Per the article, there is really very little evidence that anything is happening at all with sperm counts. There is a good chance that there is no effect, that it's all measurement artifacts and/or localized changes.