FIG. F1. Model outputs for 120^oW longitude (open symbols) and 70^oW longitude (closed symbols) are presented. We determined whether each lynx sample was collected from a population that was increasing (including the low point in the cycle) or decreasing (including the cycle peak) by visually examining published lynx harvest numbers (Elton and Nicholson 1942, Novak et al. 1987, Healy 2001, Statistics Canada 2005). Separate models relating carbon and nitrogen stable isotope ratios to longitude, latitude, season, cycle phase and their interactions were fit to the data. Since only six observations occurred in spring during a decreasing cycle, no interactions between season and cycle phase were included in the model. δ¹⁵N values were significantly greater during the decline phase than the increase phase (F_1,191 = 25.05, P < 0.001). For δ¹³C there was a significant latitude by longitude by phase interaction (F_1,188 = 4.25, P = 0.041).