Fig 1.
Samples distribution across Sweden.
Moose bones (white dots) and antlers (white triangles) are plotted over the geolithological map of Scandinavia. Main stratigraphic units by age are reported; type of lithology is indicated when explicitly reported in the USGS source file: v = volcanic, i = intrusive, m = metamorphic. The map is based on the European geo4_2l shapefile from USGS [114].
Fig 2.
Strontium, oxygen and carbon isotope data of moose antlers and bones.
A) Boxplot of strontium, oxygen and carbon isotope data of antlers (green; n = 15) and bones (orange; n = 50). B) δ18O values of antlers and bones versus their respective 87Sr/86Sr and C) δ13C values; color coding as in (A); grey lines are linear regressions through the data. D) Voronoi diagram of moose’s Sr isotope values plotted over the map of Sweden. E) Voronoi diagram of moose’s oxygen isotope values plotted over the map of Sweden.
Fig 3.
A) 87Sr/86Sr isoscape built through Random Forest with n = 7 external predictors; B) δ18O isoscape (mean annual precipitation values) downloaded from waterisotopes.org. Maps are colored through a quantile scale (q0.1, median, q0.9); min and max values for the 87Sr/86Sr isoscape (A) are: 0.705 and 0.748, respectively; while for the δ18O isoscape (B) are -17.3 and -7.1‰.
Fig 4.
Data comparison between sample and isoscape.
A) Bone (n = 50) symbols are orange; while antler (n = 15) symbols are green; males (n = 29) are triangles; females (n = 17) are circles; individuals with unknown sex (n = 19) are squares. One antler is reported as ‘sex unknown’ being labelled as ‘hermaphrodite’ in the museum ID card (see S1 Table). 1σ-outliers are labelled; gray bars on the sides are histograms of univariate sample distribution for Δ87Sr/86Srsample-isoscape and Δ18Osample-isoscape; rug bars are also sample distributions but classified by sample type (bone and antler). Light grey areas are approximate median spatial uncertainties of the isoscapes (∼0.01 for Sr and 1‰ for oxygen) depicted below and above Δ = 0. B) The linear fit (1σ-outliers included) between the 87Sr/86Srsample and the 87Sr/86Srisoscape shows an R2 of 0.20, p < 0.01; C) The linear fit (1σ-outliers included) between the δ18Osample (water) and the δ18Oisoscape shows an R2 of 0.30, p < 0.01.
Fig 5.
Moose provenance probabilities estimated through a Bayesian approach.
Sr and O calibrated isoscapes were combined to predict the place of origin of the 1σ-outliers (see the Materials and Methods section). Red dots are the places of death of the individuals. Probability estimates are scaled between 0 (low probability, purple) and 1 (high probability, yellow). Sample’s names are reported in green for antlers and orange for bones.
Fig 6.
Probability distance of origin.
Probability density plots representing the likely distance of movements determined for 1σ-outliers (see the Materials and Methods section) using assignR on the calibrated isoscapes. Bones are orange; antlers are green. Dashed line is an arbitrary cut-off value of 100 km (see Discussion).
Fig 7.
Top 5% probability of origin of two archaeological moose samples based on their Sr isotope ratio.
Sample Kanaljorden_1 is a tooth pendant (enamel 87Sr/86Sr = 0.72086) while Kanaljorden_2 is an incisor (enamel 87Sr/86Sr = 0.73489). Data are from Eriksson et al. [166]. The red dot is the site of Kanaljorden. Yellow represents highly probable areas of origin (top 5%, see [138]), while purple represents low probable areas of origin.