It has recently been argued that the geometric phase (GP) effect is not important for the differential cross-sections (DCSs) of the H + D-2 reaction at E-tot = 1.481 eV, in sharp contrast with previous calculations. It is shown here that this argument is based on an invalid extrapolation from partial wave sums truncated at J(max) = 5 to converged results. It is also shown that the GP effect is strongly affected by resonances, whose characteristics are very sensitive to the potential energy surface (PES). Finally, it is argued that reliable theoretical predictions and identifications of experimental resonances require that converged GP calculations be performed with sufficiently accurate PESs.