Fig. 3: Change in trophic position of the brown bear (Ursus arctos) from Late Pleistocene to Holocene.
From: Dynamic omnivory shapes the functional role of large carnivores under global change
a Net primary productivity (NPP, kg C m−2 a−1) and meteorological growing season length (GSL, months with T > 0 °C), based on global climate and vegetation models (HadCM3 and BIOME4)41 during the last 55,000 years. Circles, thick and thin error bars represent mean NPP and growing season length, as well as 50% and 90% CIs [Confidence Intervals], across the subfossil bone and tooth samples of brown bear and red deer (Cervus elaphus) in each time bin (Methods). Lines indicate mean temporal trends in NPP and growing season length across all sample locations shown in (b). Abbreviations are LG, Late Glacial; GS9 to GI14, Greenland Stadial 9 to Greenland Interstadial 14; GS3 to GI8, Greenland Stadial 3 to Greenland Interstadial 8; GS2 to GI2, Greenland Stadial 2 to Greenland Interstadial 2; GI1, Greenland Interstadial 1; GS1, Greenland Stadial 1; M, Meghalayan; N, Northgrippian; G, Greenlandian. Sample sizes (n) are given at the bottom of the panel. b Locations of subfossil bone and tooth samples of brown bear and red deer across Europe. The background map is based on public data from Natural Earth (www.naturalearthdata.com). c Trophic position of the brown bear as estimated by a Bayesian hierarchical model with red deer as baseline, where a value of 2 corresponds to a strict herbivore (i.e., trophic level = 2) and values larger than 2 reflect an increasing dietary contribution of animal prey. Black circles, thick and thin lines represent posterior estimates (median, 50% and 90% ETIs, respectively). Sample sizes for brown bear (nU. arctos) and red deer (nC. elaphus) are given at the bottom of the panel. d, e Component + residual plots showing the partial relationships of trophic position with NPP and growing season length, conditioned on the effect of the other predictor variable. Black lines, dark and light gray bands represent estimated relationships and uncertainty (median, 50% and 90% ETIs, respectively). pd is the posterior probability of a relationship being the same direction as the median.
