(Abridged) Recent results from the Pierre Auger Observatory (PAO) indicate that the composition of ultra-high-energy cosmic rays (UHECRs) with energies above $10^{19}$ eV may be dominated by heavy nuclei. An important question is whether the distribution of arrival directions for such UHECR nuclei can exhibit observable anisotropy or positional correlations with their astrophysical source objects despite the expected strong deflections by intervening magnetic fields. For this purpose, we have simulated the propagation of UHECR nuclei including models for both the extragalactic magnetic field and the Galactic magnetic field. Assuming that only iron nuclei are injected steadily from sources with equal luminosity and spatially distributed according to the observed large scale structure in the local Universe, at the number of events published by the PAO so far, the arrival distribution of UHECRs would be consistent with no auto-correlation at 95% confidence if the mean number density of UHECR sources $n_s >~ 10^{-6}$ Mpc$^{-3}$, and consistent with no cross-correlation with sources within 95% errors for $n_s >~ 10^{-5}$ Mpc$^{-3}$. On the other hand, with 1000 events above $5.5 times 10^{19}$ eV in the whole sky, next generation experiments can reveal auto-correlation with more than 99% probability even for $n_s <~ 10^{-3}$ Mpc$^{-3}$, and cross-correlation with sources with more than 99% probability for $n_s <~ 10^{-4}$ Mpc$^{-3}$. In addition, we find that the contribution of Centaurus A is required to reproduce the currently observed UHECR excess in the Centaurus region. Secondary protons generated by photodisintegration of primary heavy nuclei during propagation play a crucial role in all cases, and the resulting anisotropy at small angular scales should provide a strong hint of the source location if the maximum energies of the heavy nuclei are sufficiently high.