Fig. 2

Variable importance plots for gradient boosted models using presence/pseudo-absence of satellite tracked whales displaying low move persistence (presumed foraging movement behaviour) in North West WA (a), South West WA (b) and southern Australia (c); displaying low move persistence in Southeast Asia (presumed foraging/reproduction area) (d); and during high move persistence (migration movement behaviour) off western Australia (e) and Southern Australia (f). Variable importance is presented in relation to their contribution to reducing error rate, where the higher the value the more important the variable. Variable names are described in Table 1