Past studies that attempted to quantify the spatio-temporal organization of seismicity have defined the conditions by which an event and those that follow it can be related in space and/or time. In this work, we use the simplest measures of spatio-temporal separation: the interevent distances <i>R</i> and interevent times <i>T</i> between pairs of successive events. We observe that after a characteristic value <i>R</i><sup>*</sup>, the distributions of <i>R</i> begin to follow that of a randomly shuffled sequence, suggesting that events separated by <i>R</i> > <i>R</i><sup>*</sup> are more likely to be uncorrelated events generated independent of one another. Interestingly, the conditional <i>T</i> distributions for short-distance (long-distance) events, <i>R</i> ≤ <i>R</i><sup>*</sup> (<i>R</i> > <i>R</i><sup>*</sup>), peak at correspondingly short (long) <i>T</i> values, signifying the spatio-temporal clustering (separation) of correlated (independent) events. By considering different threshold magnitudes within a range that ensures substantial catalogue completeness, invariant quantities related to the spatial and temporal spacing of correlated events and the rate of generation of independent events emerge naturally.