Figure 2: Long-term differential response between trophic levels to diminishing Arctic sea ice.

(a) Time series of observed (black triangles, black line) and predicted (grey lines) timing of 50% plant species emergence at the study site near Kangerlussuaq, Greenland, based on January (or February) and June multiple regression (y=β₀+β₁Jan(or Feb)E+β₂JunE+ε, where y=day of year of 50% plant species emergence in year t, Jan(or Feb)E=January (or February) Arctic-wide sea ice extent (10⁶ km²) in year t, JunE=June Arctic-wide sea ice extent (km²) in year t). 1993 (red triangle) was not included in the parameterization of the model, but is plotted in a to show the minimal difference between observed and predicted values for that year. The gap in the January- and June-modelled phenology series reflects a gap in the satellite-based sea ice extent record (1988). (b) Observed dates of 50% caribou calves born at the study site near Kangerlussuaq, Greenland (black circles, solid black regression line) plotted over the same phenology models as a. These time series document a divergence in the phenologies of caribou and the plant community in recent years, but also demonstrate a sustained flip in the timing of plant growth relative to herbivore parturition since the year 2000, a change driven primarily by earlier plant phenology relative to the comparatively fixed timing of caribou calving.