Extended Data Fig. 2: Results of additional map assay experiments plotted on a linear scale.

In four additional experiments, turtles discriminated between a magnetic field in which they were fed and one in which they were not. Turtles differentiated between magnetic fields that exist near: (a) Delaware, U.S.A. and Cuba (two-tailed Wilcoxon signed-rank test, w = 108, p = 0.04, Hedge’s g = 0.50, n = 16); (b) Maine and Florida, U.S.A. (two-tailed Wilcoxon signed-rank test, w = 121, p = 0.004, Hedge’s g = 0.63, n = 16); (c) Newfoundland, Canada and Virginia, U.S.A. (two-tailed Wilcoxon signed-rank test, w = 115, p = 0.01, Hedge’s g = 0.60, n = 16); and (d) the Turks and Caicos Islands and Haiti (two-tailed Wilcoxon signed-rank test, w = 97, p = 0.003, Hedge’s g = 0.60, n = 14). The data in (c) represent a second conditioning experiment conducted with the same turtles used in (b) and thus indicate that turtles can learn magnetic fields that exist at multiple locations. For each pair of magnetic fields, the rewarded field for the turtle is indicated by either ● or ▲ as indicated on the figure. Remaining conventions as in Fig. 1. The maps were created using Natural Earth (https://www.naturalearthdata.com; credit Tom Patterson and Nathaniel Vaughn Kelso).