TITLE: NDVI Predicts Birth Seasonality in Historical Baja California Sur, Mexico
AUTHORS: Shane J Macfarlan1,2,3(shane.macfarlan@anthro.utah.edu), Ryan Schacht4(schachtr18@ecu.edu), Izabella Bourland1,5(izabella.bourland@utah.edu), Savannah Kapp1(savannah.kapp@utah.edu), Trevor Glad6(u0960974@utah.edu), Lauren Lewis1(lewis.lauren.abbey@utah.edu), Nathan Darmiento2,7(u0892197@utah.edu), Tanner Clegg8(tannernoahclegg@gmail.com), Cole Thorpe2,10(colethorpe96@gmail.com), Taylor Peppelar1(taypep@gmail.com), Melissa Santiago4(santiagom15@ecu.edu), Celeste Henrickson1,11(celestehenrickson@gmail.com), Kenneth Blake Vernon1(kenneth.b.vernon@utah.edu)
1 Department of Anthropology, University of Utah, Salt Lake City, UT 84112
2 Center for Latin American Studies, University of Utah, Salt Lake City, UT 84112
3 Global Change and Sustainability Center, University of Utah, Salt Lake City, UT 84112
4 Department of Anthropology, East Carolina University, Greenville, NC 27858
5 Environmental and Sustainability Studies, University of Utah, Salt Lake City, UT 84112
6 Department of Music, University of Utah, Salt Lake City, UT 84112
7 Department of History, University of Utah, Salt Lake City, UT 84112
8 College of Humanities, University of Utah, Salt Lake City, UT 84112
10 Department of Linguistics, University of Utah, Salt Lake City, UT 84112
11 General Education, Nightingale College, Salt Lake City, UT 84111
Short Title: NDVI Predicts Birth Seasonality
Keywords: Birth Seasonality; NDVI; Baja California Sur; Mexico; Behavioral Ecology; Adaptive Strategies; Historical Demography; Mexican History
Article Type: Letters
Abstract Word Count: 150
Text Word Count: 3316
Number of References: 81
Number of Figures: 1
Number of Tables: 3
Corresponding Author:
Shane Macfarlan, 260 Central Campus Drive, Department of Anthropology, University of Utah, Salt Lake City, UT 84112
Email: shane.macfarlan@anthro.utah.edu Office Phone: (801) 587-3432
SJM conceived of the study and performed all the analyses. SJM, IB, SK, TG, LL, ND, TC, CT, TP, MS, CH, KBV collected the data. SJM and RS wrote the first draft of the manuscript.
ABSTRACT
Birth seasonality is a phenomenon whereby populations can be characterized by a single month or season in which births peak. While non-human animal research suggests seasonal birth-pulses are related to variation in climate and local energy availability, social scientists debate the mechanisms responsible for it in humans. Here we investigate the role of precipitation, temperature, and energy availability on seasonal birth pulses using a historical dataset from the Baja California peninsula - a hot, arid desert that experiences seasonal fluctuations in temperature, precipitation, and energy associated with the North American Monsoonal. Analyses suggest that local energy availability (as measured through NDVI) predicts seasonal birth pulses, while temperature and precipitation do not; however, both are indirectly related to it through their direct effects on NDVI. Our analyses demonstrate the importance of local energy availability on human reproductive timing and suggest that human birth seasonality is a form of traditional ecological knowledge.
INTRODUCTION
Birth seasonality [one aspect of breeding phenology (Forchhammer et al. 1998; Sinclair et al. 2001)] is a common trait among many viviparous species whereby parturition events peak in a single season or month (Ellison et al. 2005; Wittemyer et al. 2007; Martinez-Bakker et al. 2014). Birth seasonality is particularly common among iteroparous organisms (those that reproduce more than once during their lifetime) residing in seasonal habitats (Pettorelli 2013). Comparative research in animal ethnology and evolutionary ecology indicate that breeding phenology is strongly associated with both climate and temporal variation in food availability across animal taxa (Drent & Daan 1980; Kennish 1997; Forchhammer et al. 1998; Sinclair et al. 2001; Rubenstein & Wikelski 2003; Wittemyer et al. 2007; Pettorelli 2013; Drake & Martin 2018). There are clear payoffs for organisms to synchronize reproduction with local energy availability, especially among mammals with obligate pre- and post-natal parental investment (e.g. gestation, lactation, and resource provisioning) – offspring born during peaks in energy availability are heavier and have higher survival rates than those born before or after those peaks (Pettorelli 2013; Plard et al. 2014).
Seasonal birth pulses are not unique to non-human animals and were first identified in humans nearly 200 years ago through early epidemiological research (Villerme 1831). However, despite being recognized as a near universal feature of human populations today (Cowgill 1965; Bronson 1995; Ellison et al. 2005; Dahlberg & Andersson 2019), social scientists continue debate the mechanisms responsible for this patterning (Condon & Scaglion 1982; Bobak and Gjonca 2001; Ellis et al. 2005; Herteliu et al. 2015; Dahlberg & Andersson 2018, 2019). While some have favored non-adaptive socio-cultural explanations for single case studies (see review in Ellis et al. 2005), recent large-scale, cross-national datasets suggest that human birth seasonality may be related to seasonal variation in local energy availability (Martinez-Bakker et al. 2014). For example, across the northern hemisphere, human birth seasonality has a latitudinal gradient, with populations residing between the equator and 30 degrees North Latitude typically peaking in births between the fall and early winter (i.e., September through December) and those between 30 degrees North Latitude and the Arctic peaking between the spring and summer (i.e., April through August) (Martinez-Bakker 2014). The patterning of latitudinal distribution of seasonal birth pulses generally tracks ecosystem-level energy availability, which has similar seasonal variation based on a latitudinal gradient (Becker & Boyd 1957; Lindsey 2009). Mothers who give birth during peaks in seasonal local energy availability have children with higher birth weights, larger head circumferences, and greater gestational ages, all of which are indicators of child health and survivorship (Dadvand et al. 2012; Case et al. 2016; Bakhtsiyarava et al. 2018).
Local energy availability is impacted by a number of bio-physical factors, including temperature and precipitation (Wu et al. 2001; Pettorelli 2013; Yan et al. 2015; Gilbert 2019). Not surprisingly, a large body of research has linked seasonal variation in precipitation and temperature to birth seasonality in humans (Condon & Scaglion 1982; Lam & Miron 1998; Ellis et al. 2005; Lawlor et al. 2005; Torche & Corvalan 2010; Osei et al. 2016). However, these aspects of climate serve as proxy variables for local energy availability (Pettorelli 2013) and no one has examined the direct impact of local energy availability on human birth seasonality.
Accordingly, here we investigate the role of all three (precipitation, temperature, and energy availability) on monthly birth pulses in a 19th century dataset from Baja California Sur, Mexico. Performing this research on a historical, rather than modern, dataset has several advantages. First, as global populations have become increasingly industrialized, societies have generally become reliant on infrastructure that smooths out localized environmental variability and consequent resource scarcity (e.g. drought). Hence, this reduces variability in local energy availability thereby lowering the incidence of monthly birth pulses (Dahlberg & Andersson 2018, 2019). Conversely, subsistence-level populations are more likely to be impacted by seasonal variability in local bio-physical processes (Moran 2018). Second, as human populations continue to undergo demographic transition (i.e., lower fertility), the ability to detect birth seasonality is diminished (Dahlberg & Andersson 2018). Historical datasets cover time periods that typically promoted early female reproduction and higher fertility (Kaplan 2006; Kaplan et al. 2000, 2001; Galor 2012; Macfarlan et al. 2019). Consequently, research on these data allow for more reliable tests of birth seasonality than among modern populations with lower fertility due to widespread contraception, public education, and the advent of scientific medicine.
MATERIALS AND METHODS
Study Site: Baja California Sur (hereafter, BCS) is one of two Mexican states that comprises the Baja California Peninsula (Krutch 1986) (Figure 1). Climatologically much of BCS is characterized as a hot, arid desert (Köppen-Geiger BWh) (Rebman & Roberts 2012); however, portions of the southern tip of the peninsula, known as the “Cape Region”, is characterized as hot and semi-arid (Köppen-Geiger BSh) (Rebman & Roberts 2012; Flores-Cárdenas et al. 2018; Maldonado-Enriquez et al. 2020). Both regions are subject to the North American monsoon (Barron et al. 2012), a climatological phenomenon whereby an onshore wind shift occurs between July and September that is associated with an increase in temperature and precipitation (Hasting & Turner 1965; Diaz et al. 2001; Rebman & Roberts 2012; Barron et al. 2012). This climatological system has been estimated present for the last 7500 years (Barron et al. 2012). Directly following the monsoonal rain season, biological energy availability increases, resulting in a September through December growing season (Maldonado-Enriquez et al. 2020). Minor rain events continue through January; however, between March and June, life becomes increasingly difficult as precipitation virtually vanishes and temperatures rise (Salinas Zavala et al. 1990. Numerous historical accounts attest to the difficulty of surviving this arid ecosystem (Clavigero 1937; Baegert 1952).
Figure 1 approximately here
Population History: The Baja California peninsula has been inhabited by indigenous peoples for at least the last 12,000 years (Fujita; Des Lauries et al. 2015; Macfarlan & Henrickson 2010); however, peoples of Euro-American descent began to permanently occupy the region after the Jesuits founded mission Loreto in 1697 AD (Martinez 1960; Crosby 1994). The Jesuits brought with them soldiers and their families who performed a number of economic roles to assist with the mission building process such as smithing, farming, and herding. The descendants of these initial colonists make up a portion of the peninsular population today (Martinez 1960, 1965; Crosby 1981, 1994). After 70 years of tight administrative control, the Jesuits were expelled from the Spanish Kingdom (1768 AD) and the Baja California peninsula was opened to those seeking to settle the land, resulting in a second population influx (Martinez 1960). Two additional waves of in-migration later occurred; one following Mexican Independence (1821 AD) and another that occurred throughout the Presidency of Porfirio Diaz (the Porfiriato Period: 1875-1910 AD) (Trejo Barajas 1994, 2005). Across all four migratory events, the majority of people came from Sonora and Sinaloa; however, they also included people from Asia, Europe, and North and South America (Martinez 1965; Macfarlan et al. 2019).
Throughout the 19th century, the Baja Californian economy was largely insular, with economic ventures focusing on cattle and goat ranching, crop production (largely based on Jesuit introduced species), and mining (Trejo-Barajas & Gonzalez 2002). Indeed, herding and horticulture still predominate throughout the rural, mountainous ranching communities today (Crosby 1981). While crop production was made possible through the irrigation of desert springs (the only permanent source of surficial fresh water that exists in BCS) (de Grenade et al. 2016), ranching was (and still is) dependent on seasonal rains to provide animal fodder. Because the Baja California peninsula was poorly integrated into the larger Mexican nation-state throughout the 19th century (Martinez 1965; Trejo-Barajas & Gonzalez 2002), people were highly reliant on local climatic conditions and ecology for making a living (Crosby 1981).
Data: Birth records were extracted from the, Guía Familiar de Baja California: 1700-1900 (Martinez 1965). This repository represents the largest source of vital records for the Baja California peninsula prior to the first systemic, nation-wide censuses that were initiated at the beginning of the 20th century (Platt 1998). Due to weathering and neglect over 200+ years, some of the vital records were damaged and not transcribed. However, data loss is unbiased and does not favor (or disfavor) any particular group or time period (Macfarlan et al 2020). In all, this repository represents the single best source for reconstructing demographic process on the peninsula during the 19th century (Macfarlan et al. 2019).
A total of 9111 birth, baptismal, and registration records were extracted from ten communities spanning the modern state of BCS (Loreto, Santiago, Santa Rosalía, Mulegé, Comondú, Todos Santos, La Paz, San Jose del Cabo, San Antonio, San Ignacio) (Figure 1). Records included a variety of information including: 1) the first and last name of the individual, 2) the day, month, and year of birth, baptism, and/or registration, 3) the location where the individual was born, baptized, and/or registered, 4) parental first and last names, and 5) additional notes about the individual who was born. Some records included information on individuals who were baptized or registered in one location (e.g. Santa Rosalía) but were born in a different location (e.g. Todos Santos). As such, we organized the record such that they were associated with the correct birth location. When individuals were born in locations not directly associated with one of the ten identified communities, we searched for the location using a variety of published sources (e.g. Gerhard & Guilick 1964; Secretaría de Desarrollo Social 2020; Instituto Nacional de Estadística, Geografía e Informática 2020) and placed them into the closest community. Finally, we filtered the data to include only those records where month and year of birth were available (n=8905).
Community-level climatological data were aggregated from Mexican National archives (Ruiz et al. 2006). The data are presented as monthly temperature and precipitation averages that are based on daily measurements collected from weather stations over a 41-year period spanning 1961 and 2002 in each community (Table 2). While the time frame for weather and birth data do not overlap, the modern climatological system has been stable for at least the last several hundred years, if not longer (Baegert 1952; Diaz et al. 2001; Barron et al. 2012).
Community-level energy availability data are presented as average monthly Normalized Difference Vegetation Indices (NDVI). NDVI values (which range from negative one to one) quantify the spatio-temporal variability in green biomass (i.e., vegetation) by measuring the difference between remotely sensed red and near-infrared light (which vegetation reflects and absorbs, respectively) (Hamel et al. 2009; Pettorelli 2013). Data were queried from the MODIS/VIIRS Global Subsets Tool using a 16-day interval, 250-meter resolution across a 19-year period (January 1, 2001 through December 31, 2019) (Didan 2015). For each of the ten communities, we used three square-kilometer polygons to assess intra-annual energy availability (Table 2) (ORNL DAAC 2018a,b,c,d,e,f,g,h,i,j). To account for urbanization in the communities of San Jose del Cabo and La Paz, we offset our polygons to non-developed plots immediately adjacent to the community (ORNL DAAC 2018a,e). Although the NDVI data does not overlap temporally with our birth data, and NDVI values may be lower than that observed during the historical period due to urbanization, the seasonal change in NDVI values within communities should be comparable based on the stability of the climatic regimes over the last several thousand years (Baegert 1952; Diaz et al. 2001; Barron et al. 2012).
Tables 1 and 2 approximately here
Analytic Modeling: All statistical modeling was run in STATA/IC 16.1 (StataCorp 2019) and can be replicated using the associated Stata “do file” (see Supplemental Information). To examine whether precipitation and temperature or local energy availability impact monthly birth pulses, our birth record database was transformed into a monthly birth pulse database, whereby each row represented the number of births recorded in a particular community, in a particular month, in a particular year (See Data Supplement). The outcome variable was the number of births within a month. The independent variables included average precipitation by month, average temperature by month, average NDVI score by month, and the year of birth. To deal with the fact that some communities did not have births in some months and/or years, we employ a Zero-Inflated Poisson model which accounts for those cases where no births were recorded. Because data are nested within communities, we use clustered-robust standard errors at the level of the community.
RESULTS
As an initial data check, we perform OLS regression to assess the impact of average monthly precipitation and temperature on NDVI. Because the data are nested at the level of the community, we employ clustered robust standard errors around community. We find that both climatic variables are significantly related to monthly average NDVI (R2=0.51; p<.001; n-observations=120; n-groups=10), with NDVI positively associated with precipitation (B=0.003; p<.001) and negatively associated with temperature (B=-0.006; p=.002).