What's the right approach to aggregate information from distinct components in layered software?

Suppose I have a component on the business layer that needs to aggregate information from the layer below (eg.: a CustomerBO depends on CustomerDAO data) and another information that is present on both the layer below and on a sibling component (eg.: this same CustomerBO depends on information from OrdersBO that is also on OrdersDAO).

What is the correct approach on this scenario and why? Should CustomerBO access CustomersDAO and OrdersBO, or CustomersDAO and OrdersDAO is the right way to go?