A paleodemographic assessment of mortality and fertility rates during the second demographic transition in rural central Indiana
Abstract
Objectives
Since its inception, skeletally based paleodemographic research has emphasized the utility of biocultural models for interpreting the dynamic relationship between the sociocultural and ecological forces accompanying demographic transitions and shaping populations' health and well-being. While the demographic transition associated with the Neolithic Revolution has been a common focus in bioarcheology, the present study analyzes human skeletal remains from a large 19th century cemetery in central Indiana to examine population dynamics during the second demographic transition, a period generally characterized by decreasing fertility rates and improvements in life expectancy. This study demonstrates the potential to methodologically identify regional variations in the timing and interactions between broad-scale socioeconomic changes and technological advancements that characterized the time period through observed changes in survivorship and fertility based on age-at-death distributions.
Materials and methods
This study uses three temporally distinct samples (AD 1827–1869; 1870–1889; 1890–1935) from the Bethel Cemetery (n = 503). Kaplan–Meier survival analyses with a log- rank tests are utilized to evaluate survivorship and mortality over time. Next, Cox proportional hazard analyses are employed to examine the interaction between sex and time as covariates. Finally, the D0–14/D ratio is applied to estimate fertility for each of the three temporally bounded cohorts.
Results
The Kaplan–Meier survival analyses and Cox proportional hazard modeling revealed statistically significant differences in survivorship between the three time periods. Age-specific mortality rates are reduced among adult female and male age classes in this rural community over the course of the 19th and early 20th centuries, resulting in the increasing life expectancies associated with the second demographic transition. While mortality in early adulthood was common during the first time period and decreases thereafter, sex was not identified as a meaningful covariate. The proportion of juveniles in the three temporal samples indicate that fertility rates were higher than national averages for the better part of the 19th century and subsequently declined around the turn of 20th century for this community.
Conclusions
The results indicate temporal differences between the three periods, demonstrating increased survivorship and decreased mortality and fertility over time. These findings corroborate two key features of the second demographic transition characterized by the move from high rates of both fertility and mortality to reduced rates and a general easing of demographic pressures. The observed trends likely reflect improvements in health, coinciding the industrial advance and economic development within and around Indianapolis. While the socioeconomic factors characterizing the Industrial Revolution drove demographic shifts that parallel an equally important epidemiological transition, potential regional differences are discussed to highlight variability in the timing of demographic transitions. The paleodemographic methods utilized in this study demonstrate improved accuracy and efficacy, which ultimately advances researchers' potential to disentangle population-specific socioeconomic factors that may contribute to asymmetrical experiences of health and mortality.
1 INTRODUCTION
Utilizing a variety of data sets, human biologists, bioarcheologists, and other researchers have routinely examined population dynamics derived from a variety of sources to explain shifts in human health and disease patterns, drawing upon demographic and epidemiological transition models as a framework for interpreting trends in past and contemporary societies (e.g., Gage, 1994; Johnson-Hanks, 2008; Lee, 2003; Lesthaeghe, 2014; McCaa, 2002; Schmidt & Sattenspiel, 2017; Wood, 1998). Among these researchers, Omran (1971) was the first to coin and utilize the epidemiological transition model to describe declining mortality from infectious diseases following the Industrial Revolution in the United States and Western Europe (Caldwell, 2001; Gage, 2005). Placing Omran's classic model into a larger evolutionary framework, researchers now consider there to be a series of demographic transitions of varying intensity and length, from the first that coincided with the Neolithic Revolution, to the present and ongoing transition of emerging and reemerging infectious diseases (Barrett et al., 1998; Harper & Armelagos, 2010). In this expanded model, Omran's original transition is recognized as the second major one, involving a reduction in infectious disease-related deaths and increasing life expectancy (McKeown, 2009). In light of these demographic and epidemiological changes over the last two centuries in the United States and elsewhere, the present study focuses on the demographic dimensions of this second transition by conducting bioarcheological analyses of survivorship, mortality, and fertility for a 19th century rural agrarian community from central Indiana (see Figure 1).
Source: Warner, A., Worley & Bracher & Bourquin, F. (1866). Map of Marion County, Indiana. Philadelphia: C.O. Titus, Publisher. Retrieved from the Library of Congress, https://www.loc.gov/item/2013593173/
The contemporary relevance of both transition models and the documented demographic changes revolve around their potential for interpreting and understanding the dynamic and interdependent relationships among demographic, social, economic, and environmental factors, especially as they relate to public policy, migration, population forecasting, and the evolution and spread of disease (e.g., Caldwell, ; DeWitte, 2016; Fleischer et al., 2014; Harper & Armelagos, 2010). Today, as modern populations face circumstances similar to the first demographic transition, including zoonotic infections, considerable effort has been dedicated to interpreting the historic and contemporary patterns of subsistence, social organization, and migration that may signal the emergence and reemergence of infectious disease (Zuckerman, 2014; Zuckerman et al., 2014). The intensification of agriculture and increased sedentism, associated with the Neolithic Revolution are recognized as coinciding with the first major demographic transition, which variably resulted in increased fertility and elevated mortality rates (Bocquet-Appel, 2002, 2011; Eshed et al., 2004; Wilson, 2014). While the intensity and speed of urbanization, social stratification, and globalization has increased since the Neolithic, the contemporary effects of these processes remain relatively similar (Moore, 2003). However, interpreting past demographic trends can be challenging given the absence of documentation in prehistory. As such, the analysis of human skeletal remains has been critical for examining trends in timing, elucidating the health trade-offs of urbanization, and identifying conditions that impacted the severity of past disease outbreaks and epidemics in various social and ecological contexts (e.g., DeWitte, 2010, 2014a; DeWitte & Wood, 2008; Nagaoka & Hirata, 2007; Walter & DeWitte, 2017; Zhang et al., 2016).
While archeological sequences and skeletal samples have been critical to modeling the major demographic changes accompanying the shift from foraging to farming (e.g., Bocquet-Appel & Naji, 2006), in contrast, analyses of the second demographic transition have primarily relied on historical records given their general ubiquity for the 19th and early 20th centuries. The second demographic transition has been broadly characterized as involving increasing life expectancy at birth, decreasing age-specific mortality rates, and progressive declines in fertility rates throughout the latter half of the 19th and most of the 20th century in many western countries (Caldwell, ; Riley, 2001). Omran (1971) contended that the improvements in disease load and health during this transition particularly benefited the young, including infants and children, as well as reproductive-age women, resulting in improved survivorship. The general properties of this transition include declining mortality rates that were not immediately accompanied by reductions in the average number of children per woman, resulting in population growth rates between 1 and 2% during the time period in question (Livi-Bacci, 2007). Population growth was especially pronounced when, during this transition, the birth and death rates were imbalanced, a feature that is also believed to have characterized the Neolithic Demographic Transition (Bocquet-Appel, 2002, 2009, 2011; Gage & DeWitte, 2009). In addition, this second transition was characterized by stabilization of annual death rates in Western countries by the mid-19th century. Given the interdependence of these demographic measures, life expectancy at birth (e0) has profoundly improved from under 45 years in the early to mid-19th century to over 80 years today, especially among women in several western countries (Oeppen & Vaupel, 2002). Drawing on federal census records from 1850 onward, Haines (1994) provides crucial estimates of life expectancy and total fertility for Euro-Americans in the United States. In brief, life expectancy at birth hovered around 40–45 years during the mid-to-late 19th century, while fertility rates were steadily declining from 5.4 in 1850 to 3.6 in 1900. Unlike the demographic transition in Europe where reductions in mortality preceded those for fertility, Haines (1994) contends that fertility rates, posited as beginning between seven and eight live births per woman, were already in decline in the United States by the onset of the 19th century. Meanwhile, more precipitous gains in life expectancy beyond 50 years of age are characterized as a largely 20th century phenomenon.
Collectively, these studies indicate that advances in industry and medicine instituted during the Industrial Revolution led to improvements in sanitation and other public health measures that dramatically reduced mortality from infectious disease (Cutter & Miller, 2005; Ellms, 1928; Higgs, 1979; Meeker, 1972; Okun, 1996). Cutter and Miller (2005), for instance, used census data to demonstrate how the availability of clean water and sanitary facilities, such as city sewer systems, reduced total mortality by half and infant mortality rates by three-quarters in the United States. However, several studies that have developed models for smaller communities have questioned the assumed accuracy of historical records (Funari, 1999; Johnson, 1999), indicating they are limited and biased toward larger cities and national-level data sets for which more sources exist. While many of these studies generally corroborate assertions regarding the impact of public health initiatives on declining mortality during the 19th and 20th century (Gage, 1994, 2005; Preston, 1976; Preston & Haines, 1991), they contend that far less is understood about rural, low-income, and marginal regions for which documentation is piecemeal or entirely absent, thereby suggesting a different experience during the second demographic transition (Farmer, 1996a; Farmer, 1996b; Higgs, 1973; Marinho et al., 2013). For example, Sattenspiel and Stoops's (2010) analysis of headstone data from the Columbia Cemetery in Columbia, Missouri demonstrated that the onset of major mortality shifts occurred in the late 1920s or early 30s, substantially later than similar changes in urban areas, including eastern cities and states. And yet, just knowing the numbers and ages of people who died may still not provide a clear picture of the health and disease experience of the once living population. Biased by a writer's discursive power, the contents of recorded histories often reflect privileged circumstances and potentially obscure or alter marginal experiences (Johnson, 1999), rendering interpretations of local-level changes incomplete (Beemer, 2011; Higgs, 1973; Koepke, 2014). In consolidating population-specific trends under an umbrella of aggregate data, the initial aim of these models to identify how and when shifts in mortality and fertility occurred is missed. The variety of socioecological conditions, especially those characterizing underrepresented populations, which includes the rural community in the present study, is critical for interpreting the forces that drive the speed and intensity of these transitions and provide a better understanding of the variation characterizing the second demographic transition.
Bioarcheological and paleodemographic analyses can provide empirical data on health, mortality, and fertility patterns that expands the geographic and temporal scope of research on demographic transitions. Given that individuals cannot obscure their biological response to processes affecting individual health and the lived experience, such as disease, early childhood stress, and malnutrition, analyses of skeletal data can serve as an invaluable source for building a more balanced picture of the demographic change driven by biosocial and ecological variables (DeWitte, 2014b; Krieger, 2005). Deriving meaningful data from skeletal analyses, however, can be challenging given the inherent limitations of a mortuary context to accurately represent the once-living population, including the length of cemetery use, demographic non-stationarity, and the selective nature of mortality that results in the frailest in a given age cohort entering a death assemblage at higher rates than others (DeWitte & Stojanowski, 2015; Milner et al., 2008; Milner & Boldsen, 2017; Waldron, 1994; Wood et al., 1992). Additionally, considering the proximity of the second transition to the present, the availability of skeletal samples as data sources to assess demographic change is often limited. As such, the ability to consider bioarcheological and textual data side-by-side render studies, such as the present one, invaluable.
Augmented by the wealth of regional- to national-level demographic data and vital statistics available for this time period, a primary objective of this study is to understand the potential of human skeletal remains and paleodemographic analyses to detect the aforementioned transition and inform our understanding of human biology and demography in the past. As part of the larger Bethel Cemetery Relocation Project (BCRP) undertaken in 2018 with input and feedback from descendants and community members (Peterson et al., 2019), the present study utilizes skeletal data from this 19th and early 20th century cemetery to analyze trends in survivorship, mortality, and fertility, while also assessing the divergence of our paleodemographic results from historical accounts and census records that capture the second demographic transition. Put simply, this study first examines how patterns of survivorship, mortality, and fertility changed, if at all, leading up to the height of Industrial Revolution. Second, we examine sex as a covariate with the goal of understanding to what degree a differential experience in mortality existed between females and males in this rural, agrarian community interring the deceased in their parish-associated cemetery. Third, we utilize a recently developed proportional measure of juveniles from skeletal samples (McFadden & Oxenham, 2018) to examine changes in total fertility rates (TFR) over time, comparing our results to national-level statistics on fertility for the time periods in question. Lastly, the present study aims to address several of the methodological issues and challenges raised by previous researchers regarding the analysis of cemetery samples (Bocquet-Appel & Masset, 1982; Buikstra & Konigsberg, 1985; DeWitte & Stojanowski, 2015; Wood et al., 1992), highlighting the benefits of newer age estimation and paleodemographic techniques, as well as a multidisciplinary approach.
2 BIOCULTURAL CONTEXT
2.1 Background
The skeletal sample analyzed in the present study is derived from the Bethel Cemetery, which was excavated in its entirety over 17 weeks during a 2018 mitigation project on the southwest side of Indianapolis, Indiana (Peterson et al., 2019). The excavation resulted in the assignment of 543 burial numbers. As depicted in the insets of Figure 1, 506 individuals were exhumed and underwent skeletal inventorying and analysis, while 24 were exhumed, but remained in concrete vaults. The remaining 13 burials were associated with graves lacking human skeletal remains, but including material culture, such as nails and small fragments of coffin wood. The Bethel Cemetery was in use between 1827 and 1935 and associated with the Bethel Methodist Episcopal Church, a congregation of Euro-American settlers in central Indiana. The cemetery is located some 11 km from the heart of Indianapolis, which was declared Indiana's capital in 1821, ultimately spurring economic growth in this region. The individuals and families interred at the Bethel Cemetery resided in Decatur Township with historical documentation for this area describing the community as agrarian and rural for most of the 19th century (Sulgrove, 1884). While parish and cemetery records are mostly lost or missing, it is reasonable to assume that members of the Bethel Methodist Episcopal Church community would have shared many of the experiences of other early central Indiana settlers. Tracking early settlers in rural communities through succeeding census years and land records has proven valuable for understanding these particular socioeconomic transformations. Meanwhile, economic development in rural, 19th century Indiana was characterized by the shift from subsistence to commercial agriculture, new modes of transportation and farm mechanization, the development of scientific agriculture leading to increased productivity, and rural outmigration (Carter, 1946). The speed and uniformity of this agricultural transition, however, varied considerably between various parts of the state (Atack & Batman, 1987).
During the early part of 19th century, central Indiana's inexpensive land and fertile soils drew settlers to farm with the non-Native population of the state going from several thousand in 1800 to nearly a million by 1850. Donnelly (1994) describes Decatur Township as being first settled in 1819 by Quakers from the Carolinas. US Census records for the 19th and early 20th centuries highlight a unique rural vs. urban relationship for Decatur Township and the Bethel community relative to Indianapolis and its other townships that experienced considerable growth. Between 1830 and 1870, Indianapolis' population would grow from an estimated 1900 to 48 244 individuals with major influxes to the state capital occurring during and shortly after the Civil War. As a percentage of Indiana's entire population, Indianapolis eclipsed 5% of the state's nearly two million residents in the last decade of the 19th century and 10% by 1920. In contrast, Decatur Township's population growth was exceedingly slow with 1008 individuals in 1850, 1647 in 1880, and 1550 individuals at the turn of the 20th century. The township's population would only come to exceed 3000 individuals in the 1930s, largely in response to employment and industry associated with the Indianapolis Municipal Airport, which opened in 1931. Population densities around the Bethel Methodist Episcopal Church and Cemetery, therefore, ranged from 31.2 people per square mile in 1850 to 51 people per square mile in 1880 and 51 again in 1920, paling in comparison to the nearly 7000 individuals per square mile living in the city's core during the 1920s.
Social and economic opportunities in this peripheral area to southwest of Indianapolis were accompanied by hardships, however, including a heavily wooded environment requiring clear-cutting for agriculture, a heavy disease burden, and physically demanding labor that challenged these early settlers to adapt (Daly, 2008; Ferguson, 1893; Johnson, 1951; Sulgrove, 1884). Daly (2008) notes that cholera outbreaks were a frequent occurrence in many smaller communities across Indiana between the 1830s and 1870s, attributing the elevated risk to shallow water wells and sewage contamination of drinking water. Economic growth was exceedingly slow for most of the 19th century, and while the construction of the National Road (i.e., US 40) in 1828 did spur community development, Messing (2018) characterizes traffic related to commerce and migration as fleeting and hampered by poor, underfunded road maintenance. By the mid-to-late 19th century, Indiana. Like many Midwestern states, would experience significant economic growth and demographic change, at the heart of which was the railroad industry (Abbott, 1978; Simons & Parker, 1997). The railway's arrival in 1847 connected the state's capital to the economically important Ohio River valley, connecting many Indiana farmers to regional and national markets (Thornbrough, 1965). As seen in Figure 1, this railway, as well as the Terre Haute and Vincennes Railroads, contributed to the growth of Bridgeport and Fremont, where many of the Bethel community members and local farmers would frequent small businesses and shops (Hetherington, 2000).
Advances in agricultural mechanization followed those involving transportation, with the Civil War spurring railroad construction, increasing demand for farm products, and creating a shortage of available farm laborers (Thornbrough, 1965). Indiana investments in farm machinery, such as iron plows, reapers, mowers, wheat drills, and corn planters nearly quadrupled, making possible the cultivation of three times as much land by 1900 (Nolan & Scarpino, 1988). Many changes in Indiana farming practices were linked to the explosion of agricultural information, organizations and initiatives undertaken by the federal government. Additionally, Purdue University's School of Agriculture and its agricultural experiment station began enrolling students in 1879 and became the most prolific source of information on scientific agriculture for Indiana farmers, performing research and disseminating results on topics ranging from seed germination to livestock fertility (Butz & Robertson, 1935).
Industrialization and urbanization peaked between 1880 and 1929 and new technologies in manufacturing and task-divided labor increased demand for a large labor force to maximize production (Phillips, 1968). Industrialization and urbanization reinforced one another, augmenting the growth of cities, including Indianapolis, as waves of immigrants concentrated in these urban centers for jobs and access to resources. Yet, even as cities became the new centers of American life, Indiana remained one of the nation's top agricultural producers (Baer, 2003; Butz, 1966; USDA Census of Agriculture, 2012). As farm communities overwhelmingly characterized rural Indiana throughout the 19th and 20th centuries, developments in agriculture remained an essential component of the state's progress. Connecting farm and factory maintained the bucolic aspects of rural communities with advances in agricultural technology allowing farmers to grow more using less labor (Baer, 2003; Peden, 2009). The isolation of agrarian communities, however, often meant differential access to resources (Phillips, 1968). Higgs (1973) has suggested that the mortality decline began in rural American around 1870, while in contrast a sustained decline in mortality did not take place in urban areas until after 1880 (Meeker, 1972). Urban centers characterized by crowding, social stressors, and poor air, food and water quality initially fostered high mortality compared to rural areas where land development and the mechanization of production lead to improved nutritional status and living standards (Haines, 1977; Haines, 1979b). But not all rural areas were similarly productive and over the next decade, urban mortality declined at a faster rate and health differences narrowed (Higgs, 1973). The late 19th century changes in social environments with regard to urbanization, industrialization, and improved living standards appear to have corresponded with the changing health status characterizing the second demographic transition (Higgs, 1979; Meeker, 1972; Preston & Haines, 1991). As mortality due to infectious disease transitioned to morbidity related chronic disease, rural areas were disproportionally affected, as the advantages of urban dwelling provided access to resources and opportunities that would signal improved health status (Elman & Myers, 1997). Despite the relative scarcity of records specific to the Bethel community, it is a reasonable supposition that Decatur Township's close proximity to downtown Indianapolis was advantageous for the Bethel community, offering access to transportation, markets, and new technologies.
2.2 Skeletal sample
Skeletal preservation predictably varied from poor to very good and excellent among the 506 individuals from the Bethel Cemetery analyzed by bioarcheologists from Indiana University-Purdue University Indianapolis (IUPUI) and the University of Indianapolis. The level of preservation was largely a function of several factors, including the location within the cemetery, a variable water table, type of coffin, the presence of a viewing pane, and the utilization of the coffins' shipping crates as a wooden vault. The collapse of the viewing panes, coffins, and wooden vault tended to result in very poor preservation of the scapulae, ribs, and vertebrae. With the records for the cemetery lost decades ago, erect and buried headstones (n = 151) were the only means of corroboration with the biological profiles described below. One-hundred and forty-four of the headstones were deemed to match the estimated biological profile, while seven headstones had been moved or fallen over and subsequently misplaced over another individual. In sum, over 70% of the individuals exhumed from the Bethel Cemetery originated from unmarked graves.
During the excavation, structure-from-motion photogrammetry was used to document the in situ state of skeletal preservation, coffin hardware, and location of associated material culture (Badillo et al., 2020; Peterson et al., 2019). Given the large number of unmarked graves, the analysis of coffin hardware and personal effects buried with individuals was required to develop terminus post quem (TPQ) and terminus ante quem (TAQ) estimates for each burial. The former consisted of patented decorative and non-decorative coffin hardware, including screws, hinges, handles, thumbscrews, escutcheons, ornamental tacks, caplifters, and viewing panes. Clothing-related items recovered and analyzed ranged from buttons, buckles, and fasteners through cuff and collar closures, safety pins, and shoes. Meanwhile, personal items recovered and analyzed included jewelry, beads, hair combs and pins, dentures, and coins, among other objects. The TPQs and TAQs utilized for this study are based on the burial container chronology set forth by Davidson during his work on the Freedman's Cemetery in Dallas, Texas (Davidson, 1999, 2004), the Holmes–Vardeman–Stephenson Cemetery in Lincoln County, Kentucky (Davidson, 2001), and the Becky Wright and Eddy Cemeteries in Crawford County, Arkansas (Mainfort & Davison, 2006). Davidson's chronology is particularly robust not only because it is constructed from archeological data gathered from four regionally and socioeconomically diverse cemeteries, but also because of his extensive analyses of patent records and coffin hardware trade catalogs. This archival research allowed Davidson to establish time frames for not only when certain material types were invented, but also when they became available for consumer consumption and when they appeared in the archeological record. These three dates can be very different, a fact often overlooked by many archeologists.
Because 138 burials were robustly identified through extant grave markers at Bethel, it was possible to assess the applicability of Davidson's chronology to this cemetery and make necessary adjustments prior to establishing TPQs and TAQs for unmarked graves. For the most part, the TPQs and TAQs provided by Davidson were consistent with the known dates of burial at Bethel, with one major exception. Mainfort and Davison (2006) note that while wire nails were first produced in the United States sometime between 1851 and 1875, they were not extensively used in casket manufacture until the mid-1890s. As a result, in his chronology the presence of wire nails in a burial should result in a TPQ of c.1895. However, several caskets from identified Bethel burials dating to the late 1880s and early 1890s were constructed exclusively with wire nails. The early introduction of this hardware type is likely due to Decatur Township's proximity to Indianapolis' industrial center. This included the Indianapolis Coffin Company, a commercial burial case manufacturing factory and warehouse first listed in the 1874 city directory (Swartz & Co., 1874) and detailed in the 1889 The Industries of the City of Indianapolis (Indianapolis Board of Trade, 1889). Therefore, the presence of wire nails in a burial at Bethel Cemetery resulted in an earlier TPQ of c.1890.
These temporal estimates enabled the division of the larger cemetery into three diachronic sub-samples of AD 1827–1869 (Period 1; n1 = 261), AD 1870–1889 (Period 2; n2 = 150), and AD 1890–1935 (Period 3; n3 = 92). Three of the 506 burials with skeletal remains could not be attributed to a specific time period of cemetery use. Following the initial double-blind classification of burials into the three time periods, the interment estimates were assessed and compared to the known ages of interment derived from the headstones. Accuracy was 95% (38/40) for Period 1, 96% (53/55) for Period 2, and 95% (41/43) for Period 3 (X2 = 242.428, p < .001). Meanwhile, the variability in sub-sample size is a function of both the coffin hardware's temporal specificity and usage of the cemetery by parishioners. Assuming complete recovery during the 2018 excavations and the accuracy of temporal assignments for the three time periods, the rate of interment varied over time from 6.4 and 8.5 individuals per year for Period 1 and Period 2 to 2.5 individuals per year for Period 3. The reduced rate of interment during the final period of the cemetery's usage likely reflects the gradual decline of the parish and use of other local cemeteries by descendants in the late 19th and early 20th centuries.
To facilitate data collection in a timely manner prior to reburial in 2019 for the larger BCRP, Osteoware 2.4 and the Sybase® Advantage Data Architect SQL database were employed to record and score the skeletal inventory and analysis of each individual (Peterson et al., 2019). Osteoware is largely based on the methodologies described in Standards for Data Collection from Human Skeletal Remains (Buikstra & Ubelaker, 1994) with notable additions related to, for example, dental morphology. Each individual was evaluated for preservation, taphonomy, pathology, oral health, metric and non-metric cranial and post-cranial variation, and morphological characteristics and metrics related to age-at-death and sex as described below.
3 MATERIALS AND METHODS
3.1 Age and sex estimation
Age-at-death estimation for juveniles from the Bethel Cemetery was contingent upon the relative stage of maturation of the individual (e.g., fetal vs. infant vs. child vs. adolescent), as well as skeletal and dental preservation. Age for fetal remains lacking a developing deciduous dentition was consistently estimated based on the dimensions of pars basilaris and pars lateralis of the occipital bone and pars petrosa of the temporal bone given their preservation at the Bethel Cemetery relative to other cranial and post-cranial skeletal elements (Nagaoka et al., 2012; Nagaoka & Kawakubo, 2015). Given the durable nature of the dentition, crown and root development for the deciduous and permanent teeth was the primary means of estimating age at death following Moorrees et al.' (1963a, 1963b) stages that have been recently revised by AlQahtani et al. (2010). When the dentition was poorly preserved or absent, age was estimated based on long bone diaphyseal length and the pattern of epiphyseal fusion following the reference data presented by Scheuer and Black (2000) and Cunningham et al. (2016). As a performance evaluation for the juvenile age estimates from the Bethel Cemetery prior to paleodemographic analysis, a Pearson's correlation was conducted on the known and estimated ages for the 36 juveniles with headstones, revealing a strong association between the two ages (r = .957, p < .001). The high degree of accuracy and precision associated with juvenile age estimation is depicted in Figure 2, where the green circles reflect the known ages for the 36 juveniles and the whiskers depict the upper and lower bounds of the skeletal age estimate.
Version 2.1.046 of transition analysis (Boldsen et al., 2002) was downloaded and served as the primary means of estimating age at death for adults from the Bethel Cemetery (available at: https://www.statsmachine.net/software/ADBOU2/). This technique, involving maximum likelihood estimation (MLE), orders individual components and sites within the pelvic girdle and cranium into senescent stages, requiring the analyst to estimate sex and ancestry, as well as a generalized mortality model, prior to the calculation of age. In lieu of a uniform prior described by Boldsen et al. (2002, p. 78), the informed prior based on 17th century Danish parish records was utilized with the tacit acknowledgement that the mortality profile of the Bethel Cemetery was likely to more closely resemble the pattern in “early modern northern Europe” as discussed by Milner and Boldsen (2012a, p. 100). Transition analysis has been extensively used in paleodemographic and epidemiological research over the past 15 years given the technique's ability to more accurately estimate age in individuals older than 50 or 60 years at the time of death (e.g., Boldsen, 2007; DeWitte, 2010; DeWitte, 2014a; Dewitte, 2014b; DeWitte & Wood, 2008; Walter & DeWitte, 2017; Wilson, 2014). Validation studies of transition analysis have clearly demonstrated its effectiveness (Fojas et al., 2018; Milner & Boldsen, 2012a; Milner & Boldsen, 2012b). Likewise, given the aim of the present study to detect improvements in life expectancy associated with the second demographic transition, transition analysis represents a significant improvement over older age estimation techniques involving the pubic symphyses and auricular surfaces of the ilia.
For the 108 individuals above 18 years of age at death with associated headstones, a second Pearson's correlation was performed to assess the degree of association between the known and the MLE-derived point estimate of age at death. A strong, but predictably weaker association (as compared with the juveniles) between known and point estimate of age was obtained (r = .869, p < .001), highlighting both the utility and limitations of transition analysis discussed by Milner and Boldsen (2012a). In opposition to the pattern observed for the 36 juveniles of known age at the time of death, Figure 2 also highlights the imprecision associated with adult age estimation with the whiskers around the circles representing the 95% confidence intervals for age-at-death obtained while utilizing transition analysis. Those individuals in Figure 2 with whiskers extending to 110 years of age represent one of two circumstances, both underscoring the challenges in conducting demographic research on variably preserved human skeletal remains. The first sub-set represents those individuals lacking the sites in the cranium and pelvic girdle essential for transition analysis, but reflecting other, non-traditional sites of senescence and advanced age in the axial and appendicular skeleton (e.g., advanced osteoporosis and/or osteoarthritis). The second sub-set of skeletal remains with broad “old age” interval estimates in Figure 2 represents those individuals that only had a handful of the sites routinely used in transition analysis. More specifically, due to preservation, these individuals' ages at death were estimated from cranial suture closure, which Milner and Boldsen (2012a, pp. 102–104) have demonstrated to be less precise than those age estimates derived from the pubic symphyses and sacroiliac joints.
Biological sex was the final dimension of data collection during the BCRP of importance to the current study. When preserved, macroscopic features of the pelvic girdle were heavily weighted during sex estimation following the methodologies outlined in Standards for Data Collection from Human Skeletal Remains (Buikstra & Ubelaker, 1994), as well as MorphoPASSE (Klales, 2020; Klales et al., 2012). Cranial and mandibular morphology were also assessed using the standard ordinal scoring system with the probability of male or female calculated using Walker's (2008) logistic regression formulae. Lastly, measurements of the appendicular skeleton, including humeral and femoral head size and humeral epicondylar breadth, were used to corroborate the sex estimates from the pelvic girdle and cranium following Milner and Boldsen (2012b); also Boldsen et al., 2015).To assess the accuracy of these metric and non-metric sex estimates prior to the demographic analyses, the first names and supplementary genealogical research for 123 headstones associated with 67 females and 56 males above 15 years of age at the time of death were compared to the skeletal estimates. Sex estimation accuracy for females was 98.5% (66/67), while the accuracy rate for males was 94.6% (53/56), demonstrating the efficacy of the laboratory methods (X2 = 107.444, p < .001). In total, 268 individuals above 15 years of age at the time of death were given sex estimates, including 57 females and 55 males for AD 1827–1869, 38 females and 40 males for AD 1870–1889, and 42 females and 36 males for AD 1890–1935.
3.2 Kaplan–Meier survival analysis and Cox's proportional hazard modeling
Kaplan–Meier survival analyses and the associated log-rank tests were conducted in SPSS 26.0 to examine survivorship across the three time periods of cemetery interment and within time periods to assess differences in mortality between adult females and females. This nonparametric approach to survival analysis has been utilized previously in bioarcheological research to, for example, examine early childhood stress and later mortality among adult females and males in medieval Denmark (Boldsen, 2007), pre- and post-Black Death survivorship in medieval London (DeWitte, 2014a; Dewitte, 2014b), linear enamel hypoplasias and survivorship among pre-Columbian Native peoples in midcontinental North America (Wilson, 2014), urban and rural populations in Roman Britain (Redfern et al., 2015), and high and low socioeconomic status children and adults from industrial-era London (DeWitte et al., 2016). While Kaplan–Meier survival analyses cannot simultaneously analyze multiple covariates, such as sex, skeletal lesion presence, and time period, the approach is widely used for censored data and smaller samples where parametric hazard modeling is statistically inappropriate. For the Kaplan–Meier survival analysis factoring time and examining the entirety of the human lifespan, 227 individuals for Period 1 (AD 1827–1869), 141 individuals for Period 2 (AD 1870–1889), and 105 individuals for Period 3 (AD 1890–1935) were analyzed from the Bethel Cemetery. Meanwhile, the second series of Kaplan–Meier survival analyses examining survivorship between females and males above 15 years of age at death within each time period included 57 females and 55 males for Period 1, 38 females and 40 males for Period 2, and 42 females and 36 males for Period 3.
To assess the interaction between time period and sex, Cox's proportional hazard modeling was utilized. In recent years, bioarcheologists have deployed this semiparametric approach of regression analysis to examine mortality and survivorship for linear enamel hypoplasias among Jomon people in Japan (Temple, 2014), vertebral neural canal size and linear enamel hypoplasias in late medieval and post-medieval London (Watts, 2015), long bone length, sex, and status in industrial London (Hughes-Morey, 2016), rural versus urban contexts among adult males and females in medieval London (Walter & DeWitte, 2017), and urbanization in medieval Poland (Betsinger, & &DeWitte, S., 2017). In the current study, age-at-death served as the predicted variable in a backwards conditional method where the three time periods and biological sex were identified as potential explanatory variables. In this form of stepwise regression, variables lacking explanatory power are iteratively removed, resulting in the most parsimonious model for, in this instance, survivorship. A total of 268 individuals that were 15 years of age or above at the time of death and reliably estimated to be adult females (nF = 137) and males (nM = 131) across the three time periods (n1 = 112, n2 = 78, n3 = 78) of cemetery interments were included in the Cox's proportional hazard modeling. In all of the aforementioned analyses, the maximum likelihood point estimates of age-at-death were utilized as opposed to the interval estimates, which potentially introduces a false level of confidence and precision in the paleodemographic analyses of adult mortality.
3.3 Fertility proxy
Previous studies have demonstrated a strong correlation between the proportion of juveniles in cemetery samples and the once-living population's birth and growth rates (Bocquet-Appel, 2002; Bocquet-Appel & Naji, 2006; Kohler et al., 2008). Given that age-at-death distributions in cemetery samples have been discussed as more sensitive to changes in fertility than to mortality fluctuations of the same scale, it is important to consider the role that crude birth and death rates had on population growth (Sattenspiel & Harpending, 1983). Previous research has demonstrated the relationship between fertility and mortality, such that as mortality increases the result will be an increase in fertility. However, estimating fertility using cemetery samples requires cautious interpretation, as the potential for bias may arise from multiple sources including inaccurate age estimation (Buikstra et al., 1986), infant underenumeration (Gordon & Buikstra, 1981), and inappropriately assuming demographic stationarity and population stability (Wood et al., 1992). Several fertility proxies have been developed to address these issues, thereby necessitating careful analysis of a skeletal sample's structure and history to select the appropriate proxy to generate the best fitting model. Though it is out of the scope of this study to discuss the variety of techniques and the specific analytical concerns each addresses (Bocquet-Appel, 2002; Bocquet-Appel & Masset, 1977; Buikstra et al., 1986; Robbins, 2011), several ratios were considered for their applicability to the Bethel Cemetery sample prior to selection.
McFadden and Oxenham (2018) used real, non-stationary populations to develop a linear regression that demonstrated the predictive capability of their ratio (D0–14/D) and known fertility rates for 52 countries with varying growth rates (r = .848). The D0–14/D ratio is a proportional measures of those between birth and 14 years of age over all individuals recovered from a cemetery sample, excluding fetal remains. While other methods for measuring fertility have focused on alleviating the problem of infant underenumeration in skeletal samples, the D0–14/D ratio differs in requiring a well-represented infant and young child categories. The D0–14/D ratio argues for improved correlation resulting from the inclusion of the 0–4 age category due to this cohort's high sensitivity to changes in fertility. As the Bethel Cemetery sample assumes good juvenile preservation and representation based on its large infant and child sub-samples shown in Table 1, the D0–14/D ratio was employed to estimate fertility. To test for significance, the present study computes 95% comparison intervals following the statistical procedure outlined by Buikstra et al. (1986) and derived from Sokal and Rohlf (1981). Comparison intervals are utilized as a method for multiple comparisons of proportions (i.e., D0–14/D), while controlling for the overall level of significance. This is opposed to confidence intervals, which constitute interval estimates for the location of true means.
| Age class | AD 1827–1869 | AD 1870–1889 | AD 1890–1935 |
|---|---|---|---|
| Fetal (16–36 weeks in utero) | 23 (8.8%) | 12 (8.0%) | 2 (2.2%) |
| Infant (0–2 years) | 55 (21.1%) | 45 (30%) | 12 (13%) |
| Young child (2–5 years) | 19 (7.3%) | 8 (5.3%) | 6 (6.5%) |
| Older child (5–12 years) | 33 (12.6%) | 7 (4.7%) | 2 (2.2%) |
| Adolescent (12–18 years) | 16 (6.1%) | 4 (2.7%) | 4 (4.3%) |
| Young adult (18–35 years) | 44 (16.9%) | 15 (10%) | 8 (8.7%) |
| Middle adult (35–60 years) | 24 (9.2%) | 16 (10.7%) | 7 (7.6%) |
| Older Adult (60+ years) | 35 (13.4%) | 43 (28.7%) | 50 (54.3%) |
| Adult (20+ years) | 4 (1.5%) | 0 | 1 (1.1%) |
| Unknown | 8 (3.1%) | 0 | 0 |
| Total | 261 | 150 | 92 |
4 RESULTS
As shown in Table 2, the Kaplan–Meier survival analysis and associated log-rank test examining those from birth to old age across the three periods of Bethel Cemetery's usage revealed significant differences in survivorship and the age-at-death distributions. During the first four decades of use, a considerable number of infants, children, adolescents, and young adults were interred in Bethel Cemetery with the mean and median age-at-death estimates reflecting this pattern. A similar pattern exists over the next two decades (i.e., Period 2), wherein infants and children represent some 40% of the mortality distribution. However, as depicted in the survival curves for the three time periods in Figure 3, distinct differences exist in the cumulative survivorship between Period 1 and the latter two during late adolescence and early adulthood, reflecting improved survivorship among these cohorts and older ones by the late 19th century. Meanwhile, the skewness of the distributions is reflected in the discrepancy between the mean and median age-at-death estimates for all three time periods at Bethel Cemetery shown in Table 2. During Periods 1 and 2, the lower median age-at-death estimates are strongly influenced by the fertility rates that are discussed below. In opposition, the higher median age-at-death estimate during Period 3 is likely a function of reduced age-specific mortality rates, decreasing fertility, and a reduced interment rate during the late 19th and early 20th centuries.
| Measure | Time period | Estimate | SE |
|---|---|---|---|
| Mean age at death | AD 1827–1869 | 23.846 | 1.693 |
| Median age at death | AD 1827–1869 | 15.000 | 2.371 |
| n | 227 | ||
| Mean age at death | AD 1870–1889 | 32.998 | 2.692 |
| Median age at death | AD 1870–1889 | 23.070 | 3.752 |
| n | 141 | ||
| Mean age at death | AD 1890–1935 | 52.632 | 3.059 |
| Median age at death | AD 1890–1935 | 68.770 | 3.577 |
| n | 105 | ||
| Log-rank test—Time effect | 62.706 | ||
| p | .000 | ||
In the second series of Kaplan–Meier survival analyses, sex-based survivorship among adults older than 15 years of age at the time of death was examined for each time period at the Bethel Cemetery. As seen in Table 3, no significant differences in survivorship were detected with Figure 4 depicting the similarity between female and male mortality patterns over time. While comparable within the time periods of cemetery usage, Figure 4 illustrates the steady reduction of age-specific mortality over time. Early adult mortality was high among both sexes during the cemetery's early use, but relatively negligible in Periods 2 and 3, suggesting the hazards associated with mortality in this community changed for both sexes during the demographic transition. This pattern is also reflected in mean and median age-at-death estimates in Table 3 with e15 for females and males moving from the lows 40s in the early-to-mid 19th century to the 50s by the late 19th century and 60s thereafter.
| Males | Females | ||||
|---|---|---|---|---|---|
| Measure | Time period | Estimate | SE | Estimate | SE |
| Mean age at death | AD 1827–1869 | 43.532 | 3.104 | 44.612 | 2.849 |
| Median age at death | AD 1827–1869 | 41.000 | 4.588 | 40.000 | 8.196 |
| n | 55 | 57 | |||
| Log-rank test—Sex effect | 0.002 | ||||
| p | .965 | ||||
| Mean age at death | AD 1870–1889 | 56.190 | 3.696 | 55.811 | 3.669 |
| Median age at death | AD 1870–1889 | 61.630 | 5.257 | 61.000 | 6.126 |
| n | 40 | 38 | |||
| Log-rank test—Sex effect | 0.019 | ||||
| p | .891 | ||||
| Mean age at death | AD 1890–1935 | 66.357 | 3.427 | 65.900 | 3.080 |
| Median age at death | AD 1890–1935 | 72.500 | 1.212 | 72.500 | 1.231 |
| n | 36 | 42 | |||
| Log-rank test—Sex effect | 0.069 | ||||
| p | .793 | ||||
The Cox's proportional hazard modeling of adult females and males above 15 years of age at the time of death across the three time periods of cemetery usage was statistically significant (−2 Log Likelihood = 2431.349, X2 = 46.771, df = 3, p < .001). Paralleling the results of the survival analyses, the backward stepwise regression removed sex from the model between Steps 1 and 2 of the analysis as shown in Table 4. The second model that included time as the only explanatory variable was statistically as parsimonious as the former one (−2 Log Likelihood = 2431.448, X2 = 46.682, df = 2, p < .001), thereby indicating that sex does not significantly contribute to the variation in mortality observed within the Bethel Cemetery. Figure 5 depicts the survivorship between the three time periods for adults. From late adolescence onward, the age-specific risk of mortality is attenuated across the three time periods. While the velocities (aka slope) of the survival curves resemble each other, fewer individuals, irrespective of sex, are dying in late adolescence, as well as young and middle adulthood, during the second and third time periods, resulting in increased life expectancy and thereby corroborating one aspect of the second demographic transition documented via census data and other sources.
| Step 1 | Variable | n | Wald | Exp(B) | Sig. | 95% CI |
|---|---|---|---|---|---|---|
| Sex | 268 | 0.099 | 1.040 | 0.753 | 0.816–1.324 | |
| Time period | 268 | 44.188 | 0.000 | |||
| Period 1 | 112 | 42.736 | 2.806 | 0.000 | 2.059–3.823 | |
| Period 2 | 78 | 6.608 | 1.518 | 0.010 | 1.104–2.088 |
| Step 2 | Variable | n | Wald | Exp(B) | Sig. | 95% CI |
|---|---|---|---|---|---|---|
| Time period | 268 | 44.129 | 0.000 | |||
| Period 1 | 112 | 42.663 | 2.801 | 0.000 | 2.057–3.816 | |
| Period 2 | 78 | 6.540 | 1.514 | 0.011 | 1.102–2.081 |
The D0–14/D ratios for the three periods of cemetery use and the calculated TFR are presented in Table 5 alongside national averages for the same time intervals derived from Haines (2008). While overlap is reflected in the 95% comparison intervals, the D0–14/D ratios decline with each consecutive time period of use for the Bethel Cemetery. The decline between the first and second time periods is slight (9% decrease) with a more significant decline by Period 3 (31% decrease). These results are consistent with previous findings, indicating a drop in birthrates and probable improvements in survivorship. When compared to the estimates from Haines (2008), the TFR derived from the D0–14/D ratios across all three time periods at the Bethel Cemetery remained higher with their decline being slower than national averages, ultimately suggesting variance in the second demographic transition.
| Period | D0–14/D | 95% Comparison intervals | Calculated TFR | National average TFR (Haines, 2008) |
|---|---|---|---|---|
| 1827–1869 | 0.58 | 0.48–0.68 | 6.71 | 5.08 |
| 1870–1889 | 0.49 | 0.38–0.62 | 6.07 | 3.71 |
| 1890–1935 | 0.25 | 0.12–0.38 | 4.16 | 3.01 |
5 DISCUSSION
Despite the utility and accessibility of historical records for modeling the second demographic transition, the results of these investigations have been relatively broad with regards to demography, health, and well-being, while inadequately addressing fine-scale differences within and between populations. The present paleodemographic analysis of a rural, 19th century community from central Indiana provides better estimates of mortality, fertility, sex-specific survivorship, and life expectancy, ultimately informing our understanding of this second major transition in human population dynamics. Utilizing a multidisciplinary approach that integrates bioarcheological research with parallel archeological work and information from historical accounts, the present study illuminates regional and temporal differences that would otherwise remain obscured. The paleodemographic models utilized in the present study to examine survivorship and fertility demonstrate a strong correlation between the study's skeletal sample and known data derived from the once-living population. These results have proven useful in capturing demographic trends, as well as identifying divergent patterns of survivorship, mortality, and fertility. However, as a part of our more recent past, the accessibility of archeological samples for analysis dating to this second demographic transition has been limited. As such, the few bioarcheological studies that do exist (DeWitte, 2014b; Nystrom, 2014; Perry, 2014; Western & Bekvalac, 2017), including the present one, are critical for enhancing our understanding of the social and ecological conditions that characterized the second demographic transition and variably shaped populations' experiences related to health and disease.
The present study compared the age-at-death distributions between three temporally distinct periods of the Bethel Cemetery, demonstrating patterns of survivorship and mortality consistent with historically documented trends. As anticipated with second demographic transition models, improvements in survival, as indicated by each consecutive age-at-death distribution, demonstrated significant increases in the proportion of older adults (i.e., Table 1), which was corroborated by each Kaplan–Meier survivorship analysis presented in Tables 2 and 3. While appreciating the steadily declining mortality rates, specifically adult mortality, it is important to consider factors, such as migration, which may off-set these distributions. As existing historical documents indicate (Sulgrove, 1884), the social and economic circumstances characterizing the community during the earlier parts of the cemetery's use would have likely drawn younger adults and their children to migrate to and settle the township. As such, we would expect an increased proportion young adults relative to older adults in the earlier part of the cemetery's history. The increasing proportion of old adults, when accompanied by decreasing proportions of young and middle adults within and between consecutive time periods, suggests that the larger conclusions made here about improvements in adult survivorship are parsimonious with models for the second demographic transition where the force of mortality eased across the human lifespan.
While an important contribution of the present study was the ability to generate a paleodemographic model that accurately captured the demographic processes characteristic of the second demographic transition, of equal importance are the subtle divergences from prior census-derived analyses. Previous second transition models suggest that, as survivorship increased and mortality from infectious disease declined, there were simultaneous declines in fertility (Haines, 1979a, 1980, 1998; Hirschman, 1994). Though the present study's hazard analysis demonstrates steadily declining mortality and improved survivorship, the Bethel Cemetery sample's D0–14/D ratios demonstrate minor changes in fertility and higher rates relative to recorded state-wide and national averages (US Bureau of the Census, 1975; Haines & Steckel, 2000; Modell, 1971; Vinovskis, 1976; Yasuba, 1962). At present, the relationship between changes in mortality versus changes in fertility for explaining the demographic transition are not well understood (Sattenspiel & Stoops, 2010). Several demographic studies have examined the role of socioeconomic and ecological factors that may benefit limiting fertility (Gillespie et al., 2007; Lawson & Borgerhoff Mulder, 2016; Lawson & Mace, 2010), but the empirical data to explore these trends among small, rural communities is difficult to locate. Researchers have suggested that there were differences in the timing and degree of change across and within urban and rural areas (Condran & Crimmins, 1980; Hacker, 2003; Haines, 1977, 1979b; Higgs, 1973), population subgroups (Condran, 1984), age (Meeker, 1972), and ethnicity (Condran, 1984; Haines, 1977), often finding that fertility in the 19th century tended to be higher among farming communities as compared with others (Atack & Batman, 1987; Easterlin, 1971; Steckel, 1994).The results of the current study support these findings, further evidencing the need to reexamine interpretations of the second demographic transition for their uniformity.
Much like the inherent bias related to historical accounts and records, addressing bioarchological studies' samples, methods, and the extent to which these data can be considered to reflect once-living populations has been a persistent concern in the field (Goodman & Martin, 2002; Walker et al., 2009; Wood et al., 1992). For example, are we able determine whether improvements in health and declining mortality were directly associated with changes brought on by the Industrial Revolution, or might they be artifacts of migration or changes in the cemetery's use? As discussed, the variation in this study's age-at-death distributions would be considered to reflect manipulation from migration alone in, for instance, the unlikely scenario that migrants had primarily consisted of older adults. Similarly, as fertility in this study appears to have neither influenced nor been affected by changes in survivorship and mortality, the inconsistency with rates indicated by historical records may illuminate a different experience for rural families. The challenges that accompany fertility measurements based on archeological samples, however, have driven researchers to develop and apply new estimation methods that take into account sample-specific potentials for biases. Buikstra et al. (1986), for example, demonstrated a proxy (D30+/D5+) for estimating fertility among eight Woodland and Mississippian skeletal samples from west-central Illinois that contradicted previous models for the Neolithic Demographic Transition. The results suggest that demonstrated increases in fertility were the product of a dietary shift to high carbohydrate resources rather than the result of increased mortality (Buikstra et al., 1986). Recognizing the region-specific complexity of intervening variables, the aforementioned study, congruent with the present findings, is that while previous transition models have captured a broad picture of fertility patterns, interpreting regional variability in fertility is critical to generate and enhance models that explain said change. Similarly, selecting and statistically supporting the most appropriate fertility proxy was a priority of the present study, further supporting the suggestion that the divergence in results likely reflects urban–rural difference, rather than a source of methodological bias or error.
An additional consideration of this study, like that of other paleodemographic research, is the extent to which changes in mortality or fertility are an accurate indicator of health and can reflect the variety of forces driving these shifts. There are several factors related to population health status that researchers have explored as shaping demographic transitions, such as changing social conditions and health practices. Numerous studies, for instance, have suggested that changes in health that came subsequent to the implementation of specific public health and sanitation measures during the Industrial Revolution (Gage, 1994; Higgs, 1979; Preston, 1976). As has become increasingly evident, a single agent for causation is unlikely given the variability both within and between populations (Farmer, 1996a; Farmer, 1996b; Krieger et al., 2005; Sattenspiel, 2011). An aim of the present research was to contribute to the development of a framework for methodologically identifying and examining covariates that may contribute to this differential experience. Similar to the present study's results exhibiting variable shifts in fertility patterns between rural and urban communities, paleodemographic models of age-at death distributions can address for whom, and in what geographic contexts changes in health and mortality arose differently. For example, the second demographic transition has been generally characterized by increased sex differentials in mortality, interpreted to likely reflect changing socioeconomic and environmental conditions as both men and women's health saw early improvement, subsequently marked by asymmetrically timed gains (Mooney, 2002; Vaupel et al., 1979). Prior to the second transition, demographic models have interpreted health and census data to demonstrate higher rates of mortality initially for women, with subsequent improvements in social status and reproductive technologies generating more forceful decline in mortality rates (Retherford, 1975). Inversely, these models suggest males to be increasingly disadvantaged with respect to mortality and longevity owing to a higher susceptibility to the chronic conditions that characterize the latter part of the second transition (Preston, 1976). The results of the present study, however, did not find any significant differences in adult mortality based on sex within or between time periods. These findings, which contradict expected sex differentials, advance our understanding of the diversity surrounding demographic transitions, significantly contributing to the field by demonstrating methods through which studies can tease apart numerous covariates.
The extent to which differentials in health status change and reflect biological, environmental, and social structures over time may help elucidate the conditions under which demographic transitions occur. This study contributes to the theoretical framework for examining the second transition, considering jointly primary source information and skeletal data to interpret geographic and demographic differences that may have shaped population dynamics. Despite the aforementioned limits and inherent biases associated with mortuary contexts, archeological recovery, and skeletal analyses, this study's application and statistical evaluation of newer methods offers promising implications for improved accuracy and more visible efficacy of paleodemographic studies for modeling the second transition.
6 CONCLUSION
Previous research using large data sets from major US cities has illustrated the significant impact of the second demographic transition in the United States, but the resulting homogeneous picture obscures complexity and variance made evident by the regionally diverse experiences of how and when these changes occurred. The present study analyzed skeletal data to interpret trends is mortality, fertility, and longevity among a primarily agrarian community in rural Indiana, but contemporary to the Industrial Revolution in neighboring Indianapolis. The study's empirical data, measured against known historical accounts, is generally analogous, but underscores some more nuanced, rural–urban differences. Though the results do not generate assertions regarding the specific social and ecological forces shaping differential experience, they contribute to a foundation for future, complementary research. In growing the field's body of evidence, this study suggests that a multidisciplinary approach may provide opportunities for inferring causation and measuring the impact of forces governing the variable trajectories of demographic transitions. Moreover, this study carries contemporary relevance as modern populations are on the brink of a third transition characterized by decreasing fertility, increasing life expectancy, continued social stratification, increased globalization, and indiscriminate use of antibiotics. These conditions have paved the way for increased transmission and pathogen virulence, giving rise to what these models have long predicted to be largely devastating global pandemics. This study suggests that solidifying a methodological approach for modeling these transitions and disentangling social determinates of population dynamics and health can allow future research to more comprehensively address the consequences of inequality, irresponsibility, and the return of infectious disease crises, potentially offering suggestions for health crisis mitigation.
ACKNOWLEDGMENTS
The authors would like to thank the community of Decatur Township in Marion County, Indiana and the descendants of those interred at the Bethel Cemetery who actively provided input, feedback, and support during the field and laboratory components of the project. In addition, this manuscript has greatly benefited from input on 19th century Indianapolis and central Indiana provided by Paul R. Mullins, as well as Sharon DeWitte, who provided input and feedback on the paleodemographic analyses.
AUTHOR CONTRIBUTIONS
Gretchen Zoeller: Conceptualization; data curation; formal analysis; investigation; methodology; resources; writing-original draft; writing-review and editing. Brooke Drew: Formal analysis; resources. Christopher Schmidt: Formal analysis; resources. Ryan Peterson: Project administration; supervision. Jeremy Wilson: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; resources; software; supervision; validation; visualization; writing-original draft; writing-review and editing.
Open Research
DATA AVAILABILITY STATEMENT
The data associated with this research and the larger Bethel Cemetery Relocation Project are available from the corresponding author(s) upon request in multiple file formats (i.e., .xlsx, .csv, .sav). The data are also permanently curated and available from the Indiana Department of Natural Resources' Division of Historic Preservation & Archaeology. Please reference the Bethel Cemetery (12Ma1025; CR49-12) when corresponding with the state agency.
