are in poorer condition more generally, and more likely that fractures contributed to the mobility complaints that led men to stop hunting and tree chopping. As with hunting and tree chopping, among adults who are unable to walk all day, those with fracture are younger than those without fracture (mean age = 59 vs. 62) (there are no differences by fracture status in self-reported reasons for inability to walk all day [see Appendix 1—table 12]).

After adjusting for age and fat-free mass, men who are no longer able to hunt, chop trees, and walk all day do not differ in their BMD vs. men who still hunt (95% CI: −18.1 to 4.4 mg/cm3), chop trees (95% CI: −15.0 to 7.1 mg/cm3), and can walk all day (95% CI: −10.2 to 7.9 mg/cm3). This suggests that men who cease these activities do not experience greater risk of overall skeletal fragility. For women, inability to walk all day is associated with reduced BMD (95% CI: −25.3 to −6.1 mg/cm3) after adjusting for age and fat-free mass. Lastly, if hunting or tree chopping cessation preceded fracture (e.g. because reduced meat consumption weakened bone structural integrity, or because muscle degeneration from task cessation reduced bone loading and structural integrity), then one might expect anthropometric status to mediate the association between hunting or tree chopping cessation and fracture. Yet these associations strengthen after adjusting for anthropometrics (i.e. weight, fat-free mass, and fat mass). Anthropometric status also does not mediate the association between inability to walk all day and fracture. Together, these results provide suggestive evidence that fracture preceded task cessation, rather than the reverse.

Discussion

We find among adult Tsimane forager-horticulturalists that compromised bone strength – as indicated by thoracic vertebral body fracture and lower thoracic vertebral body BMD – is associated with diminished subsistence involvement. Compromised bone strength is associated with diminished participation in diverse habitual subsistence tasks, particularly those requiring high levels of strength or endurance (i.e. hunting, tree chopping, and walking long distances). In contrast, participation in less physically demanding but skill-intensive tasks (i.e. tool manufacture/repair) is only modestly associated with bone strength. Since strength and endurance are required to exploit the suite of resources upon which omnivorous hominins rely, functional disability may hinder resource production in costly ways for a comparatively long-living, extractive foraging primate. This inference is supported by prior studies in other contemporary small-scale populations that highlight illnesses or injuries of shorter duration (e.g. acute infection, laceration, and animal bite) as constraints on productivity (Bailey, 1991; Sugiyama and Chacon, 2000; Sugiyama, 2004). As most activities of daily living generate loads on human vertebrae (Myers and Wilson, 1997; Rohlmannt et al., 2001; Stewart and Hall, 2006), and since for many activities the greatest compressive loads along the spine are in the thoracolumbar region (Bruno et al., 2017), thoracic vertebral body fracture in particular may be debilitating for an obligate biped even after behavioral modifications that minimize its burden.

Ample bioarchaeological evidence of partially or fully healed fractures (e.g. Lovejoy and Heiple, 1981; Pfeiffer, 2012), usually from excessive trauma, and developmental abnormalities (Cowgill et al., 2015; Trinkaus, 2018) suggest in past populations longer-term survival with varying levels of disability. Fragility fractures resulting from reduced bone strength are relatively rare in bioarchaeological studies (Agarwal, 2008; Agarwal and Grynpas, 1996), even among the elderly, leading some to conclude that osteoporosis was rare or absent in past populations (but see Curate et al., 2010; Dequeker et al., 1997). This and the fact that age-specific osteoporotic fracture incidence rates appear to be increasing over time in Western populations (Cooper et al., 2011) lends support to the notion that osteoporosis is a ‘disease of modernity’ (cf. Kralick and Zemel, 2020; Lieberman, 2013). However, few if any bio-archaeologists systematically study thoracic vertebral compression fractures, and a paucity of observed fragility fractures in the archaeological record is partly due to biased preservation of skeletal remains. Accelerated bone loss with age appears to be a basic feature of human aging in past and present populations (Aspray et al., 1996; Kneissel et al., 1997; Mays et al., 1998; Mays, 1996; Wallace et al., 2014) and has also been documented among free-ranging chimpanzees (Gunji et al., 2003; Morbeck and Galloway, 2002; Sumner et al., 1989). It would thus be premature to conclude that fragility fractures were rare or of minimal consequence throughout human evolution.

We find suggestive evidence that Tsimane fracture precedes subsistence task cessation, rather than the reverse. But there are several possible explanations for the empirical associations reported in this paper (see Figure 6) and given the study design we cannot definitively parse among them. Associations between reduced thoracic vertebral body BMD and task cessation may reflect effects on functional ability of other properties of senescence that were not examined here. The present study does not consider diverse mechanisms (e.g. spinal osteoarthritis, restricted range of motion for a given anatomical region) and future research is needed to establish causal relationships between somatic condition, functional ability, and productivity. A candidate model that is only partially evaluated in this paper (Figure 6) proposes that senescence induces anatomical and sensory declines – including loss of strength in muscles, bones, and connective tissues – which jointly increase risk of fracture, body pain or discomfort, and functional disability. Senescence-induced declines can also induce pain and disability in the absence of fracture or reduced BMD (e.g. from osteoarthritis). The magnitude and pace of anatomical and sensory decline may be influenced not just by senescence but also by habitual participation in strenuous subsistence labor (i.e. ‘mechanical wear and tear’), which can directly cause fracture (e.g. via excessive mechanical stress) and body pain. The manifestation of pain and functional disability following fracture appears to be quite variable both across individuals and within individuals over time (e.g. Begerow et al., 1999; Nevitt et al., 1998; O'Neill et al., 2004; Suzuki et al., 2008) and samples are usually not representative of the population since many fractures remain unobserved. Back pain can arise directly from vertebral fracture or indirectly from consequences of spinal deformity or degenerative changes involving paravertebral muscles, intervertebral discs, associated ligaments, and facet joints. Fractures in particular can entail longer-term productivity constraints as they provoke systemic bone losses, incomplete recovery of bone quality (especially for older adults), increased risk of future fractures, and longer-term disability (Osipov et al., 2018). Future research is needed to test the empirical adequacy of the hypothesized, sometimes bidirectional relationships shown in Figure 6. It is also possible that our results reflect influences of earlier life factors (e.g. activity levels, peak bone mass, and economic productivity) that we do not examine in the present study.

A conceptual model linking somatic condition to economic productivity during aging in small-scale subsistence societies.

Variables in red are a focus of the present study.

Tsimane thoracic vertebral fractures appear to result from both trauma and senescence (Stieglitz et al., 2019) and are not just prevalent among the oldest adults (Figures 2 and 3). Tsimane can thus experience a significant portion of adult life with fracture. Tsimane women’s thoracic vertebral body BMD is inversely associated with thoracic vertebral body fracture risk, suggesting a role of compromised bone strength in precipitating fracture. For men, we suspect that work-related trauma plays a role in precipitating fracture due to a weaker association between BMD and fracture risk (Stieglitz et al., 2019), and our anecdotal observations of high levels of mechanical stress on men’s vertebrae from frequent heavy load carrying (e.g. of animal carcasses, and timber for constructing houses). This of course does not preclude a contributing role of compromised bone strength in precipitating men’s fracture. Likewise, for women, trauma may also precipitate fracture; some of the greatest compressive vertebral loads occur when weights are carried in front of the body (Bruno et al., 2017; Rohlmannt et al., 2001), which is how Tsimane women routinely carry infants and toddlers. High reproductive effort (i.e. early onset of reproduction, on-demand breastfeeding, short interbirth intervals, and high parity) is associated with reduced bone strength for Tsimane women (Stieglitz et al., 2015a; Stieglitz et al., 2019) and may thus constrain the ability of bone tissue to respond to mechanical loading from a physically active lifestyle. Interpersonal violence likely does not precipitate much fracture among men because such violence is relatively rare and generally does not result in severe injuries when it does occur. However, intimate partner violence against women is prevalent (Stieglitz et al., 2018), particularly among younger women, and may precipitate traumatic fracture at diverse skeletal sites.

To more precisely identify the economic cost of disability, we estimated age-specific energetic deficits from hunting cessation by men with compromised bone strength. An advantage of our approach to estimating productivity loss is that it is not subject to recall inaccuracy or subjectivity regarding the meaning of foregone production. Energetic deficits are substantial, averaging from ages 40 to 74 years 261 lost hunting kcals/day for men with vs. without fracture (Figure 4 and Appendix 1—table 4). Average production deficits are even greater for men with both fracture and lower BMD (vs. men without fracture and higher BMD): 397 lost hunting kcals/day, equivalent to 15% of the mean daily per-capita energy intake, and 44% of the mean intake from protein and fat (Kraft et al., 2018; Figure 5 and Appendix 1—table 5). By crude comparison, adults in Western industrialized settings with musculoskeletal problems show productivity losses of 10–15% for frequent computer users across various industries (Hagberg et al., 2002) and up to 28% for construction workers with more physically demanding jobs (Meerding et al., 2005). For osteoporotic adults aged 50+ years in the Netherlands, the cost of clinical spine fracture in terms of work absenteeism exceeds 25% of the mean annual income (Eekman et al., 2014). Tsimane age-specific energetic deficits peak in the early 60s (Figures 4 and 5 and Appendix 1—tables 4 and 5), which is near the age at which Tsimane adults achieve peak numbers of dependent offspring and grand offspring (Gurven and Kaplan, 2007). The magnitude of these disability costs and their timing in the life cycle may thus pose for Tsimane significant challenges to meeting adult survival needs and/or providing assistance (material or other) to descendant kin. These challenges should vary in severity across populations depending on fertility schedules and dependency loads. Our results suggest that the scope of behavioral responses to compensate for productivity loss is limited to activities with lower strength and endurance requirements. Given the ubiquity across subsistence populations of the sexual division of labor, with men typically focusing on high strength tasks, aging women may maintain peak or near-peak subsistence involvement with relatively minimal disability-related productivity losses, despite increasing frailty. Relatedly, in industrialized populations there is some evidence from population-based cohorts that severe thoracolumbar vertebral deformity is more strongly associated with functional impairments for men than women (Burger et al., 1997), perhaps because men load their spines more heavily.

Faced with protein and fat deficits from hunting cessation, one might expect adults with compromised bone strength to increase time spent in other, less physically demanding food acquisition tasks that reliably yield substitutable macronutrients. Fishing is the other major source of protein and fat in small-scale Amazonian subsistence populations, most of whom acquire protein more efficiently from animal- rather than plant-based foods (Hames, 1989). Prior behavioral studies of other contemporary Amazonian populations (Gurven and Kaplan, 2006) indicate a shift with age in men’s time allocation away from hunting and instead toward fishing; for both sexes time spent in fishing increases monotonically throughout adulthood. Fishing is also an important source of protein and fat for Tsimane (Gurven et al., 2013; Hooper, 2011). We find that, to wholly offset their own hunting productivity losses, men with fracture and lower BMD would have to considerably increase their time spent fishing. A perceived need by adults to offset their own current and/or future energetic deficits (Figures 4 and 5) could influence residential decisions by increasing the value of maintaining proximity to fishing locations (e.g. rivers and lakes), particularly during periods of greater fishing efficiency (e.g. dry season). Disability-related production deficits could influence various economic decisions (e.g. by influencing decisions on where to forage on a given day, or where to farm). Minimizing travel costs during food acquisition may be important in light of restricted day ranges among adults with compromised bone strength (see Results). These economic decisions also depend on local demographics and social networks, as productivity losses can be mitigated with the availability of substitute workers. If, for instance, able-bodied adults within the multi-generational sharing unit can substitute for impaired adults in hunting and meat sharing, then impaired adults may take fewer risks to acquire protein and fat compared to those without buffered networks.

Our estimates of the ways in which Tsimane with compromised bone strength may compensate for hunting productivity loss are crude for several reasons. They do not, for example, consider the possibility of adults being provisioned by others, or pursuing other means of protein and fat acquisition beyond fishing. Tsimane can pursue other low-intensity subsistence tasks (e.g. harvesting cultigens), thereby offering additional opportunities for adults to produce and transfer resources outside of a strict hunting and gathering economy. Our crude estimates nevertheless imply that productivity losses entail behavioral modifications and might affect social organization more broadly. The multi-generational production and sharing units that structure human sociality (Bird et al., 2019; Isaac, 1978; Migliano et al., 2017; Washburn and Lancaster, 1968) incorporate divisions of labor whereby impaired adults can still make valuable contributions beyond food acquisition (e.g. through conflict mediation, pedagogy, and leadership). Older adults play an important fitness-enhancing role not just by providing caloric surpluses and transfers, but also by providing childcare, skills and knowledge, and other resources. Tsimane cognitive function is generally well preserved throughout adulthood among both schooled and non-schooled adults (Gurven et al., 2017a), consistent with a late-life service niche emphasizing information transmission of accumulated wisdom, experience, and judgment. Aging Tsimane adults, despite their declining physical abilities, are regularly consulted and respected for their knowledge of the natural and spiritual world (e.g. profitable hunting locations; proper ethics), family lineage histories and past events, healing sick group members, making household crafts, and traditional stories and music (Schniter et al., 2015; Schniter et al., 2018). Aging adults accommodate their changing roles within the family and society and gradually compensate for their increasing frailty by providing valuable expertise across diverse skill- and knowledge-intensive domains that facilitate achieving the childrearing and subsistence goals of the residential group (cf. Levitin, 2020).

The extent to which our results generalize across time and space depends on various considerations including age profiles of production, rates of mortality and senescence, and subsistence ecology (e.g. reliance on hunted versus other foods). Fertility patterns are not explicitly considered in the present study, but they can influence both somatic condition, particularly for women (Jasienska, 2020; Ryan et al., 2018; Stieglitz et al., 2015a; Stieglitz et al., 2019; Ziomkiewicz et al., 2016), and economic productivity (Hooper et al., 2015; Kramer, 2005). Extensive longevity may be a novel life history feature of modern H. sapiens that was absent among prior hominin species; extant hunter-gatherers with relatively minimal acculturation exhibit a modal adult lifespan of 68–78 years (Gurven and Kaplan, 2007). It is thus possible that our estimates of disability-related productivity loss are more relevant for understanding lifeways of past hominin populations with more similar life histories as modern H. sapiens and perhaps some archaic humans (e.g. H. heidelbergensis), but less applicable to Lower Paleolithic and earlier hominins with shorter adult lifespans. Whether lifespans were too short for some fossil hominins to expect significant post-reproductive longevity is vigorously debated (see Gurven and Kaplan, 2007; Hawkes, 2003; Hill et al., 2007 and references therein). One study using dental-wear seriation and relative macro-age categories (ratio of old to young) demonstrated an increase in the relative presence of older adults from Australopithecines to early Homo and Early Upper Paleolithic humans (Caspari and Lee, 2004; but see Hawkes and O'Connell, 2005; Minichillo, 2005). Re-estimation of several paleo-mortality curves based on hazard analysis and maximum likelihood methods shows a life course pattern more similar to that of modern human hunter-gatherers than previous methods (Konigsberg and Herrmann, 2006). Development of reliable paleo-demographic reconstructions of hominin life histories (e.g. see DeWitte and Stojanowski, 2015) are critical for determining generalizability of the results presented here. Moreover, disability-related productivity losses could change based on variation in ancestral hominin vertebral morphology that can affect interpretation of fracture results using Genant’s semi-quantitative technique (Trinkaus, 2018).

Conclusion

In comparative cross-species perspective, modern human life histories are characterized by delayed peak foraging efficiency (e.g. Kaplan, 1994; Kaplan, 1997; Koster et al., 2020; Walker et al., 2002), complex cooperative strategies to produce and rear altricial, slow-growing offspring (e.g. Hawkes, 2003; Kaplan et al., 1985; Richerson and Boyd, 2008; Wood and Marlowe, 2013), low adult mortality rate (e.g. Hill et al., 2001; Wood et al., 2017), and long post-reproductive lifespan (e.g. Gurven and Kaplan, 2007; Hawkes et al., 1998; Kaplan et al., 2010). Resource transfers from older to younger generations help mitigate risk of production deficits in a skill- and strength-based extractive foraging niche (Gurven et al., 2012; Hawkes et al., 1998; Hooper et al., 2015; Kaplan et al., 2010). In this paper we considered energetic deficits associated with compromised bone strength, which may have been a source of disability and constrained economic productivity and resource transfers for extended periods throughout much of human history (Bailey, 1991; Gurven et al., 2000; Lambert and Welker, 2017; Sugiyama, 2004). While we provide suggestive evidence of a causal relationship between bone properties and functional disability, changes in bone strength may reflect a combination of earlier life factors (e.g. trauma and activity levels) or other senescent processes.

Perhaps not surprisingly, Ache hunters cite fear of disability as an important reason why they willingly conform to band-wide conventions of need-based food sharing (Hill and Hurtado, 2009). It has been proposed that individuals may have survived periods of prolonged disability only via costly provisioning and care by others (Gurven et al., 2000; Sugiyama and Chacon, 2000), although the extent to which others adjust their activity budgets, food production, and sharing to compensate for disability-related productivity loss is beyond the scope of the present study. Our findings nevertheless suggest that disability is economically costly.

Vertebral deformities are observed in diverse hominin skeletal samples yet the social and behavioral implications of these deformities remain obscure. In the absence of direct evidence of disability-related productivity loss in the fossil record, bone properties may represent a potential proxy for inferring behavioral consequences of skeletal deformity in prior hominin populations. Quantification of microstructural variation in the appendicular and axial skeleton of fossil hominins may provide evidence for inferring foraging behavior, as detailed in the present study. While our ability to make inferences about the past using data collected in contemporary populations is limited, studies of contemporary small-scale subsistence populations like the Tsimane can offer a unique opportunity to examine potential consequences of skeletal deformity in a population-representative adult sample. The present study provides a framework for interpreting morphological signals identified in the fossil record; while this study is narrowly focused on specific morphological features, and while it overlooks important subsistence tasks performed by women, it can nonetheless be an exemplar for future ethno-archaeological inquiry into potential economic and social consequences of disability in past human populations.

Appendix

Methods

Study design and participants

Since 2002 the Tsimane (pop. ~16,000) have participated in the ongoing Tsimane Health and Life History Project (THLHP; see Gurven et al., 2017b for an overview). All Tsimane residing in study villages are eligible to participate and most participate at least once. THLHP physicians have conducted annual medical exams on Tsimane of all ages since 2002. A mobile team of physicians and Tsimane research assistants collects data in villages on medical histories, functional ability, and aspects of lifestyle (e.g. food production and sharing).

Between July 2014 and September 2015, men and women aged 40+ years from 59 Tsimane villages were invited to participate in the CT scanning project. At the time, there were 1214 eligible adults living in these villages (the only eligibility criteria were being 40+ years old, self-identifying as Tsimane, and willing to participate). A total of 731 adults were present in their villages at the time and subsequently received a CT scan. Transporting participants from their village to the nearby market town of San Borja was logistically complicated (requiring trekking through the forest, dug-out canoes, rafts propelled by poles pushed off the river bottom, trucks, and cars) and can require up to 2 days of travel each way. From San Borja to the Beni department capital of Trinidad (where the hospital containing the CT scanner is located) is an additional 6 hr car ride. Due to these logistical complications, participants not in their village at the time we arrived were not sampled. The Tsimane are semi-mobile and often build secondary houses deep in the forest near their horticultural plots, not returning to their village for extended periods of time. Hunting and fishing trips can last days or weeks, and some men engage in wage labor in San Borja or elsewhere (e.g. rural cattle ranches). In an average village, approximately one-third of individuals are away hunting, fishing, working in their horticultural plots, or in San Borja at any given time. Additionally, a major flood in February 2014 resulted in mass migration from some villages, and the creation of several new villages that were not sampled as part of this study, further reducing the number of individuals that could be sampled in this study. To address potential sources of sample bias, analyses comparing Tsimane who received CTs and those who did not but who participated in the THLHP’s medical exams in Tsimane villages were conducted. There were no significant differences in sex, blood pressure, or body fat (see Kaplan et al., 2017 for these and additional participant details) and thus the CT sample analyzed here is thought to be representative of all Tsimane aged 40+ years.

Of the 731 Tsimane who received a CT scan, CT data from 506 (69%) were selected with no particular criteria to estimate thoracic vertebral body bone mineral density (BMD) and fracture prevalence and severity (radiologist time constraints precluded analysis of all 731 adults). Of the 506 Tsimane contributing BMD and fracture data, 14 (3%) were missing data on subsistence involvement (e.g. whether still able to hunt or walk all day) because of absent or sick THLHP personnel who were unable to collect data, resulting in a final sample of 493 adults. Among these 493 adults, some anthropometric data were missing because of missing or broken equipment (see Appendix 1—table 1 for sample sizes and descriptives for study variables).

Appendix 1—table 1
Descriptives for study variables*.
VariableNMeanSDMinMax
Thoracic vertebral body BMD (mg/cm3)493165.541.368.9315.0
Any thoracic vertebral body (T6–T12) fracture
(proportion grade ≥1)
4930.280.4501
Any thoracic vertebral body (T6–T12) fracture
(proportion grade ≥2)
4930.090.2901
Age (years)49355.99.941.291.0
Sex (proportion male)4930.520.501
Height (cm)489156.17.6136.0176.3
Weight (kg)48958.79.834.696.9
Body fat (%)44521.58.05.046.7
Fat mass (kg)44512.96.11.942.1
Fat-free mass (kg)44545.87.927.873.1
  1. *Data were missing for various reasons (see Appendix for details).

Thoracic vertebral body BMD

CT-measured vertebral body BMD is increasingly used for osteoporosis screening in vivo because of its ability to provide three-dimensional information compared to traditional dual X-ray absorptiometry two-dimensional images.

Intra-observer variability in BMD measurements using the same measurement technique in a different sample (see Budoff et al., 2010) was tested on 120 scans by one radiologist, with 1 week intervals between the two readings. To measure inter-observer variability on 67 randomly selected scans, the results obtained by two radiologists who were blinded to all clinical information were compared using this other sample. Intra- and inter-observer variations in BMD measurements were 2.5% and 2.6%, respectively.

Thoracic vertebral body fracture

Genant’s semi-quantitative technique does not distinguish between wedge (i.e. reduced anterior height), biconcave (i.e. reduced central height), or crush (i.e. reduced posterior height) fractures; most fractures contain a combination of these features and are influenced by the local biomechanics of the spinal level involved. Vertebral body fractures are differentiated from other, non-fracture deformities (e.g. osteoarthritis), although these other deformities were not systematically coded.

Socio-demographics and anthropometrics

Birth years were assigned based on a combination of methods described in detail elsewhere (Gurven et al., 2007), including using known ages from written records, relative age lists, dated events, photo comparisons of people with known ages, and cross-validation of information from independent interviews of kin. Each method provides an independent estimate of age, and when estimates yielded a date of birth within a 3-year range, the average was generally used.

Results

Hunting cessation is associated with thoracic vertebral body fracture and, albeit weakly, with lower BMD

Appendix 1—table 2
Percentage of men (95% CI) who completely ceased hunting by age and thoracic vertebral body fracture status (fracture grade ≥2).
Age category (years)% ceased hunting with fractureN% ceased hunting without fractureN
40–490 (-----)43 (<1–6)77
50–5921 (<1–46)1414 (6–22)78
60–6960 (<1–1)532 (19–45)50
70+100 (-----)673 (53–93)22
Total41** (22–60)2920 (15–25)227
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact Test [vs. no fracture]).

    Max age = 83 years (mean ± SD = 74 ± 4).

  2. Max age = 83 years (mean ± SD = 74 ± 3).

Appendix 1—table 3
Binary logistic regression: effect of thoracic vertebral body fracture on the probability of hunting cessation after adjusting for age (model 1; n = 256 men).

Model 2 additionally includes as a covariate thoracic vertebral body bone mineral density (BMD). Shown are odds ratios (95% CIs); continuous variables are z-scored.

ParameterModel 1Model 2
(+BMD)
Any thoracic vertebral body fracture (grade ≥1; vs. none)7.3***(3.3–17.6)7.4***(3.3–18.2)
Age (years)5.2***(3.4–8.5)4.1***(2.5–7.0)
Thoracic vertebral body BMD (mg/cm3)-----0.62*(0.38–0.99)
AIC178.55176.61
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (refer to main text for FDR q-values)

Hunting cessation associated with thoracic vertebral body fracture and lower BMD generates substantial productivity losses

Fracture
Appendix 1—table 4
Tsimane men’s age-specific daily hunting production by thoracic vertebral body fracture status.
Age (years)(A) Hunt cals/day(B) Probability still hunting: no fracture(C) Probability still hunting: fracture(D) Hunt cals/day: no fracture§(E) Hunt cals/day: fracture
4018720.9950.96818641812
4118840.9950.96218741813
4218880.9940.95518751803
4318820.9920.94718681783
4418700.9910.93818531754
4518550.9890.92718351720
4618360.9870.91518131680
4718150.9850.90017871634
4817900.9820.88417581582
4917620.9790.86517251524
5017310.9750.84316871459
5116970.9700.81916461389
5216590.9650.79216011314
5316190.9580.76215511233
5415750.9510.72914981148
5515280.9420.69414401060
5614780.9320.6561378969
5714250.9200.6161311878
5813690.9060.5741241786
5913100.8910.5311166696
6012470.8730.4881088609
6111810.8520.4451007526
6211130.8290.403923449
6310410.8030.362836377
649660.7750.323748313
658880.7430.287660255
668070.7090.253572204
677230.6720.222486160
686360.6330.193402123
695460.5920.16832392
704520.5490.14524966
713560.5060.12518044
722570.4630.10711928
731540.4210.0926514
74490.3790.078194
7500.3390.06700
7600.3020.05700
7700.2670.04800
7800.2340.04100
7900.2050.03500
8000.1780.02900
  1. Predicted values (loess fit). From Jan 2005 to Dec 2009 adults were interviewed once or twice per week about time allocation and production for each co-resident individual over age 6 during the previous 2 days (n = 1245 individuals from 11 villages).

    Predicted from binary logistic regression of whether one still hunts on thoracic vertebral body fracture status (i.e. grade ≥1; vs. none) and age.

  2. §Derived by multiplying value in column A by value in column B.

    Derived by multiplying value in column A by value in column C.

Lower BMD
Appendix 1—table 5
Tsimane men’s age-specific daily hunting production by thoracic vertebral body fracture status and thoracic vertebral body bone mineral density (BMD).

For illustrative purposes we report estimates holding BMD at +1 SD and −1 SD of the mean.

Age (years)(A) Hunt cals/day*(B) Probability still hunting: no fracture and +1 SD BMD(C) Probability still hunting: no fracture and
−1 SD BMD
(D) Probability still hunting: fracture and +1 SD BMD(E) Probability still hunting: fracture and
−1 SD BMD
(F) Hunt cals/day: no fracture and +1 SD BMD(G) Hunt cals/day: no fracture and −1 SD BMD§(H) Hunt cals/day: fracture and +1 SD BMD(I) Hunt cals/day: fracture and −1 SD BMD**
4018720.9960.9900.9710.9301865185318191741
4118840.9950.9880.9670.9201875186218221732
4218880.9950.9870.9620.9081878186318161714
4318820.9940.9840.9560.8951870185318001684
4418700.9930.9820.9500.8801857183617761646
4518550.9920.9790.9420.8641840181617471602
4618360.9900.9760.9330.8451819179217141552
4718150.9890.9720.9240.8251795176416761497
4817900.9870.9680.9130.8021767173216331436
4917620.9850.9630.9000.7781736169715861371
5017310.9830.9570.8860.7511701165715331301
5116970.9800.9510.8700.7231664161414761226
5216590.9770.9430.8520.6921621156514141148
5316190.9740.9350.8330.6601576151413481068
5415750.9700.9250.8110.626152714581278986
5515280.9650.9150.7880.591147413971203902
5614780.9600.9020.7620.554141813341126819
5714250.9530.8880.7340.518135912661046738
5813690.9460.8730.7040.48112961195964658
5913100.9380.8560.6720.44412291121881582
6012470.9290.8360.6390.40811591043797509
6111810.9190.8150.6040.3731085963714440
6211130.9070.7920.5680.3391010881633377
6310410.8940.7660.5320.307931798554319
649660.8790.7390.4950.276849714478267
658880.8620.7090.4580.248766630407220
668070.8440.6780.4220.221681547340178
677230.8230.6450.3860.197595466279142
686360.8010.6100.3520.174509388224111
695460.7760.5750.3190.15442431417484
704520.7500.5380.2880.13633924313061
713560.7210.5010.2580.1192571789242
722570.6900.4640.2310.1051771195927
731540.6580.4280.2060.092101663214
74490.6240.3920.1830.080311994
7500.5880.3580.1620.0700000
7600.5520.3240.1430.0610000
7700.5150.2930.1250.0530000
7800.4790.2630.1100.0460000
7900.4420.2360.0960.0400000
8000.4060.2100.0840.0350000
  1. *Predicted values (loess fit). From Jan 2005 to Dec 2009 adults were interviewed once or twice per week about time allocation and production for each co-resident individual over age 6 during the previous 2 days (n = 1245 individuals from 11 villages).

    Predicted from binary logistic regression of whether one still hunts on thoracic vertebral body fracture status (i.e. grade ≥1; vs. none), BMD (mg/cm3) and age (years).

  2. Derived by multiplying value in column A by value in column B.

    §Derived by multiplying value in column A by value in column C.

  3. Derived by multiplying value in column A by value in column D.

    **Derived by multiplying value in column A by value in column E.

Tree chopping cessation is associated with thoracic vertebral body fracture but not lower BMD

Appendix 1—table 6
Binary logistic regression: effect of thoracic vertebral body fracture on the probability of tree chopping cessation after adjusting for age (model 1; n = 256 men).

Model 2 additionally includes as a covariate thoracic vertebral body bone mineral density (BMD). Shown are odds ratios (95% CIs); continuous variables are z-scored.

ParameterModel 1Model 2
(+BMD)
Any thoracic vertebral body fracture (grade ≥1; vs. none)6.9***(3.1–16.6)6.8***(3.1–16.6)
Age (years)8.0***(4.9–13.9)6.8***(4.0–12.4)
Thoracic vertebral body BMD (mg/cm3)-----0.75 (0.47–1.16)
AIC183.41183.75
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (refer to main text for FDR q-values).

Weaving cessation is not associated with thoracic vertebral body fracture but is associated, albeit weakly, with lower BMD

Appendix 1—table 7
Binary logistic regression: effect of thoracic vertebral body fracture on the probability of weaving cessation after adjusting for age (model 1; n = 237 women).

Model 2 additionally includes as a covariate thoracic vertebral body bone mineral density (BMD). Shown are odds ratios (95% CIs); continuous variables are z-scored.

ParameterModel 1Model 2
(+BMD)
Any thoracic vertebral body fracture (grade ≥1; vs. none)2.2 (0.8–6.4)1.8 (0.6–5.4)
Age (years)4.9***(3.1–8.6)3.3***(1.8–6.6)
Thoracic vertebral body BMD (mg/cm3)-----0.51^(0.23–1.05)
AIC134.52133.18
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (refer to main text for FDR q-values).

Appendix 1—table 8
Self-reported reasons for weaving cessation by thoracic vertebral body fracture status.

Shown are percentages of women reporting a given reason.

Reason for weaving cessationFracture
(n = 9)
No fracture
(n = 24)
Total
(n = 33)
Problem with hips100100100
Problem with back100100100
Problem with fingers100100100
Problem with hands (other than fingers)100100100
Any problem with hips, back, or hands100100100
Difficulty sitting100100100
Difficulty seeing898888
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact test [vs. no fracture]).

Restricted day range is associated with thoracic vertebral body fracture and, albeit weakly, with lower BMD

Appendix 1—table 9
Binary logistic regression: effect of thoracic vertebral body fracture on the probability of not being able to walk a full day after adjusting for age and sex (model 1; n = 493 adults).

Model 2 additionally includes as a covariate thoracic vertebral body bone mineral density (BMD). Shown are odds ratios (95% CIs); continuous variables are z-scored. Interaction terms between sex and either fracture, age or BMD do not yield significant parameter estimates and are not shown.

ParameterModel 1Model 2
(+BMD)
Any thoracic vertebral body fracture (grade ≥1; vs. none)8.2***(4.8–14.5)7.8***(4.5–13.8)
Age (years)4.0***(3.1–5.4)3.4***(2.4–4.8)
Sex = male0.19***(0.11–0.31)0.21***(0.12–0.34)
Thoracic vertebral body BMD (mg/cm3)-----0.77^(0.57–1.03)
AIC479.60478.52
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (refer to main text for FDR q-values).

Suggestive evidence of fracture preceding task cessation, rather than the reverse

Appendix 1—table 10
Self-reported reasons for hunting cessation by thoracic vertebral body fracture status.

Shown are percentages of men reporting a given reason.

Reason for hunting cessationFracture
(n = 34)
No fracture
(n = 22)
Total (n = 56)
Problem with hips29**‡018
Problem with back535955
Problem with arms122316
Problem with legs565555
Any problem with hips, back, or limbs778680
Feels weak29^5038
Tires easily596461
Tires easily or feels weak657770
Difficulty hearing385043
Difficulty seeing627768
Difficulty hearing or seeing658271
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact test [vs. no fracture]).

    Fracture data are missing for one man. For another man self-reported data are missing.

  2. FDR q = 0.012.

    Note: only q-values ≤0.05 are reported.

Appendix 1—table 11
Self-reported reasons for tree chopping cessation by thoracic vertebral body fracture status.

Shown are percentages of men reporting a given reason.

Reason for tree chopping cessationFracture
(n = 41)
No fracture
(n = 35)
Total (n = 76)
Problem with hips29*‡920
Problem with back597164
Problem with arms66^8374
Problem with legs292628
Any problem with hips, back, or limbs839488
Feels weak545454
Tires easily637167
Tires easily or feels weak808382
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact test [vs. no fracture]).

    Fracture data are missing for one man. For another man self-reported data are missing.

  2. FDR q = 0.036.

    Note: only q-values ≤0.05 are reported.

Appendix 1—table 12
Self-reported reasons for inability to walk all day by thoracic vertebral body fracture status.

Shown are percentages of adults (pooled sexes) reporting a given reason.

Reason for inability to walk all dayFracture
(n = 95)
No fracture
(n = 134)
Total (n = 229)
Problem with hips393939
Problem with back616966
Problem with arms888
Problem with legs818684
Any problem with hips, back, or limbs889391
Feels weak293131
Tires easily716065
Tires easily or feels weak766972
  1. ^p≤0.1, *p≤0.05, **p≤0.01, ***p≤0.001 (χ² or Fisher’s Exact test [vs. no fracture]).

    Fracture data are missing for one adult.

References

  1. 1
    Light and Broken Bones: Examining and Interpreting Bone Loss and Osteoporosis in Past Populations
    1. S Agarwal
    (2008)
    In: M Katzenberg, S Saunders, editors. Biological Anthropology of the Human Skeleton. Hoboken, New Jersey: Wiley. pp. 387–410.
    https://doi.org/10.1002/9780470245842
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
    The Behavioral Ecology of Efe Pygmy Men in the Ituri Forest, Zaire
    1. R Bailey
    (1991)
    Ann Arbor: University of Michigan, Museum of Anthropology.
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
    The optimal timing of teaching and learning across the life course
    1. MD Gurven
    2. RJ Davison
    3. TS Kraft
    (2020b)
    Philosophical Transactions of the Royal Society B: Biological Sciences 375:20190500.
    https://doi.org/10.1098/rstb.2019.0500
  45. 45
  46. 46
  47. 47
  48. 48
    Time, efficiency, and fitness in the Amazonian protein quest
    1. R Hames
    (1989)
    Research in Economic Anthropology 11:43–85.
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
    Cooperative breeding in South American hunter–gatherers
    1. K Hill
    2. AM Hurtado
    (2009)
    Proceedings of the Royal Society B: Biological Sciences 276:3863–3870.
    https://doi.org/10.1098/rspb.2009.1061
  56. 56
  57. 57
    The Structure of Energy Production and Redistribution Among Tsimane' Forager-Horticulturalists
    1. P Hooper
    (2011)
    Albuquerque: University of New Mexico.
  58. 58
    Social Underpinnings of Human Biodemography. Advances in Biodemography: Cross-Species Comparisons of Social Environments and Social Behaviors, and Their Effects on Health and Longevity
    1. P Hooper
    2. M Gurven
    3. H Kaplan
    (2014)
    Washington, D.C: National Academy Press.
  59. 59
  60. 60
  61. 61
    Costs of reproduction and ageing in the human female
    1. G Jasienska
    (2020)
    Philosophical Transactions of the Royal Society B: Biological Sciences 375:20190615.
    https://doi.org/10.1098/rstb.2019.0615
  62. 62
  63. 63
  64. 64
  65. 65
    The evolution of the human life course
    1. H Kaplan
    (1997)
    In: C Finch, editors. Between Zeus and the Salmon: The Biodemography of Aging. Washington, D.C: National Academy of Sciences. pp. 175–211.
    https://doi.org/10.17226/5740
  66. 66
  67. 67
    The evolutionary and ecological roots of human social organization
    1. HS Kaplan
    2. PL Hooper
    3. M Gurven
    (2009)
    Philosophical Transactions of the Royal Society B: Biological Sciences 364:3289–3299.
    https://doi.org/10.1098/rstb.2009.0115
  68. 68
  69. 69
  70. 70
  71. 71
    The osteological evidence for human longevity in the recent past
    1. L Konigsberg
    2. N Herrmann
    (2006)
    In: K Hawkes, R Paine, S Fe, editors. The Evolution of Human Life History. School of American Research Press. pp. 267–306.
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
    Successful Aging: A Neuroscientist Explores the Power and Potential of Our Lives
    1. D Levitin
    (2020)
    New York: Dutton.
  80. 80
    The Story of the Human Body: Evolution, Health, and Disease
    1. D Lieberman
    (2013)
    New York: Vintage.
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
  91. 91
    Getting old at Gombe: skeletal aging in wild-ranging chimpanzees
    1. M Morbeck
    2. A Galloway
    (2002)
    In: J Erwin, P Hof, editors. Interdisciplinary Topics in Gerontology: Aging in Nonhuman Primates. Basel: Karger. pp. 48–62.
  92. 92
  93. 93
  94. 94
  95. 95
  96. 96
  97. 97
  98. 98
  99. 99
    Two disparate instances of healed cranial trauma from the later stone age of South Africa
    1. S Pfeiffer
    (2012)
    South African Archaeological Bulletin 67:256–261.
  100. 100
  101. 101
  102. 102
  103. 103
    Not by Genes Alone: How Culture Transformed Human Evolution
    1. PJ Richerson
    2. R Boyd
    (2008)
    University of Chicago Press.
  104. 104
  105. 105
  106. 106
  107. 107
  108. 108
  109. 109
  110. 110
  111. 111
    Work to live and live to work: Productivity, transfers, and psychological well-being in adulthood and old age
    1. J Stieglitz
    2. A Jaeggi
    3. A Blackwell
    4. B Trumble
    5. M Gurven
    6. H Kaplan
    (2014)
    In: M Weinstein, M Lane, editors. Sociality, Hierarchy, and Health: Comparative Biodemography. Washington DC: National Academies Press. pp. 197–221.
  112. 112
  113. 113
  114. 114
  115. 115
  116. 116
  117. 117
  118. 118
  119. 119
    Effects of Illness and Injury on Foraging Among the Yora and Shiwiar: Pathology Risk as Adaptive Problem
    1. L Sugiyama
    2. R Chacon
    (2000)
    In: L Cronk, N Chagnon, W Irons, editors. Human Behavior and Adaptation: An Anthropological Perspective. New York: Aldine. pp. 371–395.
  120. 120
  121. 121
  122. 122
  123. 123
    Theory and Practice in the Bioarchaeology of Care
    1. L Tilley
    (2015)
    Springer International Publishing.
  124. 124
  125. 125
  126. 126
  127. 127
  128. 128
  129. 129
  130. 130
  131. 131
    The evolution of hunting
    1. S Washburn
    2. C Lancaster
    (1968)
    In: R Lee, I Devore, editors. Man the Hunter. Chicago: Aldine. pp. 293–303.
  132. 132
  133. 133
  134. 134
  135. 135
    Human adaptation to the control of fire
    1. R Wrangham
    2. R Carmody
    (2010)
    Evolutionary Anthropology: Issues, News, and Reviews 19:187–199.
    https://doi.org/10.1002/evan.20275
  136. 136

Decision letter

  1. George H Perry
    Senior and Reviewing Editor; Pennsylvania State University, United States
  2. Ian Wallace
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study examines relationships among bone fragility, disability, and subsistence costs in a forager-horticulturalist society, the Tsimane (Bolivia). Vertebral fracture and lower bone mineral density are associated with substantial productivity losses from the cessation of hunting, tree chopping, and long-distance walking activities. These results help improve our understanding of the potential complex selective pressures associated with human longevity and our species' propensity for age-related fragility.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Productivity loss associated with physical impairment in a contemporary small-scale subsistence population" for consideration by eLife. Your article has been reviewed by three peer reviewers, including George H Perry as the Senior and Reviewing Editor and Reviewer #1, The following individual involved in review of your submission has agreed to reveal their identity: Susan Pfeiffer (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife at this time. I do not want to completely close the door on our future consideration of this work, as we truly liked much about it and we believe that it has much promise. However, the consensus outcome of our reviewer consultation process was that at the least some major core work is necessary that likely exceeds the scope of revision acceptable in the eLife model, and resulting in substantial changes to the paper requiring a new full review if you were to choose to submit a new manuscript to eLife. Note that while the reviewers are enthusiastic in some respects they wanted to emphasize the fundamental nature of some of the concerns that they feel need to be fully considered and addressed.

Reviewer #1:

I'm reviewing this manuscript from a general biological anthropology perspective only, without direct expertise in bone health measures and interpretations of those data, so I will default to other reviewers accordingly.

I thought that the overall study was fairly solid and insightful. The authors integrate ethnographic data on Tsimane activity and economy with thoracic vertebrae bone mineral density estimates (and fracture incidence) from computed tomagraphy (CT) scans for a large sample of n=493 individuals.

1) The beginning investigations of correlation vs. causation are valuable and interesting, but I do not think that they are definitive to warrant the causative statements made by the authors throughout the paper. I think that such statements should be changed to “associated with”, with discussion of correlation/causation supported by the post-hoc analyses that help in this direction (but are not definitive, given so many other things that cannot be accounted for) in the Discussion.

2) I don't agree with the practice applied throughout the paper of discretizing the data by age categories and positive versus negative BMD residuals, etc., for analyses and visualization. Such data-reducing discretization should be limited as much as possible; please analyze and show the full data. Also, anytime histograms are necessary or useful for visualization, use jitter to show the underlying data points.

3) The multiple testing burden is also not being considered at present. The authors should compute false discovery rate q-values for all of the tests conducted within each major hypothesis test and report these as the main probability estimate.

4) I would also like to have a visualization of the hunting activity data alongside the BMD and fracture data as a function of age. Presenting the results of the models are fine but these are the critical data of the study and there needs to be an accompanying strong data visualization. What is being shown in Figures 3, 4, and 5? If these are summary models they should only be shown as supporting panels to main figures with the actual data. The quality of the figures in general in the paper needs to be improved.

Reviewer #2:

The evolutionary hypotheses are of interest with relevant objectives. The statistical analyses are rigorous and correctly utilized.

However, there are several key issues with the overarching design and interpretation of the study methods (bone health as a proxy for impairment) and wider contextualization of archaeological and evolutionary studies of bone health that are needed in this study in order to be published.

Biological contextualization and understanding of the method of bone analysis is lacking, and discussion of the finding of poor bone health (low bmd, fracture in youngish age males) needs to be given. While I realize this (bone biology/osteoporosis data and interpretation) is not the focus of this study, not including this discussion/data causes the study design and method to appear flawed, and in many places the paper has sweeping statements that are simply incorrect. The measure of BMD is a crude measure, and data for different groups (sexes, and ages) are likely needed to be considered or discussed separately as they have likely different etiologies and different outcomes/costs because of this. Readership is perhaps better suited to a specialty journal (ex Bone, JBMR) but this data in current form is not rigorous to be published there.

Specifically:

Adult physical impairment (API) is described as degenerative or other origin (e.g. trauma, infection). But this is a very wide statement – this includes a multitude of pathological conditions, that are obviously part of the human (long-lived vertebrate/mammalian?) condition – so of course these things will hinder foraging in any animal. The authors go on to cite “specifically” degenerative conditions that are osteoarthritic as things that are API in human evolution, and specifically cause back pain (the closest link to this study) but then many other conditions and diseases (trauma, infectious, degenerative, developmental….):

1) There is an entire discipline (disability studies) and a multitude of social science data studies that have critiqued the use of all disease and all pathology as indicative of impairment, or worse the use any disease, as a proxy for lack of productivity and/or ability. I strongly suggest the authors "pick" a disease to build a better case for why this maker (in this case low bmd and spinal fracture) can be used as indicative of impairment.

2) The authors do not build legitimate case or correct lit review that low bmd and mild fracture is related to back pain. In fact, until there is severe vertebral collapse the correlation to clinical back pain is not usually made and is not consistent. What is linked to back pain and impairment very clearly is spinal osteoarthritis – a pathology that has nothing to do with BMD or osteopenic fracture. The authors go on to cite in fact spinal arthritis studies that found in hominin studies (Chapman; Cook; Trinkaus etc);

More inconsistencies in data design and discussion:

Why are thoracic vertebrae "are more common among Tsimane than matched US (Los Angeles) controls "? It is not true that thoracic fractures are the most common in human populations – lower thoracic and upper lumbar are. Is this why the Tsimane look higher %- as in US controls lumbar fractures are higher? Why did the authors not look at lumbar vertebra: The CT work in this study was part of an atherosclerosis study of the thoracic cage- so lumbar vertebrae that are the most diagnostic of bone loss impairment we not included in this study.

It is relevant to know why Tsimane males in particular have low bmd and thoracic fracture – this is not an expected age-related phenomena. Is the low bmd a sign of another causative pathology that is causing the impairment of activity and pain? If so, low bmd is only a secondary measure of an underlying impairment that may not have to do with bone health at all – maybe dietary, parasitic, related to blood disorder that causes bone loss and impairment? Is this not relevant to discuss or mention?

What is going on with male bone loss and female bone loss is clearly different, as discussed by authors that point to previous studies that show female bone loss is related to post-menopausal or reproductive reasons. And both sexes show different "costs" of the bone health measure as API – they sexes needed to be discussed differently – and again why the "cause” of the male bone loss is important to explain. They likely have different etiology in both sexes, different costs associated because of this.

Again wide sweeping statements with no basis for any meaningful interpretation: "Sixty-eight percent of adults age 50+ are also clinically diagnosed with some musculoskeletal problem by Tsimane Health and Life History Project (THLHP)" and "some physical discomfort during activities". – what is "musculoskeletal problem" and how can we use this as an equivalent of the BMD measure used in this study?

Statements that are not true or well researched/cite:

"Ample bioarchaeological evidence of partially or fully healed fractures (e.g. (Lovejoy and Heiple, 1981; Pfeiffer, 2012)) and developmental abnormalities (Cowgill et al., 2015; Trinkaus, 2018) suggests longer-term survival with varying levels of disability.” In fact, numerous bioarcheological studies have shown that “osteopenic” fractures with low BMD are very rare in the past – traumatic/violent fractures are very common

"Tsimane thoracic vertebral fractures appear to result from both trauma and senescence (Stieglitz et al., 2019) and are not just prevalent among the oldest adults.”

So are we looking at violence or accidental trauma fracture in males here – or related to another primary disease? This needs to be discussed/expanded. It is not senescence in young age males.

Again – both sexes -age groups have different etiologies of low BMD and likely different outcomes/costs because of this.

Reviewer #3:

The submitted manuscript describing productivity loss linked to bone mass decline is well written, novel and pertinent in several domains. It is certainly a worthwhile contribution. The suggestions made here for adjustments and improvements reflect this reviewer's responses to the work. The clear statements at beginning and end regarding the importance of this kind of research for bioarchaeological and paleoanthropological studies is welcome.

Perhaps the broadest concern with the current version of the manuscript is the tendency to under-play the imbalance of impact of bone mass loss between men and women. Clearly, the paper's focus is on hunting and fishing, which is all good. However, from Abstract to conclusions the different pattern in women is underplayed. The type of loom(s) used by the weavers could be pertinent to the study and should be specified. The brief section on the lack of significant effect on weaving should be moved up so that it precedes the section on long distance walking. The relative inattention to women's productivity is a related concern. There should be acknowledgment, as the authors discuss the limitations of the work, of the fact that their data are at least somewhat inconsistent with modern industrialized countries' clinical data, with regard to sex patterns in BMD and the disability load associated with compression fractures.

The research presented here links to a research area of which the authors appear to be unaware. In a 1996 paper (Anatomical Record 246:4:423-432), Agarwal and Grynpas asserted that osteoporotic fracture is absent in past populations. While it has only been cited 41 times, those citations include recent work. I urge the authors to become better informed on just how novel – and helpful – their work is, in a bioarchaeological context. In my opinion, the best explanation for the "pattern" reported in that 1996 paper is that no one ever systematically studies thoracic compression fractures. Nevertheless, there is a literature arguing that osteoporosis is a modern disease (cf. Kralick and Zemel, 2020 Frontiers in Endocrinology).

With regard to which vertebrae form the basis of the dataset: If T7 or T8 form the starting point for the assessment, the technician might be providing bone mass input from groups as different as T5-T7 to T8-T10. These are very different parts of the thoracic column, from the bone mass perspective. This is a concern that should be addressed, if possible. The morphological assessment of T6 to T12 helps to give the reader more confidence in the data.

The discussion of the contributions of older group members would benefit from stretching beyond physical activities to intellectual assets, along the lines of D.J. Levitin's argument that older adults contribute experience and judgement. The impressive breadth of the framework constructed by the authors could and should be even broader and more interdisciplinary.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Productivity loss associated with physical impairment in a contemporary small-scale subsistence population" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by George Perry as the Senior and Reviewing Editor. The following individual involved in review of your submission has agreed to reveal their identity: Ian Wallace (Reviewer #1).

The reviewers have discussed the reviews with one another and the Senior and Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Due to pandemic-related constraints, the two reviewers besides myself of your related manuscript submission were not able to reconsider this new version. However, each of the two new reviewers were made aware of the previous round of review, had access to your response to the previous reviewer comments, and I took this factor into consideration for this decision letter. I also did re-review the paper myself, though I did not write a formal review given that you had addressed my most substantive comments in your new submission.

One major concern about your previous submission had been the need to avoid conflating association and causation. I noticed how you quite thoroughly addressed this concern in the Results section of your revised manuscript. Thank you. However, interestingly, this was still the overarching major concern noted by the two new reviewers, with this concern now focused on the Discussion section and your interpretations. We had a great consultation discussion on this issue. We collectively do find much to like about the study and paper, but agree that this overarching concern needs to be resolved robustly. The following is a summary of the essential revisions, with the detailed reviewer comments appended below for expanded direction and explicit instruction:

The authors need to 1) reduce statements of implied causation in the discussion (and also in the general setup of the paper regarding the possible causal links between bone fragility and disability) and fully acknowledge all possibilities and 2) carefully establish potential mechanisms for any proposed causality, even when reduced explicitly to speculation. Currently the discussion is written as if the correlation is directional and infers causation, but with so many unstudied covariates (since lower BMD is a function of aging) this is inappropriate. The appropriate null hypothesis for interpretation should be carefully considered. I support some hypothesis generation and speculation in the discussion but this should be clearly indicated as such, and clearly separated from the conclusive interpretations. Even as such, the potential mechanistic links here would need to be much more carefully detailed and supported given the reliance on indirect inference only and the inconsistency with existing literature. Please see the reviewer comments for explicit suggestions and details.

Reviewer #1:

This is a very nice study of associations between bone fragility, disability, and subsistence costs among a small-scale forager-horticulturalist society, the Tsimane of Bolivia. The study builds on a great recent paper in eLife by the same authors showing that the Tsimane (especially women) have lower average vertebral BMD values than might be expected given their high physical activity levels, as well as surprisingly high rates of vertebral fractures. In this study, they investigate the impact of low vertebral BMD and fracture incidence on disability (or, "adult physical impairment" as they call it) and ultimately the economic costs of such disability. The paper nicely frames the study within evolutionary theory and provides novel insights on the selective pressures that may have shaped human longevity and our species' propensity for age-related fragility. The sample analyzed is impressively large, particularly for a remote-living population like the Tsimane. The CT methods used to collect the bone data are the gold standard. The paper is clear and well written, and the interpretations of the results are all reasonable. Also, the figures are nice. Overall, I think this paper deserves consideration for publication in eLife, but I have a couple of concerns.

First, in my opinion, the possible causal links between bone fragility and disability are not considered or described in enough detail. The only mention of such a link is: "Vertebral deformities are reliably associated with chronic pain…(no citation)" Do the authors have data on back pain? If so, the link between bone fragility and disability would be a lot stronger if they could show that bone fragility is indeed associated with chronic pain among the Tsimane. If not, then it remains possible that bone fragility and disability are only indirectly related. For example, perhaps in many cases bone fragility is just one manifestation of a more general somatic decline, and disability is engendered primarily by other aspects of the general phenomenon and not bone fragility per se. If the authors do not have pain data, I suggest they at least make a stronger argument for the link between bone fragility and pain/disability by citing prior research demonstrating such links.

Second, on the same topic of causal links between bone fragility and disability, I believe the authors must provide a more detailed explanation for how they interpret the contribution of low BMD to disability. Independent of a fracture, there is no clear reason why one would expect low BMD to directly affect disability. Again, it is possible that low BMD and disability could be indirectly related if low BMD is only one of multiple outcomes of general somatic decline. But my understanding is that osteoporosis is generally an asymptomatic disease until fracture occurs. My impression is that the authors may not necessarily be arguing that low BMD, independent of fracture, affects disability, but rather that fracture plus low BMD leads to more disability than a fracture alone. Even so, why does this scenario make sense? Are fractures experienced by a bone with low BMD inherently more painful? Maybe the cooccurrence of fracture and low BMD leads to more disability because such fractures are generally more severe. Or, given that bone fractures can trigger bone loss, maybe the cooccurrence of fractures and low BMD represent cases where fractures took place longer ago and thus potentially have a greater effect on chronic pain and disability. An extreme way of resolving this uncertainty would be for the authors to disregard the BMD data entirely and focus just on the fracture data. The overall storyline would not change too much. However, I am not suggesting the authors do this. The BMD data are potentially important. But the links with disability need to be more carefully considered.

Reviewer #2:

This paper aims to estimate the magnitude of productivity losses associated with compromised bone strength among the Tsimane. The authors specifically test "whether a specific morphological characteristic which may indicate physical impairment, i.e. compromised bone strength, is indeed associated with diminished participation in routine subsistence tasks". They "identify economic correlates of compromised bone strength (i.e. thoracic vertebral body fracture and lower thoracic vertebral BMD) by examining involvement in four common tasks in foraging economies: hunting, tree chopping, tool manufacture (i.e. weaving), and walking long distances".

The data presented in this paper are important and have the potential to make a valuable contribution to the literature, and the authors are to be congratulated for conducing important, unique, and valuable fieldwork. The questions they have posed in this paper are interesting and worthy of consideration. That being said, I have some concerns with the analysis, what it assumes, what it excludes, and the conclusions.

The paper is set to test the "probability of completely ceasing to perform a given task increases with compromised bone strength", and estimate the "magnitude of productivity losses (i.e. lost kcals/day) from hunting cessation associated with compromised bone strength" The authors claim "We do not assume that compromised bone strength is itself a proxy for functional impairment or reduced productivity. Instead we posit that individuals with compromised bone strength who are no longer able to perform specific subsistence tasks facultatively adjust their time allocation based on various factors (e.g. phenotypic condition, household demand) to maximize productivity." This is really unclear and woolly. What, specifically then, are they predicting? If they don't think compromised bone strength influences productivity, why are they testing it, and why is the text replete with indications that they are interpreting this as a direct influence?

The authors make their assumptions of the direction of this relationship clear throughout the paper, through the use of language such as "To estimate men's hunting productivity losses associated with fracture and lower BMD". The paper is replete with the inference of this relationship, and while the authors might carefully choose their words in places to try to avoid causation, the inference is there for the reader to make. Perhaps another way of putting this is the following: the authors show correlations between measures of later life activity/productivity and both increase fracture prevalence and lower BMD. There are three parsimonious explanations for this correlation:

1) The loss of bone quality is the cause of a decline in involvement in economic activity

2) The loss of bone quality is correlated with age, which independently corresponds with later life declines in economic activity, and the two are not directly related but collinear

3) A later life decline in activity is the cause of a decline in bone quality.

These are simplistic explanations, and there are likely other factors involved (see comments below), but for the purposes of simplicity, let's consider the above. The language of the paper is only oriented around the first explanation and the direction of causality is assumed through most of the text, but only really directly acknowledged when they briefly address the possibility of “reverse causality” (reductions in activity causing a reduction in bone strength). They attempt to test for “reverse causality” through indirect methods: comparing fracture status to self-reported reasons for task cessation, and the use of GLM to compare BMD of those who cease versus those who continue participation. I'm not convinced that these adequately address the challenges of the direction of the relationship. Self-reported causes may, in some cases, related to injury (particularly in severe cases), but are much more likely to relate to other factors entirely, and to be unrelated to BMD. The GLM comparisons of BMD among groups defined by continued participation in activity may be influenced by study design, and I am not certain it is testing the reverse causation that the authors suggest. The most significant influence on later life bone mass is earlier life activity. Peak bone mass is commonly achieved between the ages of 20 and 30, and is the single most important predictor of later life bone mass. The more you build early on, the more bone you have to lose. Activity at age 20 may thus be stronger correlate of bone mass after the age of 40 than recent activity patterns. It is curious that the authors point out that "Physical strength peaks in the mid to late 20s, then declines continuously thereafter, reaching ~60% of maximum adult capacity by age 70 (Gurven et al., 2006)." This is essentially the same trajectory of bone strength acquisition and loss, and there is considerable evidence that bone strength corresponds with muscle induced loads. One could argue the most parsimonious explanation is that the most active and economically productive individuals during young adulthood have the highest peak bone mass, and are more likely to remain productive (or with higher than average productivity) later in life.

Apart from the issues noted above, I remain sceptical about causation. There is, in theory, a direct causal relationship between fractures and decline or cessation of economic activity. This is supported by the relatively strong results, for example, the declines in weaving and ability to walk for a day associated with earlier fractures. What is the causal mechanism for the relationship with BMD? Going back to the three possible interpretations I mentioned earlier, I would have no issue with the statement of a hypothesis that vertebral fracture is the cause of a decline in involvement in economic activity. If I were to hypothesize what would explain the correspondence with BMD, I would go for a variant of the second hypothesis, that it is correlated with age, which independently correlates with other measures that are associated with later life declines. There are multiple collinear biological trends associated with senescence. While the authors provide no consideration of a biological mechanism through which BMD would be associated directly with productivity, the inference of causation is made without consideration of the particular relationship of declines in BMD associated with other factors of aging.

I am also sceptical of the consideration of both BMD and trauma as indicators of bone strength. BMD is highly correlated with bone mechanical function, this is uncontroversial, but considering trauma as a similar indicator may be problematic. Demographers and epidemiologists have often noted that trauma occurs in different societies with a relatively consistent frequency (at least when compared to infectious or metabolic disease, for example), because they result from accidental injury which tend to occur with some statistical regularity in larger populations. The data in Figure 2 suggest that there is no real difference the relationship between BMD and Age between groups of men with and without fracture, which could be interpreted as evidence that fracture prevalence among is predominantly determined by accidents rather than a loss of bone strength, regardless of the age or BMD of the individual. In contrast, there appear to be considerable differences among the women measured, suggesting that BMD and Fractures are more strongly correlated in women. The male data, at least, would seem to contradict the assertion that fracture prevalence is a proxy for bone strength.

https://doi.org/10.7554/eLife.62883.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

I'm reviewing this manuscript from a general biological anthropology perspective only, without direct expertise in bone health measures and interpretations of those data, so I will default to other reviewers accordingly.

I thought that the overall study was fairly solid and insightful. The authors integrate ethnographic data on Tsimane activity and economy with thoracic vertebrae bone mineral density estimates (and fracture incidence) from computed tomagraphy (CT) scans for a large sample of n=493 individuals.

1) The beginning investigations of correlation vs. causation are valuable and interesting, but I do not think that they are definitive to warrant the causative statements made by the authors throughout the paper. I think that such statements should be changed to “associated with”, with discussion of correlation/causation supported by the post-hoc analyses that help in this direction (but are not definitive, given so many other things that cannot be accounted for) in the Discussion.

We have modified text accordingly (in the Introduction, Results and Discussion), so as not to imply that the study definitively establishes causality.

2) I don't agree with the practice applied throughout the paper of discretizing the data by age categories and positive versus negative BMD residuals, etc., for analyses and visualization. Such data-reducing discretization should be limited as much as possible; please analyze and show the full data. Also, anytime histograms are necessary or useful for visualization, use jitter to show the underlying data points.

We revised all figures based on both this comment and Dr. Perry’s comment below (comment #4). We no longer discretize data by age categories or BMD residuals, and we now show the full raw data for primary bone strength measures (i.e. fracture and BMD; see Figures 2-3).

3) The multiple testing burden is also not being considered at present. The authors should compute false discovery rate q-values for all of the tests conducted within each major hypothesis test and report these as the main probability estimate.

We agree that it is important to consider the multiple testing burden for certain analyses, and appreciate the recommendation. We now report FDR q-values in the revision (in the text and Appendix 1—tables 10-11).

4) I would also like to have a visualization of the hunting activity data alongside the BMD and fracture data as a function of age. Presenting the results of the models are fine but these are the critical data of the study and there needs to be an accompanying strong data visualization. What is being shown in Figures 3, 4, and 5? If these are summary models they should only be shown as supporting panels to main figures with the actual data. The quality of the figures in general in the paper needs to be improved.

The quality of all figures has been enhanced by redoing all the figures from scratch. We agree that it is more intuitive and useful for the reader to show actual production data, rather than productivity differentials. Figures 4-5 now show actual hunting/fishing productivity estimates (both age-specific and expected cumulative future) by fracture status and BMD, rather than the difference in hunting productivity by fracture status and BMD. We do not include raw fracture and BMD data (now shown in Figures 2-3) in the same figure as hunting/fishing productivity data (now shown in Figures 4-5) because we lack all of these data at the individual-level (as explained in Materials and methods). The values shown in Figures 4-5 (reported in Appendix 1—tables 4-5) thus represent synthetic, model derived estimates rather than raw values. Lastly, we deleted the final figure in the original manuscript (old Figure 5) summarizing subsistence involvement (i.e. 1-probability of task cessation) across all tasks by age, fracture status and BMD because it was not critical to our central thesis, and to eliminate data-reducing discretization of BMD based on positive and negative residuals.

Reviewer #2:

The evolutionary hypotheses are of interest with relevant objectives. The statistical analyses are rigorous and correctly utilized.

However, there are several key issues with the overarching design and interpretation of the study methods (bone health as a proxy for impairment) and wider contextualization of archaeological and evolutionary studies of bone health that are needed in this study in order to be published.

We do not consider “bone health” as a proxy for functional ability or impairment: whether and to what extent compromised bone health is associated with impaired productivity are empirical questions which we address in this manuscript (without assuming an answer ex ante). We have clarified our position in the Introduction and Discussion to avoid potential misinterpretations.

Biological contextualization and understanding of the method of bone analysis is lacking, and discussion of the finding of poor bone health (low bmd, fracture in youngish age males) needs to be given.

We make no assessment or judgment about whether specific BMD values or fracture reflects “poor bone health” (we do not impose BMD thresholds [e.g. based on DXA-derived T-scores] denoting osteopenia or osteoporosis). Instead we empirically test for associations between subsistence involvement and bone strength measures, including continuous BMD values, and we use those parameter estimates in our calculations of productivity losses. To derive those calculations we make rather conservative assumptions regarding the relationship between compromised bone strength and productivity (e.g. we do not assume that fracture or lower BMD values are associated with lower foraging efficiency).

While I realize this (bone biology/osteoporosis data and interpretation) is not the focus of this study, not including this discussion/data causes the study design and method to appear flawed, and in many places the paper has sweeping statements that are simply incorrect.

We do not believe any statements in the revised manuscript are incorrect.

The measure of BMD is a crude measure, and data for different groups (sexes, and ages) are likely needed to be considered or discussed separately as they have likely different etiologies and different outcomes/costs because of this.

First, BMD accounts for ~70% of the variance in bone strength and is widely used in research and clinical settings to indicate bone strength. Second, the sexes are indeed analyzed separately, as the sexual division of labor would require. Third, we follow the recommendation of the Senior and Reviewing Editor and treat age as a continuous not categorical variable.

Readership is perhaps better suited to a specialty journal (ex Bone, JBMR) but this data in current form is not rigorous to be published there.

We respectfully disagree, and we thank the Senior and Reviewing Editor for providing us with the opportunity to resubmit.

Specifically:

Adult physical impairment (API) is described as degenerative or other origin (e.g. trauma, infection). But this is a very wide statement – this includes a multitude of pathological conditions, that are obviously part of the human (long-lived vertebrate/mammalian?) condition – so of course these things will hinder foraging in any animal. The authors go on (p 4) to cite “specifically” degenerative conditions that are osteoarthritic as things that are API in human evolution, and specifically cause back pain (the closest link to this study) but then many other conditions and diseases (trauma, infectious, degenerative, developmental….):

We agree that this original description was too wide-ranging, and we thank reviewer #2 for pointing this out. We have narrowed the focus of the Introduction accordingly, so as not to detract attention from the potential impairments that we consider in this paper, i.e., arising from compromised bone strength. When other health conditions (e.g. infection) are mentioned in the Introduction, it is very brief and meant to acknowledge prior studies of extant subsistence populations that have similarly examined associations between somatic condition and productivity.

1) There is an entire discipline (disability studies) and a multitude of social science data studies that have critiqued the use of all disease and all pathology as indicative of impairment, or worse the use any disease, as a proxy for lack of productivity and/or ability. I strongly suggest the authors "pick" a disease to build a better case for why this maker (in this case low bmd and spinal fracture) can be used as indicative of impairment.

We do not view all disease and all pathology as indicative of impairment, nor do we view the use of any disease as a proxy for lack of productivity and/or ability. We never intended to imply this. Whether and to what extent a given condition is associated with impairment or reduced productivity is an empirical question, and one that our study is designed to address. We have clarified our position in the Introduction in light of reviewer #2’s comment.

2) The authors do not build legitimate case or correct lit review that low bmd and mild fracture is related to back pain. In fact, until there is severe vertebral collapse the correlation to clinical back pain is not usually made and is not consistent. What is linked to back pain and impairment very clearly is spinal osteoarthritis – a pathology that has nothing to do with BMD or osteopenic fracture. The authors go on to cite in fact spinal arthritis studies that found in hominin studies (Chapman; Cook; Trinkaus etc);

Back pain is not the focus of our study, and not something we are trying to explain as a direct result of lower BMD or fracture (mild or other). Back pain can also be prevalent and severe in adults whose bone strength is not compromised. Nevertheless, we do analyze data on self-reported reasons for subsistence task cessation, including back problems (Appendix 1—tables 8; 10-12). These data suggest few differences in the experience of back problems (including pain) for adults with vs. without fracture, and indicate a fairly high overall prevalence of back problems. Back pain may be one of several mechanisms linking thoracic vertebral fracture and diminished subsistence involvement.

More inconsistencies in data design and discussion:

Why are thoracic vertebrae "are more common among Tsimane than matched US (Los Angeles) controls "?

The population-level difference in thoracic vertebral body fracture prevalence is now clarified in Materials and methods, Study population and Discussion, although an expanded explanation is beyond the scope of the present study (see Stieglitz et al., 2019, for an expanded explanation).

It is not true that thoracic fractures are the most common in human populations – lower thoracic and upper lumbar are.

We never said “the most common…” but instead we said “among the most common”, which is not inaccurate.

Is this why the Tsimane look higher %- as in US controls lumbar fractures are higher?

In Stieglitz et al., 2019 (see Figure 3) we report that Tsimane (both sexes) show higher age-standardized thoracic vertebral body fracture prevalence than Los Angeles residents, using directly comparable measures of thoracic vertebral fracture generated by the same lab.

Why did the authors not look at lumbar vertebra: The CT work in this study was part of an atherosclerosis study of the thoracic cage- so lumbar vertebrae that are the most diagnostic of bone loss impairment we not included in this study.

This was explained in the original manuscript: lumbar vertebrae do not reliably appear in the CT image field of view. We only CT-scanned the thoracic cage, and not more of the body, to minimize radiation exposure for study participants.

It is relevant to know why Tsimane males in particular have low bmd and thoracic fracture – this is not an expected age-related phenomena. Is the low bmd a sign of another causative pathology that is causing the impairment of activity and pain? If so, low bmd is only a secondary measure of an underlying impairment that may not have to do with bone health at all – maybe dietary, parasitic, related to blood disorder that causes bone loss and impairment? Is this not relevant to discuss or mention?

With respect to BMD, we do not know what reviewer #2 is referring to when he/she states that “Tsimane males in particular have low bmd”. Low relative to what? Actually, Tsimane males do not have lower BMD compared to age-matched Los Angeles males (see Figure 2 in Stieglitz et al., 2019), suggesting that the former do not experience greater risk than the latter of more generalized skeletal fragility. Tsimane males do, however, show higher thoracic vertebral body fracture prevalence (see Figure 3B in the Stieglitz et al., 2019 eLife paper), which might result from high levels of mechanical stress on men’s vertebrae from frequent heavy load carrying. In the Discussion of the present manuscript we elaborate on potential causes of skeletal fragility for Tsimane, though we consider this issue at length in our other published papers (https://elifesciences.org/articles/48607; https://asbmr.onlinelibrary.wiley.com/doi/full/10.1002/jbmr.2730; https://onlinelibrary.wiley.com/doi/abs/10.1002/ajpa.23214; https://onlinelibrary.wiley.com/doi/full/10.1002/ajpa.22681). We elected not to focus more on fracture etiology in the present manuscript because of its focus on potential economic consequences of compromised bone strength.

What is going on with male bone loss and female bone loss is clearly different, as discussed by authors that point to previous studies that show female bone loss is related to post-menopausal or reproductive reasons. And both sexes show different "costs" of the bone health measure as API – they sexes needed to be discussed differently – and again why the “cause” of the male bone loss is important to explain. They likely have different etiology in both sexes, different costs associated because of this.

The revised Discussion addresses some of these sex differences. But as revised Figures 2-3 show, age-associated BMD declines in thoracic vertebral bodies might not be so disparate between sexes (in terms of intercepts and slopes), and we prefer to refrain from mischaracterizing etiology and excessive speculation.

Again wide sweeping statements with no basis for any meaningful interpretation: "Sixty-eight percent of adults age 50+ are also clinically diagnosed with some musculoskeletal problem by Tsimane Health and Life History Project (THLHP)" and "some physical discomfort during activities". – what is "musculoskeletal problem" and how can we use this as an equivalent of the BMD measure used in this study?

We have clarified these statements and we thank reviewer #2 for their attention to detail.

Statements that are not true or well researched/cite:

"Ample bioarchaeological evidence of partially or fully healed fractures (e.g. (Lovejoy and Heiple, 1981; Pfeiffer, 2012)) and developmental abnormalities (Cowgill et al., 2015; Trinkaus, 2018) suggests longer-term survival with varying levels of disability.” In fact, numerous bioarcheological studies have shown that “osteopenic” fractures with low BMD are very rare in the past – traumatic/violent fractures are very common

We have since clarified and expanded upon these statements.

"Tsimane thoracic vertebral fractures appear to result from both trauma and senescence (Stieglitz et al., 2019) and are not just prevalent among the oldest adults.”

So are we looking at violence or accidental trauma fracture in males here – or related to another primary disease? This needs to be discussed/expanded. It is not senescence in young age males.

We have since clarified and expanded upon these statements.

Again – both sexes -age groups have different etiologies of low BMD and likely different outcomes/costs because of this.

Reviewer #3:

The submitted manuscript describing productivity loss linked to bone mass decline is well written, novel and pertinent in several domains. It is certainly a worthwhile contribution. The suggestions made here for adjustments and improvements reflect this reviewer's responses to the work. The clear statements at beginning and end regarding the importance of this kind of research for bioarchaeological and paleoanthropological studies is welcome.

Perhaps the broadest concern with the current version of the manuscript is the tendency to under-play the imbalance of impact of bone mass loss between men and women. Clearly, the paper's focus is on hunting and fishing, which is all good. However, from Abstract to conclusions the different pattern in women is underplayed. The type of loom(s) used by the weavers could be pertinent to the study and should be specified.

We have included a new figure (Figure 1) showing the actual looms used by women, and the finished woven products.

The brief section on the lack of significant effect on weaving should be moved up so that it precedes the section on long distance walking.

Done.

The relative inattention to women's productivity is a related concern.

We agree that this relative inattention to women’s direct production is a major study limitation.

There should be acknowledgment, as the authors discuss the limitations of the work, of the fact that their data are at least somewhat inconsistent with modern industrialized countries' clinical data, with regard to sex patterns in BMD and the disability load associated with compression fractures.

Here we are not exactly sure what Dr. Pfeiffer has in mind when she states that the Tsimane data are “at least somewhat inconsistent with modern industrialized countries' clinical data, with regard to sex patterns in BMD and the disability load associated with compression fractures.” Revised Figures 2-3 show linear age-related BMD declines for both sexes, as is typically observed in modern industrialized nations. Moreover, we now mention in the Discussion a previous study of Eekman et al., 2014, albeit using slightly different methods that are not directly comparable to those we use, that estimates costs of clinical spine fracture in terms of work absenteeism among osteoporotic adults aged 50+ years in the Netherlands. That study reports a cost of 25% of the mean annual per-capita income, which actually is within the range of productivity losses associated with fracture and lower BMD that we report, depending on whether one uses as a reference mean daily per-capita energy intake [cost of 15%] or intake from protein and fat [cost of 44%].

The research presented here links to a research area of which the authors appear to be unaware. In a 1996 paper (Anatomical Record 246:4:423-432), Agarwal and Grynpas asserted that osteoporotic fracture is absent in past populations. While it has only been cited 41 times, those citations include recent work.

Thank you, we now cite this paper.

I urge the authors to become better informed on just how novel – and helpful – their work is, in a bioarchaeological context. In my opinion, the best explanation for the "pattern" reported in that 1996 paper is that no one ever systematically studies thoracic compression fractures.

This interpretation seems reasonable, and we now mention it in the revised manuscript. We thank Dr. Pfeiffer for her insight.

Nevertheless, there is a literature arguing that osteoporosis is a modern disease (cf. Kralick and Zemel, 2020).

We thank Dr. Pfeiffer for mentioning this relatively new paper, which we now cite.

With regard to which vertebrae form the basis of the dataset: If T7 or T8 form the starting point for the assessment, the technician might be providing bone mass input from groups as different as T5-T7 to T8-T10. These are very different parts of the thoracic column, from the bone mass perspective. This is a concern that should be addressed, if possible. The morphological assessment of T6 to T12 helps to give the reader more confidence in the data.

We now address this.

The discussion of the contributions of older group members would benefit from stretching beyond physical activities to intellectual assets, along the lines of D.J. Levitin's argument that older adults contribute experience and judgement. The impressive breadth of the framework constructed by the authors could and should be even broader and more interdisciplinary.

We definitely agree, and have in fact published on other kinds of contributions of older Tsimane adults (some of these papers are cited). Thanks for bringing this up. We now expand our Discussion of the contributions of older group members, and cite Levitin’s work.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The following is a summary of the essential revisions, with the detailed reviewer comments appended below for expanded direction and explicit instruction:

The authors need to 1) reduce statements of implied causation in the discussion (and also in the general setup of the paper regarding the possible causal links between bone fragility and disability) and fully acknowledge all possibilities and 2) carefully establish potential mechanisms for any proposed causality, even when reduced explicitly to speculation. Currently the discussion is written as if the correlation is directional and infers causation, but with so many unstudied covariates (since lower BMD is a function of aging) this is inappropriate. The appropriate null hypothesis for interpretation should be carefully considered. I support some hypothesis generation and speculation in the discussion but this should be clearly indicated as such, and clearly separated from the conclusive interpretations. Even as such, the potential mechanistic links here would need to be much more carefully detailed and supported given the reliance on indirect inference only and the inconsistency with existing literature. Please see the reviewer comments for explicit suggestions and details.

Thank you for this detailed summary of the reviews—we found it very useful. In the revised Introduction and Discussion we now: i) omit any statements implying that our study definitively establishes causation; ii) emphasize that our results may be consistent with several possible explanations, not all of which are explicitly considered here; and iii) include a new conceptual model (see Figure 6 and accompanying text) that identifies diverse pathways linking bone fragility and diminished productivity. As suggested by Dr. Perry we are careful in the revision to recognize speculation in the discussion as such.

Reviewer #1:

This is a very nice study of associations between bone fragility, disability, and subsistence costs among a small-scale forager-horticulturalist society, the Tsimane of Bolivia. The study builds on a great recent paper in eLife by the same authors showing that the Tsimane (especially women) have lower average vertebral BMD values than might be expected given their high physical activity levels, as well as surprisingly high rates of vertebral fractures. In this study, they investigate the impact of low vertebral BMD and fracture incidence on disability (or, "adult physical impairment" as they call it) and ultimately the economic costs of such disability. The paper nicely frames the study within evolutionary theory and provides novel insights on the selective pressures that may have shaped human longevity and our species' propensity for age-related fragility. The sample analyzed is impressively large, particularly for a remote-living population like the Tsimane. The CT methods used to collect the bone data are the gold standard. The paper is clear and well written, and the interpretations of the results are all reasonable. Also, the figures are nice. Overall, I think this paper deserves consideration for publication in eLife, but I have a couple of concerns.

First, in my opinion, the possible causal links between bone fragility and disability are not considered or described in enough detail. The only mention of such a link is: "Vertebral deformities are reliably associated with chronic pain…(no citation)" Do the authors have data on back pain? If so, the link between bone fragility and disability would be a lot stronger if they could show that bone fragility is indeed associated with chronic pain among the Tsimane. If not, then it remains possible that bone fragility and disability are only indirectly related. For example, perhaps in many cases bone fragility is just one manifestation of a more general somatic decline, and disability is engendered primarily by other aspects of the general phenomenon and not bone fragility per se. If the authors do not have pain data, I suggest they at least make a stronger argument for the link between bone fragility and pain/disability by citing prior research demonstrating such links.

We agree, in retrospect, that we did not describe in sufficient detail potential mechanisms (e.g. back or other pain or discomfort) linking bone fragility and reduced economic productivity. We now make these connections more explicit, as suggested by all reviewers. We do this in the Introduction and Discussion (also see new Figure 6); the Discussion includes a new conceptual model linking bone fragility and reduced economic productivity during aging. We also note throughout the paper that there are several possible explanations for the findings that we report. We added relevant citations, including to the sentence mentioned by Dr. Wallace. Lastly, we added that prior anthropological studies of impairment and foregone productivity are similarly limited in that they overlook specific mechanisms. Unfortunately we lack systematic data on back pain specifically, but we are now considering ways to analyze existing medical data to obtain indirect insights into experiences of pain.

Second, on the same topic of causal links between bone fragility and disability, I believe the authors must provide a more detailed explanation for how they interpret the contribution of low BMD to disability. Independent of a fracture, there is no clear reason why one would expect low BMD to directly affect disability. Again, it is possible that low BMD and disability could be indirectly related if low BMD is only one of multiple outcomes of general somatic decline. But my understanding is that osteoporosis is generally an asymptomatic disease until fracture occurs. My impression is that the authors may not necessarily be arguing that low BMD, independent of fracture, affects disability, but rather that fracture plus low BMD leads to more disability than a fracture alone. Even so, why does this scenario make sense? Are fractures experienced by a bone with low BMD inherently more painful? Maybe the cooccurrence of fracture and low BMD leads to more disability because such fractures are generally more severe. Or, given that bone fractures can trigger bone loss, maybe the cooccurrence of fractures and low BMD represent cases where fractures took place longer ago and thus potentially have a greater effect on chronic pain and disability. An extreme way of resolving this uncertainty would be for the authors to disregard the BMD data entirely and focus just on the fracture data. The overall storyline would not change too much. However, I am not suggesting the authors do this. The BMD data are potentially important. But the links with disability need to be more carefully considered.

We appreciate Dr. Wallace’s request for clarification on mechanisms, especially pathways linking lower BMD and functional disability. His interpretation is broadly consistent with ours, and we agree that in the prior manuscript we did not explain this in enough detail. We now elaborate on links between lower BMD and disability in the text and in Figure 6, which we hope clarifies these mechanisms.

Reviewer #2:

This paper aims to estimate the magnitude of productivity losses associated with compromised bone strength among the Tsimane. The authors specifically test "whether a specific morphological characteristic which may indicate physical impairment, i.e. compromised bone strength, is indeed associated with diminished participation in routine subsistence tasks". They "identify economic correlates of compromised bone strength (i.e. thoracic vertebral body fracture and lower thoracic vertebral BMD) by examining involvement in four common tasks in foraging economies: hunting, tree chopping, tool manufacture (i.e. weaving), and walking long distances".

The data presented in this paper are important and have the potential to make a valuable contribution to the literature, and the authors are to be congratulated for conducing important, unique, and valuable fieldwork. The questions they have posed in this paper are interesting and worthy of consideration. That being said, I have some concerns with the analysis, what it assumes, what it excludes, and the conclusions.

The paper is set to test the "probability of completely ceasing to perform a given task increases with compromised bone strength", and estimate the "magnitude of productivity losses (i.e. lost kcals/day) from hunting cessation associated with compromised bone strength" The authors claim "We do not assume that compromised bone strength is itself a proxy for functional impairment or reduced productivity. Instead we posit that individuals with compromised bone strength who are no longer able to perform specific subsistence tasks facultatively adjust their time allocation based on various factors (e.g. phenotypic condition, household demand) to maximize productivity." This is really unclear and woolly. What, specifically then, are they predicting? If they don't think compromised bone strength influences productivity, why are they testing it, and why is the text replete with indications that they are interpreting this as a direct influence?

We agree that our original phrasing was unclear, and we have since clarified this. Specifically, we write:

“We do not assume ex ante that compromised bone strength per se is the only cause of disability or reduced productivity, or that these factors necessarily co-vary in some coordinated fashion—these are empirical questions partially addressed in this paper. We do expect that individuals with compromised bone strength who are no longer able to perform specific subsistence tasks will adjust their time allocation depending on various factors (e.g. own somatic condition, household demand).”

We also hope that the new Figure 6 will clarify this point; we have also omitted opaque language throughout the manuscript on this point.

The authors make their assumptions of the direction of this relationship clear throughout the paper, through the use of language such as "To estimate men's hunting productivity losses associated with fracture and lower BMD". The paper is replete with the inference of this relationship, and while the authors might carefully choose their words in places to try to avoid causation, the inference is there for the reader to make. Perhaps another way of putting this is the following: the authors show correlations between measures of later life activity/productivity and both increase fracture prevalence and lower BMD. There are three parsimonious explanations for this correlation:

1) The loss of bone quality is the cause of a decline in involvement in economic activity

2) The loss of bone quality is correlated with age, which independently corresponds with later life declines in economic activity, and the two are not directly related but collinear

3) A later life decline in activity is the cause of a decline in bone quality.

We thank reviewer 2 for their suggestion to describe in more detail different interpretations about statistical relationships between bone fragility and reduced economic productivity. We also thank reviewer 2 for outlining potential scenarios. We agree with reviewer 2 (and with reviewer 1) that unpacking the relationships by more explicitly considering mechanisms merits greater attention. As we mention above in our response to reviewer 1, we now describe in more detail potential mechanisms in the Introduction and Discussion, and alternative explanations that may be consistent with observed statistical relationships.

These are simplistic explanations, and there are likely other factors involved (see comments below), but for the purposes of simplicity, let's consider the above. The language of the paper is only oriented around the first explanation and the direction of causality is assumed through most of the text, but only really directly acknowledged when they briefly address the possibility of “reverse causality” (reductions in activity causing a reduction in bone strength). They attempt to test for “reverse causality” through indirect methods: comparing fracture status to self-reported reasons for task cessation, and the use of GLM to compare BMD of those who cease versus those who continue participation. I'm not convinced that these adequately address the challenges of the direction of the relationship. Self-reported causes may, in some cases, related to injury (particularly in severe cases), but are much more likely to relate to other factors entirely, and to be unrelated to BMD.

This is a valid point, and we repeatedly emphasize that these analyses are not direct or definitive tests of causality. We added a figure (Figure 6) to show multiple potential pathways linking somatic condition and economic productivity.

The GLM comparisons of BMD among groups defined by continued participation in activity may be influenced by study design, and I am not certain it is testing the reverse causation that the authors suggest.

We are not sure that we understand what is meant here by “comparisons of BMD…may be influenced by study design”. We have no evidence of specific sample biases, which we now clarify in the text (also see Appendix).

The most significant influence on later life bone mass is earlier life activity. Peak bone mass is commonly achieved between the ages of 20 and 30, and is the single most important predictor of later life bone mass. The more you build early on, the more bone you have to lose. Activity at age 20 may thus be stronger correlate of bone mass after the age of 40 than recent activity patterns. It is curious that the authors point out that "Physical strength peaks in the mid to late 20s, then declines continuously thereafter, reaching ~60% of maximum adult capacity by age 70 (Gurven et al., 2006)." This is essentially the same trajectory of bone strength acquisition and loss, and there is considerable evidence that bone strength corresponds with muscle induced loads. One could argue the most parsimonious explanation is that the most active and economically productive individuals during young adulthood have the highest peak bone mass, and are more likely to remain productive (or with higher than average productivity) later in life.

This is an excellent point, which we now acknowledge in the Discussion including in the new Figure 6.

Apart from the issues noted above, I remain sceptical about causation. There is, in theory, a direct causal relationship between fractures and decline or cessation of economic activity. This is supported by the relatively strong results, for example, the declines in weaving and ability to walk for a day associated with earlier fractures. What is the causal mechanism for the relationship with BMD? Going back to the three possible interpretations I mentioned earlier, I would have no issue with the statement of a hypothesis that vertebral fracture is the cause of a decline in involvement in economic activity.

Thank you for this suggestion, we now acknowledge this explicitly, as follows: “Human thoracic vertebrae track mechanics of both lower and upper limbs, and thoracic vertebral fracture can directly impede mobility.”

If I were to hypothesize what would explain the correspondence with BMD, I would go for a variant of the second hypothesis, that it is correlated with age, which independently correlates with other measures that are associated with later life declines. There are multiple collinear biological trends associated with senescence. While the authors provide no consideration of a biological mechanism through which BMD would be associated directly with productivity, the inference of causation is made without consideration of the particular relationship of declines in BMD associated with other factors of aging.

We agree, and the fact that both reviewers 1-2 converged on this suggestion convinces us that this was an important shortcoming of the prior submission. For this reason we included a new conceptual model (Figure 6) and text describing potential pathways linking bone fragility (including BMD) and diminished productivity. We emphasize repeatedly that our results may be consistent with several possible explanations, not all of which are explicitly considered here.

I am also sceptical of the consideration of both BMD and trauma as indicators of bone strength. BMD is highly correlated with bone mechanical function, this is uncontroversial, but considering trauma as a similar indicator may be problematic. Demographers and epidemiologists have often noted that trauma occurs in different societies with a relatively consistent frequency (at least when compared to infectious or metabolic disease, for example), because they result from accidental injury which tend to occur with some statistical regularity in larger populations. The data in Figure 2 suggest that there is no real difference the relationship between BMD and Age between groups of men with and without fracture, which could be interpreted as evidence that fracture prevalence among is predominantly determined by accidents rather than a loss of bone strength, regardless of the age or BMD of the individual. In contrast, there appear to be considerable differences among the women measured, suggesting that BMD and Fractures are more strongly correlated in women. The male data, at least, would seem to contradict the assertion that fracture prevalence is a proxy for bone strength.

It is fair to say that some fractures result from trauma, and some from fragility, and we are unable to determine the relative etiological contribution of each. We agree with reviewer 2’s general point that it would be problematic to conflate trauma with bone mechanical function. We also agree that Tsimane results suggest that men’s fracture may disproportionately result from trauma, including accident-induced trauma (as we acknowledge in the text; in Figure 6 and in a separate paper: https://elifesciences.org/articles/48607). These observations do not, in our view, contradict the idea that a fractured bone is more fragile (“strength-compromised”) than a non-fractured bone, all else equal. This idea is supported by studies showing that fractures provoke systemic bone losses, incomplete recovery of bone quality, and increased risk of future fractures and longer-term disability. Reviewer 2 is correct in stating that the association between BMD and fracture is stronger for Tsimane women than men. Men’s BMD is not significantly associated with fracture, defined as grade ≥1 on the Genant scale (i.e. ~20–25% reduction in vertebral height, and a 10–20% reduction in area). However, men’s BMD is inversely associated with borderline vertebral deformity, defined as grade 0.5 on the Genant scale (i.e. <20% reduction in height; for details see Appendix 1—table 17 here: https://elifesciences.org/articles/48607). For all of these reasons, we do not think that results for men invalidate the use of fracture status to ascertain degree of bone fragility.

https://doi.org/10.7554/eLife.62883.sa2

Article and author information

Author details

  1. Jonathan Stieglitz

    1. Université Toulouse 1 Capitole, Toulouse, France
    2. Institute for Advanced Study in Toulouse, Toulouse, France
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    jonathan.stieglitz@iast.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5985-9643
  2. Paul L Hooper

    Economic Science Institute, Chapman University, 1 University Drive, Orange, United States
    Contribution
    Data curation, Visualization
    Competing interests
    No competing interests declared
  3. Benjamin C Trumble

    1. Center for Evolution and Medicine, Life Sciences C, Arizona State University, Tempe, United States
    2. School of Human Evolution and Social Change, Arizona State University, Tempe, United States
    Contribution
    Supervision, Funding acquisition, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Hillard Kaplan

    Economic Science Institute, Chapman University, 1 University Drive, Orange, United States
    Contribution
    Supervision, Funding acquisition, Investigation, Project administration
    Competing interests
    No competing interests declared
  5. Michael D Gurven

    Department of Anthropology, University of California, Santa Barbara, Santa Barbara, United States
    Contribution
    Supervision, Funding acquisition, Investigation, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5661-527X

Funding

National Institutes of Health (R01AG024119)

  • Jonathan Stieglitz
  • Benjamin C Trumble
  • Hillard Kaplan
  • Michael D Gurven

National Science Foundation (1748282)

  • Jonathan Stieglitz

Arizona State University

  • Benjamin C Trumble

University of California, Santa Barbara

  • Michael D Gurven

Agence Nationale de la Recherche (ANR-17-EURE-0010)

  • Jonathan Stieglitz

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the Tsimane for participating and THLHP personnel for collecting and coding data. We also thank the HORUS Study Team for assistance with CT data collection. Amélie Beaudet, George Perry, Susan Pfeiffer, Ian Wallace, and two anonymous reviewers provided useful comments that improved the quality of the manuscript. Data on bone properties used in this study were generated under the supervision of Matthew Budoff. Funding was provided by the National Institutes of Health/National Institute on Aging (R01AG024119), National Science Foundation (1748282), the Center for Evolution and Medicine at Arizona State University, and the University of California-Santa Barbara Academic Senate. JS acknowledges IAST funding from the French National Research Agency (ANR) under the Investments for the Future (Investissements d’Avenir) program, grant ANR-17-EURE-0010. Funding sources had no role in research conduct, study design, or article preparation.

Ethics

Human subjects: Institutional IRB approval was granted by UNM (HRRC # 07-157) and UCSB (# 3-16-0766), as was informed consent at three levels: (1) Tsimane government that oversees research projects, (2) village leadership, and (3) study participants.

Senior and Reviewing Editor

  1. George H Perry, Pennsylvania State University, United States

Reviewer

  1. Ian Wallace

Publication history

  1. Received: September 10, 2020
  2. Accepted: November 30, 2020
  3. Accepted Manuscript published: December 1, 2020 (version 1)
  4. Version of Record published: December 16, 2020 (version 2)

Copyright

© 2020, Stieglitz et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 270
    Page views
  • 22
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Epidemiology and Global Health
    2. Genetics and Genomics
    Arslan A Zaidi, Iain Mathieson
    Research Article Updated

    Population stratification continues to bias the results of genome-wide association studies (GWAS). When these results are used to construct polygenic scores, even subtle biases can cumulatively lead to large errors. To study the effect of residual stratification, we simulated GWAS under realistic models of demographic history. We show that when population structure is recent, it cannot be corrected using principal components of common variants because they are uninformative about recent history. Consequently, polygenic scores are biased in that they recapitulate environmental structure. Principal components calculated from rare variants or identity-by-descent segments can correct this stratification for some types of environmental effects. While family-based studies are immune to stratification, the hybrid approach of ascertaining variants in GWAS but reestimating effect sizes in siblings reduces but does not eliminate stratification. We show that the effect of population stratification depends not only on allele frequencies and environmental structure but also on demographic history.

    1. Epidemiology and Global Health
    2. Genetics and Genomics
    Jennifer Blanc, Jeremy J Berg
    Insight

    A simulation study demonstrates a better method for separating genetic effects from environmental effects in genome-wide association studies, but there is still some way to go before this becomes a "solved" problem.