Dear Joel,
Cladogenetic events happen exactly *at* the speciation event (the node), and can be either sympatry (each descendant inherits the same range as the ancestor), vicariance (each descendant inherits part of the range of the ancestor; can be narrow or wide vicariance), subset sympatry (one of the descendants inherits a single area from the ancestor's range, while the other inherits exactly the same range as the ancestor) or founder-event speciation (one of the descendants inherits the same range as the ancestor, while the other descendant "moves" to a single area that is not contained in the ancestor's range).
Anagenetic events (any number of dispersals or extinctions) occur along the branchs. So you can easily have a cladogenetic event (at the node) followed by an anagenetic event (along the branch). When you look at a BioGeoBEARS output figure, you will notice that there are probabilities of ranges at the bottom of a branch, by the sides of the node. This is the probability of the range occupied by the descendants, immediately after the cladogenetic event. On the other hand, you have probabilities at the top of the branch. In the example below, you have a dispersal event (A >AB) followed by a vicariant event (ancestor: AB > descendant 1: A + descendent 2: B), followed by a dispersal event (B > BC in descendant 2).
A
------------------------
A
(A > AB) |
A ----------------- AB
| (B > BC)
B ----------------------- BC
(probability (probability at the
immediately
end of the branch)
after cladogenetic
event)
So, to sum up, it is not contradictory. Each node can have both cladogenetic events and anagenetic events associated (the former happens at the node itself, the latter happens along the branch that immediately precedes the node). I don't exactly recall the output for BSM tables, but I think that for the example above, it would show both a cladogenetic event (AB > A + B) and an anagenetic event (A > AB) for the same node.
Best,
Ivan