This article mentions but talks around the idea of their simply being a step-change in sperm counts due to lifestyle factors - i.e. the observation that sitting

This article mentions but talks around the idea of their simply being a step-change in sperm counts due to lifestyle factors - i.e. the observation that sitting can reduce sperm count, and more people have office jobs. If you take an average in a population, and then re-arrange the population distribution to the lower-side of the step change, then take another average your result would look exactly like a decline over time, even though there's no actual ongoing component to it - the sitting population, if identified would have a lower normal rate that's not decreasing otherwise. Which would give you a real effect on average, but not an actual trend you can extrapolate since you're just observing two averages for two population groups, where the membership of each is varying. Which again - is alluded to, but I found it weird this mechanism wasn't called out explicitly (except under the "mysteriously stops" conclusion).