I recomputed the times for the Spring Equinox using JPL's HORIZONS Web-Interface for the years 2010-2020 and got similar results to those for MICA for the same period. The times were all shifted by a few hours and the mean value for the period differed slightly but the plot was similar overall.
I couldn't find an explanation for magnitude of the deviations. The changes in the Earth's orientation due to nutation are of the order of a few seconds of arc. The effects of gavitational perturbations accelerate and retard the Earth in its orbit about the Sun and can produce cummulative changes.