Good one. If anyone else is curious, analyses like these usually bring in multiple types of data together, but often rely primarily on cluster sampling. Botanists will go to a specific plot of land and very thoroughly count the number of species (amongst many other metrics). These data points are abstracted (while taking into account factors like biome types of course) to obtain an distribution of expected species in different areas worldwide. We then take this figure and compare it with how many species we currently have documented
Accounting for all the different factors is obviously much more complex than what I described, but that's the gist of it
It seems like it would be very sensitive to outliers or just regular power law stuff where 1% of the biomes have 99% of the species-level diversity. If you that biome isn't one of the ones you thoroughly measure, you miscount: if it is, your efforts on the other biomes aren't actually that important.
Is not (just) unseen, is unclassified. Recognized as different stuff.
Everybody going to a coastal area in the west coast of US had seen Western gulls. The species split in the last decades by genetic analysis. Now we have western gulls and yellow-footed gulls. Same history with the European Herring gull and the Caspian gull. If you can still find a new species of big bird in California or Europe, hidden in plain sight, why not a shrub or a tree? There are plenty of ornithologists and much less people interested in botany, (not to mention virus, bacteria or fungus)
Several hybrids can become species also. Plants are clonal, so non fertile hybrids can still multiply and eventually became fertile. There are many species complex in trees that are really arcane. Oaks or Willows for example. Lots and lots of plants have changed its taxonomy in the last ten years.
