Stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotope analysis was conducted on modern and archaeological polar bear bone collagen from the Canadian Arctic Archipelago to investigate potential changes in polar bear foraging ecology over four millennia. Polar bear δ¹³C values showed a significant decline in the modern samples relative to all archaeological time-bins, indicating a disruption in the sources of production that support the food web, occurring after the Industrial Revolution. The trophic structure, indicated through δ¹⁵N, remained unaltered throughout all time periods. The lower δ¹³C observed in the modern samples indicates a change in the relative importance of pelagic (supported by open-water phytoplankton) over sympagic (supported by sea ice-associated algae) primary production. The consistency in polar bear δ¹³C through the late Holocene includes climatic shifts such as the Medieval Warm Period (MWP, A.D. 950–1250) and the early stages of the Little Ice Age (LIA, A.D. 1300–1850). These findings suggest that polar bears inhabit a food web that is more pelagic and less sympagic today than it was through the Late Holocene. We suggest that modern, anthropogenic warming has already affected food web structure in the Canadian Arctic Archipelago when modern data are contextualized with a deep time perspective.

Ancient bone samples were collected from 35 polar bears from 10 archaeological sites in the region inhabited by the Lancaster Sound polar bear subpopulation. The archaeological samples came from pre-Dorset (N= 10; 4000–2800 BP), Dorset (N= 15; 1500–700 BP), and Thule (N= 10; 700–500 BP) sites, and consisted of several different anatomical elements but each element represented a distinct individual. The archaeological samples could not be sexed or aged. Modern bone samples (N= 11; 1998–2007 CE) were obtained from 11 individuals harvested within a 150 km radius of the communities of Ikpiarjuk (Arctic Bay), Aujuittuq (Grise Fiord), and Qausuittuq (Resolute) between 1998 and 2007. The modern samples consisted of polar bear bacula, collected as a mandatory sample by subsistence hunters. The modern samples were, therefore, male and ranged in age from three to 8-years-old.

 For the archaeological specimens, chunks of bone weighing ~200 mg were sampled using an NSK dental drill equipped with a diamond-tipped cutting wheel. Samples were demineralized in 16×100 mm glass culture tubes with 0.5 M HCl at room temperature. After the samples were demineralized they were rinsed three times with Type I water (resistivity >18.2 MΩ·cm). The samples were treated with 0.1 M NaOH to remove humic contaminants from bones that exhibited a dark coloration. After 30 min, the NaOH was removed and samples were rinsed twice with Type I water. The samples were then placed in a solution of 0.01 M HCl and placed in a dry bath at 75˚C for 36 h, to solubilize the collagen. The collagen was extracted from the modern specimens using the same protocol with the following exceptions. First, powdered bone was removed from the samples using a Dremel equipped with a rotary burr. Because modern bone contains significant quantities of lipids whereas ancient bone does not, the samples were first treated with 2:1 chloroform:methanol under sonication for 1 h. The powdered samples were demineralized in 0.5 M HCl for 4 h. Demineralization in HCl does not alter the stable carbon or nitrogen isotopic composition of the bone collagen.

The solution containing the solubilized collagen was then transferred into pre-weighed 4 ml glass vials and freeze-dried. Collagen samples (0.45−0.55 mg) were weighed into tin capsules for analysis with a EuroEA 3000 (Euro Vector SpA) Elemental Analyzer coupled to a Nu Horizon (Nu Instruments) continuous flow isotope ratio mass spectrometer at the Water Quality Centre at Trent University. Ten percent of samples were analyzed in duplicate to assess homogeneity. Analytical sessions were calibrated using international standards USGS40, USGS41a and USGS66. Accuracy and precision were assessed with in-house check standards: SRM-1 (caribou bone collagen), SRM-2 (walrus bone collagen), and SRM-14 (polar bear bone collagen). The standard uncertainty across analytical sessions was calculated to be ±0.13 ‰ for δ¹³C and ±0.24 ‰ for δ¹⁵N.

Data Treatment

Modern bone samples may be particularly prone to contamination with lipids under certain circumstances. Despite lipid extraction with a commonly used protocol, we observed a correlation between the atomic C:N ratios and the δ¹³C values of the collagen even though our highest C:N ratio was 3.30. This observation is consistent with increasing quantities of residual lipids causing lower δ¹³C values as C:N ratios rise. No additional collagen from the modern specimens was available for further chemical pretreatment after the initial analyses. We therefore applied a mathematical correction to the modern bone collagen δ¹³C values using the following equation:

Equation 1:  δ¹³C_{lipid corrected} = δ¹³C_measured + (10.122)*(C:N_measured - C:N_expected)

δ¹³C_measured and C:N_measured are the isotopic and elemental compositions determined for the sample prior to any adjustments. We used an expected C:N ratio for mammalian bone collagen of 3.23 (Guiry and Szpak, 2020), and 10.122 represents the slope of the regression line when plotting our modern C:N_measured and δ¹³C_measured.

When comparing the δ¹³C values of modern and ancient samples, it is important to correct for the global decrease in δ¹³C of atmospheric and oceanic CO₂ since the beginning of the industrial revolution. A correction was calculated using Equation 2:

Equation 2:  ?¹³?_????? = (?_{????? ????}) × ? (? − 1850)^?

where α is the annual rate of decrease in δ¹³C specific to the water body (we used a value of 0.014 ‰ for the Northwest Atlantic), y is the year of sample harvest and b is the global oceanic decrease in δ¹³C (0.027 ‰).

The carbon isotopic compositions of the modern samples have undergone correction through Equations 1 and 2 and the revised carbon isotope values are denoted δ¹³C_corr. Both of these corrections have been applied to account for confounding causes of low δ¹³C values in modern samples to avoid an exaggeration of any differences between modern and ancient samples since both of these adjustments increase the δ¹³C values of the modern samples.