Table 1. Akaike weights for North Pacific bigeye tuna data.

a, Properly defined Akaike weights [37], calculated here from the raw data (all individuals pooled together) using the equations in Box 1 of [14]. Respective log-likelihoods are and , giving Akaike Information Criteria of 236,256 and 232,599. b, Data for each individual were binned using the log-binning with normalization (LBN, [13]) technique, and regression lines fitted to all the points plotted on one figure (see Supplementary Fig. 3 of [27]). c, LBN method for all individuals pooled together [27]. d, LBN method with generalised linear mixed-effect models, using individual as a random factor [27]. e, Bayesian (rather than Akaike) Information Criteria [37] weights based on fitting linear regressions to rank/frequency plots [27] for all individuals pooled together. f, Same method as e but calculated here for just two models (result can also be deduced from Supplementary Table 7 of [27]).

a, 明确定义的赤池权重（Akaike weights）[37]由合并所有个体的原始数据计算得到，计算过程采用文献[14]方框1中的公式。对应的对数似然分别为和，据此得到的赤池信息准则（Akaike Information Criteria, AIC）值分别为236256与232599。b, 针对每个个体的数据，采用归一化对数分箱（log-binning with normalization, LBN）[13]技术进行分箱处理，并为所有绘图点拟合回归线后展示于同一张图表中（详见文献[27]的补充图3）。c, 针对合并后的全部个体采用归一化对数分箱方法[27]。d, 采用以个体作为随机因子的广义线性混合效应模型（generalised linear mixed-effect models）完成归一化对数分箱分析[27]。e, 基于对合并所有个体的秩频图（rank/frequency plots）[27]进行线性拟合的结果，计算得到贝叶斯信息准则（Bayesian Information Criteria, BIC）权重（而非赤池权重）[37]。f, 采用与e完全一致的方法，但仅针对两个模型开展计算，其结果也可通过文献[27]的补充表7推导得出。