For open surfaces, there are counterexamples. The first one was constructed by P. Myrberg:
Ueber die analytische Fortsetzung von beschrankten Funktionen, Ann. Acad. Sci. Fenn., Ser. A. I N:o 58 (1949)
Since this paper is difficult to obtain (and written in German), I refer to another paper
Heins, Maurice, Riemann surfaces of infinite genus. Ann. of Math. (2) 55 (1952), 296–317,
Which proves an even stronger result: there is an open Riemann surface, such that if you remove a disk from it, then on the remaining surface every non-constant meromorphic function takes all complex values, except at most two of them.