Extended Data Fig. 4: Percent change in turtle dancing responses for all turtles.

Percent change = \(\frac{{Rewarded\; field\; turtle\; dancing}-{Unrewarded\; field\; turtle\; dancing}}{{Unrewarded\; field\; turtle\; dancing}}\,* \,100\). Red dotted lines indicate 0% change relative to the unrewarded field. Dots represent the percent change for individuals; dot color corresponds to the rewarded magnetic field as indicated on the figure. All data were analyzed with one-tailed Wilcoxon signed-rank tests. (a) Turtles conditioned to magnetic fields near New Hampshire, U.S.A. and the Gulf of Mexico had a percent change in dancing behavior significantly greater than zero (w = 127, p = 0.0005, Hedge’s g = 0.80, n = 16). (b) When these same turtles were tested four months after conditioning ended, without exposure to either field in the interim, percent change in dancing was again significantly greater than zero (w = 124, p = 0.001, Hedge’s g = 0.85, n = 16). (c) Turtles conditioned to Delaware, U.S.A. and Cuba had a percent change in dancing behavior significantly greater than zero (w = 118, p = 0.004, Hedge’s g = 0.45, n = 16). (d) Turtles with a rewarded field of Maine, U.S.A. had a percent change in dancing significantly greater than zero (w = 121, p = 0.002, Hedge’s g = 0.80, n = 16). (e) Turtles conditioned to Newfoundland, Canada and Virginia, U.S.A., had a percent change in dancing significantly greater than zero (w = 120, p = 0.003, Hedge’s g = 0.63, n = 16). (f) Turtles conditioned to Haiti and the Turks and Caicos had a percent change in dancing significantly greater than zero (w = 99, p = 0.0009, Hedge’s g = 0.94, n = 14). Collectively, these analyses of percent change corroborate the findings based on raw data in Figs. 1 and 2.