Highest-ranked GLMMs (ΔAICc<2) for observer chick-aging accuracy (correct or incorrect estimation of chick age) in learning-phase trials.

Fixed factors included tool use (“Tool”), number of previous trials of the same type (either with or without the tool [“Learning”]), field experience (≥1 yr field experience with Common Terns = experienced, [“Exp”]), chick age group (2–6 as there were no Group 1 chicks in tests at Gull Island [“Group”]), and colony site location (“Colony”). Tool use was retained in 67% of the top six ranked models: models ranked second and fourth were the only ones not to retain tool use as a factor. Model selection (reduction) began from the maximal model (not shown) that included all two-way and three-way interactions. Number of parameters (k), and AICc weights are given.Highest-ranked GLMMs (ΔAICc