S1 Fig (A) Bayesian network model of nest site of Kentish plovers.

We set the possibility of nest site designation at 100% of the nest site. The 97% category, in the sand cover node. DV = dry vegetation, TV = total vegetation, LV = live vegetation, A = absence, P = presence. (C) Bayesian network model of sand cover of Kentish plovers. We set the possibility of sand cover at 100% of each category. The BN models demonstrated that the possibility of a nest site in the node of nest site designation decreased as the ambient sand cover increased. (D) Bayesian network model of total vegetation cover of Kentish plovers. We set the possibility of total vegetation cover at 100% of each category. The BN models exhibited that the possibility of a nest site in the node of nest site designation was higher when the ambient total vegetation cover was present to some extent. (E) Bayesian network model of nest site selection of Kentish plovers depending on locations. We set the possibility of each location at 100%. The BN models exhibited the local differences of vegetation and sand cover. (F) Bayesian network model of nest success and fail of Kentish plovers prepared. (A) We set the possibility of nest fate at 100% of nest success. The categories exhibiting the highest possibility of each node were type A (major type: sand) in the cover type node, type 3 (major type: very fine sand) in the sand type node, and the 230 m categories showed higher possibility than did the other categories in the NWD node. (B) We set the possibility of nest fate at 100% of nest failure. The categories exhibiting the highest possibility of each node were type A (major type: sand) in the cover type node, type 1 (major type: fine sand) in the sand type node, the S2 Fig (A) Code of nest site selection model. (B) Code of cover type cluster analysis. (C) Code of soil type cluster analysis. (D) Code of nest fate model. (E) Code of nest failure PCA analysis. (ZIP)