Technically, this is a false statement. Normalization identifies how many more tables your data topic requires to isolate the data elements to avoid duplication or excessive convolution of data. Normalization is a type of analysis to optimize data layout given your topic's elements and the general concepts of relational databases. If you don't want to normalize, fine. But at some point you might as well use spreadsheets. Using a massively complex schema that doesn't include normalization is like painting yourself into a corner. I acknowledge it is your corner to paint. But understand that when you come to us for advice, we will give it even if it is advice you didn't want to hear.
OK. Lemmings go over a cliff because other lemmings want it to be that way and that is the end of that, too. Your response presumes that the ones who want it a certain way understand relational database concepts. Care to wager on that?
Taxonomy is a real pain in the patootie because it is a tree structure with several uneven branches involved in full identification. I say "uneven branches" because by the time you get to hominids, there are only seven living member species in four genuses (including homo sapiens... us, on a one-member genus). But if you look at Astrolagus, it has 3300 species at the genus level. Very unbalanced. All I can say is, good luck dealing with a seriously unbalanced tree structure without normalizing to some degree.