We use a kinetic-equation approach to describe the propagation of ultra high energy cosmic ray protons and nuclei and calculate the expected spectra and mass composition at the Earth for different assumptions on the source injection spectra and chemical abundances. When compared with the spectrum, the elongation rate $X_{max}(E)$ and dispersion $sigma(X_{max})$ as observed with the Pierre Auger Observatory, several important consequences can be drawn: a) the injection spectra of nuclei must be very hard, $sim E^{-gamma}$ with $gammasim 1-1.6$; b) the maximum energy of nuclei of charge $Z$ in the sources must be $sim 5Ztimes 10^{18}$ eV, thereby not requiring acceleration to extremely high energies; c) the fit to the Auger spectrum can be obtained only at the price of adding an {it ad hoc} light extragalactic component with a steep injection spectrum ($sim E^{-2.7}$). In this sense, at the ankle ($E_{A}approx 5times 10^{18}$ eV) all the components are of extragalactic origin, thereby suggesting that the transition from Galactic to extragalactic cosmic rays occurs below the ankle. Interestingly, the additional light extragalactic component postulated above compares well, in terms of spectrum and normalization, with the one recently measured by KASCADE-Grande.