Branch & bound: why does the lower bound increase?

For a minimization problem, when running a branch & bound algorithm, I understand that:

• Every integer feasible solution provides an upper bound on the optimal objective value of the original problem.
• The lower bound for each branch-and-bound node arises from solving the linear programming relaxation at that node, and the minimum of lower bounds (across all active leaf nodes) is a lower bound on the optimal objective value of the original problem.

I am having some troubling understanding the following:

1. How can the minimum of lower bounds not be at the root node?
2. In the graphs of this link, we can see that the lower bound is increasing with time. Why (or how) does this bound change during thevprocess?
• Don't forget any lower bound is a global lower bound. It's a guarantee that there is no solution with an objective value lower than this bound. It certainly wouldn't make sense for it to decrease over time. Commented Apr 28 at 0:06