Fig. 3

The 5-day averaged success rate of the turtles in the mixed numerosity tests (Experiment 2). Each numerosity pair was tested twice a day in a random order on each of the five turtles for five days. A Phase I comprised numerosity pairs of small numbers (1–5) with a ratio ranging 0.2–0.8; B Phase II comprised numerosity pairs of at least one large number (6–10) with a ratio ranging 0.2–0.8; C Phase III comprised numerosity pairs of at least one large number (6–10) with a ratio ranging 0.2–0.9. The solid lines with shaded areas are predicted mean probabilities of success ± 1 standard error, estimated from the best-fit model (Table 2 & Additional file 1: Table S3). The dots are the observed success rates with 10% jittering using jitter function in R v3.6.1, and the horizontal dashed line denotes random choice of the smaller and larger quantity