Supplement materials

Journal name: Fisheries Science

Title: Spatiotemporal catch distribution of age-0 Pacific bluefin tuna Thunnus orientalis caught by the Japanese troll fishery in relation to surface sea temperature and seasonal migration

Author names: Momoko Ichinokawa1*, Hiroshi Okamura1, Kazuhiro Oshima2, Kotaro Yokawa2, and Yukio Takeuchi2

Affiliation: 1 National Research Institute of Fisheries Science, Fisheries Research Agency, 2-12-4 Fukuura, Kanazawa, Yokohama, Kanagawa 236-8648, Japan; 2 National Research Institute of Far Seas Fisheries, Fisheries Research Agency, 5-7-1 Orido, Shimizu, Shizuoka, Shizuoka 424-8633, Japan

E-mail address of the corresponding author: Momoko Ichinokawa

(email: , tel: +81-45-788-7645)

Content description: Diagnostics and results of model selection in the binomial (S1) and lognormal (S2) models. S3 represents catches predicted by area and month in binomial and lognormal models.

S1. Diagnostics and results of model selection in the binomial model

Table S1. Results of model selection with different random effects and the same set of fixed effects, month, area and month×area

Random effects / AIC / Standard deviation estimated in the random effects
year / port / year×port
year, port, year×port / 2068.5 / 0.30 / 0.77 / 1.05
year×port / 2110.1 / - / - / 2.16
port, year / 2136.0 / 0.27 / 0.66 / -
port / 2183.5 / - / 0.59 / -
year / 2279.1 / 0.20 / - / -
no random effects / 2318.1 / - / - / -

Fig. S1. Plots of conditional Pearson residuals for the best selected model in the binomial model

S2. Diagnostics and results of model selection in the lognormal model

Table S2. Results of model selection with different random effects and the same set of fixed effects, month, area and month×area

Random effects / AIC / Standard deviation estimated in random effects
year / port / year×port
year, port, year×port / 6610.7 / 0.66 / 0.27 / 0.42
year×port / 6687.1 / - / - / 1.29
port, year / 6649.2 / 0.71 / 0.32 / -
port / 6868.8 / - / 0.26 / -
year / 6738.7 / 0.64 / - / -
no random effects / 6927.4 / - / - / -

Fig. S2. Plots of Pearson residuals for the best selected model in the lognormal model

S3. Catches predicted by area and month in the binomial and lognormal models

The standardized catch by area and month was a product of the positive catch predicted by the lognormal model and probability of positive catch predicted by the binomial model as described by Eq. 3 in the main text. Fig. S3 shows the predicted catches from the two separate models. Those results indicate that the standardized catches by area and month (Fig. 5) were determined mainly by spatiotemporal variability of the positive catches.

Fig. S3. Predicted positive catches (a, b) and probability of positive catches (c, d) in the Tsushima Current region (a, c) and Kuroshio Current region (b, d)