Sexual dimorphism in bison, characterized by larger males and smaller females, allows cranial metrics to be used for sex differentiation due to relatively little metric overlap (Shackleton et al., 1975). For this study, five measurements were evaluated to separate small bulls from large cows (Table 2).

Using analysis of variation (ANOVA), morphology was compared to the North American dataset (Widga, 2007) to see if there were specific differences. Horn Core Curvature Indices ((SK4/SK5) * 100) were calculated using Skinner and Kaisen (1947: 142). Since the North American dataset only uses one measurement for both SK4 and SK5, while this paper presents left and right measurements, the average horn core curvature was calculated for comparison. When necessary, Tukey’s pairwise tests are used to clarify differences between groups.

Stable isotope analyses have successfully reconstructed life histories from archaeological and paleontological bison specimens (Chisholm et al., 1986; Hobson, 1999; Hoppe, 2006; Cannon, 2007; Widga, 2013; McKetta, 2017). Twenty-three of the most complete bison from the historic dataset with the most reliable provenience were sampled for radiocarbon dating and stable carbon and nitrogen isotope analyses of collagen from bison bone. Five modern bison samples, all from the Antlers Ranch, were submitted for stable carbon and nitrogen isotope analysis to compare with historic specimens. These modern samples are limited in sample size (n =5) and to a ranch where bison receive supplemental food during seasons when food is scarce, but the authors felt it was important to compare historic and modern bison to see how similar their diets were and whether any differences warrant further investigation.

Radiocarbon dating (¹⁴C) and stable isotope analyses (δ¹³C and δ¹⁵N) were conducted at the University of California, Irvine W.M. Keck Carbon Cycle Accelerator Mass Spectrometer Lab. Dates were calibrated using OxCal 4.4 (Ramsey, 2024) using the Int20 Calibration Curve (Reimer et al., 2020) and reported with 2-sigma ranges (see Table 3). All samples had C:N atomic ratios ranging between 2.9 to 3.6, indicating collagen is well-preserved (Ambrose, 1991; Davies et al., 2019).

Samples for stable isotope analyses came from the petrous portion (n = 20) and zygomatic arch (n = 3) for the historic dataset. Modern samples were taken from the petrous portion (n = 3) and maxillary molars (n = 2). A correction of -1.5‰ was applied to all historic specimens to account for anthropogenic CO₂ added to the atmosphere in the 20^th century (Tieszen, 1994). Collagen δ¹³C values were further corrected to an estimate of dietary input, -5.1‰ (Carlson et al., 2018). These corrected values are notated as δ¹³C_diet.

Bison forage can be divided into two plant groups based on their photosynthetic pathways: C₃ plants and C₄ plants. The δ¹³C of C₃ plants ranges between -35 and -22‰, with an average of -26.5‰, while C₄ plants range between -16 and -9‰, averaging -12.5‰ (Smith and Epstein, 1971; Chisholm et al., 1986; Carlson et al., 2018). To calculate the proportion of C₃ and C₄ plants in a bison diet, this study uses an equation adapted from Carlson et al. (2018):

As herbivores that both graze and browse (Craine et al., 2015; Hecker et al., 2021; McDonald, 1981; Meagher, 1986; Widga, 2006), bison are expected to consume a mixture of the available plants in their range (Chisholm et al., 1986; Tieszen, 1994). Sagebrush and desert shrubland dominate the Bighorn Basin, flanked by juniper woodland that transitions into Engelmann spruce–subalpine or lodgepole pine forests (Knight et al., 2014). Annual rain in the Bighorn Basin ranges from less than 254 mm in the center to nearly 508 mm in the foothills surrounding the Basin (Knight et al., 2014). C₃ plants dominate this semi-arid environment, although C₄ plants are present during the summer months (Knight et al., 2014).

Stable nitrogen isotope analysis (δ¹⁵N) has been used to study weaning behaviors (Schurr, 1998; Ambrose, 2002), assess trophic levels (Szpak, 2014), evaluate diet (Bird et al., 2021; Deniro and Epstein, 1981), and gauge nutritional stress (Ambrose, 1991, Ambrose, 2002; Funck et al., 2020). The dynamics of δ¹⁵N at the landscape level, however, are complex, with shifts attributed to numerous factors, such as water stress (Ambrose and Deniro, 1986) or diet (Bergmann et al., 2015; Berini and Badgley, 2017; Blackburn et al., 2021; Craine, 2021; Metcalfe, 2021). Research on bison δ¹⁵N suggests values vary with age, sex-group dynamics, and body size (Metcalfe and Olson, 2025).

Comparative analyses draw on both modern and historic datasets from various regions. The modern datasets include samples from the Bighorn Basin (n = 10), Theodore Roosevelt National Park (n = 5)—hereafter referred to as the modern Northern Plains (Davies et al., 2019)—and Yellowstone National Park (n = 15) (Feranec, 2007). Historic bison are derived from archaeological sites. Specimens referred to as “late Holocene Northern Plains” originate from North and South Dakota (n = 43) (Davies et al., 2019). Additional samples come from the Scoggin Site (n = 5), an archaeological site in the south-central portion of Wyoming. Previously reported radiocarbon dates from charcoal at the site place the Scoggin Site in the middle Holocene between 4,500 – 5,000 years B.P (McKetta, 2017). Recent redating using bison petrous portions (Todd, 2024, p. 10) revises these dates, placing the site in the late Holocene (UCIAMS #288515, 3590 ± 20 uncal B.P.; UCIAMS #299516, 3910 ± 20 uncal B.P.). Regardless of this update in radiocarbon ages, Scoggin provides a temporal contrast to bison from later in the late Holocene, immediately before the population bottleneck.

Although regional bison datasets with broader temporal coverage do exist (Cannon, 2008; Cannon et al., 2023), they were not included in this study for the following reasons. This study intentionally employs a tight temporal scale, focusing on the late Holocene. This temporal constraint attempts to reduce the potential confounding effects of major climatic shifts, which could obscure relevant patterns or introduce unrelated variability. Stable isotope analysis comparisons were limited to the geographic areas bordering the Rocky Mountains – namely, the northern and central Plains. Future analyses will focus on smaller, regional datasets and varying time scales. All statistical analyses in this study were conducted using PAST version 4.17 (Hammer, 2024).

Radiocarbon dates (Table 3) place the study bison immediately before the 19^th century population bottleneck event. Mass bison kills were increasingly common in the Rocky Mountains and Northern Plains from 200–1500 cal BP (Reher, 1977; Kornfeld et al., 2010). This was a period of extensive changes both climatic, such as the Little Ice Age (500–250 cal BP) (Alt et al., 2024), and anthropogenic, including the re-introduction of the horse in Wyoming (268 ± 26 uncal BP) (Thornhill, 2021); and the westward expansion of European peoples and livestock (Dobson, 2013).

Late Holocene bison from the Bighorn Basin exhibit a mean δ¹³C_diet value of -20.3‰ and range from -21.7 to -18.3‰, with an average consumption of 7.9% C₄ plants in their diets. An ANOVA conducted on the percentage of C₄ between the six datasets was significant, F(14.91) = 17.84, p ≤ 0.001. A Tukey’s pairwise test, summarized in Table 5, showed significant differences between several populations, including the Scoggin site and all other populations, as well as Yellowstone National Park and the late Holocene Northern Plains.

Late Holocene bison from the Bighorn Basin exhibit a mean δ¹⁵N value of 7‰ and range from 5.9 to 7.9‰. An ANOVA conducted on the δ¹⁵N showed significant differences (F (13.1) = 48.77, p = 0.000). A Tukey’s pairwise test revealed these differences are significant (p ≤ 0.01) between the modern Northern Plains dataset and late Holocene datasets of the Northern Plains and Bighorn Basin, as well as the Scoggin Site (Figure 4).

Bison horn cores respond to a series of both environmental (Cannon, 1997) and social (Guthrie, 1990) factors. If the observed trends are consistent with those in modern herds, the tendency for modern bison to exhibit higher horn core proportions (i.e., be longer and more curved) may result from several variables. As previously noted, an inherent bias from the modern dataset is that the bison tended to be especially large or long-lived. A larger, more representative sample would help evaluate the extent of this bias. The modern herds studied have controlled male-to-female ratios. In these settings, male bison do not need to compete for breeding access to the females and are sometimes segregated from the larger herd. Without the need to compete directly for mates, a herd structure where the short, upward pointing horn cores of Bison bison are advantageous, bull horn cores may become longer and more curved to visually signal reproductive fitness (e.g., Guthrie, 1990, p. 168-172). If modern bulls are not engaging in direct competition but still have access to females, then horn development may shift toward display features. The apparent difference between modern and historic bison could also result from bison spending time in different habitats (sensu lato Widga, 2013).

Stable carbon isotope results align with the expectation that bison diets will reflect a mixture of the C₃ and C₄ plants available in the area (Hoppe, 2006; Bouvier, 2022) and are comparable to contemporaneous bison from the Northern Plains (Davies et al., 2019). Significant differences occur in the percentage of C₃ plants between the diets of late Holocene bison and those from the Scoggin Site (middle to early late Holocene). Bison from the late Holocene Bighorn Basin consumed the most C₃ plants (92.1%), followed by those on the Northern Plains (90.6%), with bison from the Scoggin site consuming only 74.3% on average. Greater consumption of C₄ plants is expected for a herd further south, where the availability of C₄ plants increases (Lohse et al., 2014; Carlson et al., 2018). The greater consumption of C₄ plants may reflect the variability of the middle Holocene, when the climate shifted toward warmer, drier conditions (Knight et al., 2014; Davies et al., 2019), but systematic redating of the site as recommended by Todd (2024) would help to clarify these relationships.

Bison from both late Holocene populations consumed high amounts of C₃ plants but incorporated some C₄ plants into their diets. This is a departure from the diets of Yellowstone National Park bison, where no C₄ plants are observed (Feranec, 2007). Bull bison have been observed consuming a higher percentage of C₄ plants than cow bison (Post et al., 2001; Blackburn et al., 2021). The Bighorn Basin δ¹³C dataset is derived from male bison, so the mean 7.9% of C₄ plants in bison diets from the area may represent the upper range of C₄ plants utilized by bison in the area during the late Holocene.

The populations examined generally exhibit broad δ¹⁵N variability, except for modern Northern Plains bison, which display a more constrained isotopic range. At the population-level, historic bison from the Bighorn Basin possess the highest mean δ¹⁵N values. The complexity of the nitrogen cycle and expected fractionation warrant investigation into several potential causes of the increased mean δ¹⁵N of bison from this study over contemporaneous bison in the Northern Plains. These differences could result from diet or habitat.

Elevated mean δ¹⁵N values in the bone collagen of historic Bighorn Basin sample could reflect dietary differences between the mixed grass prairie of Theodore Roosevelt National Park (Morgan, 2020) and the sagebrush steppe of the Bighorn Basin (Knight et al., 2014). It might also reflect differences between bulls and cows. In Yellowstone National Park, bull bison show higher δ¹⁵N values than cows from the same area, indicating lower-quality diets (Berini and Badgley, 2017). In contrast, research from Elk Island National Park found a negative correlation between body size and δ¹⁵N, with bull bison displaying lower δ¹⁵N than cows (Metcalfe and Olson, 2025). These contrasting findings indicate that sex-based dietary differences are context-dependent and warrant further investigation. As a result, the elevated mean values observed in the historic Bighorn Basin bison cannot confidently be attributed to sex-based dietary differences. Other contributing dietary elements could include seasonal changes in diet, such as the selection of nitrogen-fixing forbs and shrubs in the early and late growing season (Bergmann et al., 2015; Blackburn et al., 2021; Craine, 2021), or a diet composed of relatively more wetland plants (Metcalfe, 2021). Bison graze on initial growth following fires both today (Shaw and Carter, 1990) and in the past (Roos et al., 2018), possibly contributing toward elevated δ¹⁵N (Pate and Anson, 2008; Turner et al., 2011; Dunnette et al., 2014).

A final possibility is that variation in mean δ¹⁵N reflects differences in the elevation gradients used by bison. Bison faunal material has been found throughout the Rocky Mountains (Fryxell, 1926; Cannon, 1997, Cannon, 2004, Cannon, 2007; Tirlea et al., 2022) and within the Absaroka and Beartooth Mountains on the periphery of the Bighorn Basin (Lee and Puseman, 2017; Widga et al., 2024). The bison at these sites were likely escaping heat, pests, and parasites of lower elevations during the summer months (Benedict and Barboza, 2022; Dixon, 2022; Tirlea et al., 2022). Nitrogen occurs in relatively higher amounts in forested areas than in grassland ecosystems (Asner et al., 1997). Bison moving from the Bighorn Basin to the surrounding mountains would pass through forests and up to the alpine tundra where nitrogen concentrations are high (Seastedt and Vaccaro, 2001; Knight et al., 2014). Plant growth in the alpine is restricted (Fisk et al., 1998) and thus is not likely to comprise a significant component of the bison diet; however, readily accessible water may provide the mechanism for elevated δ¹⁵N levels. This explanation would require more research, including comparison with the stable isotope ratios of bison recovered from ice patches. Both modern herds have lower δ¹⁵N values, possibly resulting from supplementary food (i.e., higher-quality diets) or changes in the composition of range plants.