replacement level robustness

[beanumber]

Are the replacement levels robust? Can the robustness of these estimates be improved with data from multiple seasons?