The possibility that nucleosynthesis in neutron star mergers may reach fissioning nuclei introduces significant uncertainties in predicting the relative abundances of r-process material from such events. We evaluate the impact of using sets of fission yields given by the 2016 GEF code for spontaneous (sf), neutron-induced ((n, f)), and β-delayed (βdf) fission processes which take into account the approximate initial excitation energy of the fissioning compound nucleus. We further explore energy-dependent fission dynamics in the r process by considering the sensitivity of our results to the treatment of the energy sharing and de-excitation of the fission fragments using the FREYA code. We show that the asymmetric-to-symmetric yield trends predicted by GEF 2016 can reproduce the high-mass edge of the second r-process peak seen in solar data and examine the sensitivity of this result to the mass model and astrophysical conditions applied. We consider the effect of fission yields and barrier heights on the nuclear heating rates used to predict kilonova light curves. We find that fission barriers influence the contribution of 254Cf spontaneous fission to the heating at ∼100 d, such that a light curve observation consistent with such late-time heating would both confirm that actinides were produced in the event and imply the fission barriers are relatively high along the 254Cf β-feeding path. We lastly determine the key nuclei responsible for setting the r-process abundance pattern by averaging over thirty trajectories from a 1.2–1.4 M⊙ neutron star merger simulation. We show it is largely the odd-N nuclei undergoing (Z, N)(n, f) and (Z, N)βdf that control the relative abundances near the second peak. We find the 'hot spots' for β-delayed and neutron-induced fission given all mass models considered and show most of these nuclei lie between the predicted N = 184 shell closure and the location of currently available experimental decay data. [ABSTRACT FROM AUTHOR]