Ed- tropical tropospheric temperature responds to the whole SST field, not just ENSO. But the plot in fig. 1 only compares Nino3.4, not the whole SST pattern (or, for example, the tropical mean SST, which one can take as a simple measure of what we expect to most closely influence the tropospheric temperature) to tropospheric temperature. When there is a big ENSO event, that's what you see in TT, because it dominates the tropical SST field. Nino3.4 in fact is very closely correlated to the tropical mean SST over time (with a few months' lag), for reasons explained in the post - the Nino3.4 influences the rest of the tropical SST via the troposphere. But if there is any other factor going on - either some forcing (such as a volcano) or non-ENSO internal variability - then there is no reason why the Nino3.4 in particular would have to keep the same relationship to the tropospheric temperature, because it would not have the same relationship to the SST outside the Nino regions either, and those other regions' SSTs can influence the tropospheric temperature. That in fact is the point of the post; the anomaly in the early 90s is just another example, in broad terms (though probably not in detail) similar to what is going on now. I don't have a more detailed explanation for exactly what was going on in the 90s - that would require some work - but Pinatubo could well have played a role.