You don't have to reproduce the results to implicitly verify them. As I understand it, the normal pattern is that if the results are important, researchers will do stuff based upon then, and if the results were wrong their experiments won't go right. Eventually they try reproducing the original to figure out what's wrong, and that can lead to people deciding the original was wrong.
If the results aren't important, they don't tend to get built upon, they mostly just matter for someone getting tenure, the next grant, whatever.
For the rest of us it all boils down to trusting those held in respect by their peers and that is a belief system. Occasionally reinforced by shunning the nay-sayers as "anti-science"...