It depends a bit on what level you are seeking for an answer:
One can say: Because a simulation says so. After all, the simulation is nothing but a bunch of calculations, which you could theoretically do by hand. (Mind that I have no idea whether your particular statement is based on a simulation.)
One can consider the Reynolds number and the Hagen–Poiseuille equation. However, bear in mind that the Reynolds number is only a rough heuristics. You can make robust predictions for very high and low Reynolds numbers, but in between, the type of flow depends on the exact shape of the obstruction.
Thus, one can say that a narrower segment is faster and flow speed begets turbulence. Additionally irregularities tend to beget turbulence. Thus, if you are at the border between laminarity and turbulence, these things can make a crucial difference.
One can answer from a mesoscopic point of view and observe: If a fluid needs to go around a corner (such as after an obstruction), it’s already doing half an eddy and thus more likely to form one.
Neither of these answers is wrong, but they address different aspects and have different usefulness depending on what you want to know. For example, a simulation probably gives you the most robust result for a given situation, but it won’t give you conceptual understanding or yield general patterns such as “bottlenecks are potential sources of turbulence”.