This is one of my favourite figures in Zieminska et al 2013 paper. It is a relationship between wood density and fibre fraction across 24 species (trees and shrubs from Australia, each donut represents a single species, the subscript ‘NV’ is not important here). On a cross section, fibres look like donuts (see below). They can have from thin to thick cell wall (the pink area). The more cell wall in wood the higher wood density.
The black donuts on the graph represent actually a fibre of a given species. And one can quickly notice why two species with the same fibre fraction (for example the ones with nearly 0.8 fibre fraction) can have very different wood densities. The species at the bottom right of the graph has a very narrow donut ring, which means there is little cell wall. While the species at the right left of the graph has a very thick donut ring, which means there is a lot of cell wall. So within the same fibre fraction there can be a lot of wall resulting in high wood density all the way down to very little wall resulting in low density. So the density is not driven by the fraction of fibres but rather by the fraction of fibre cell wall and fibre lumen (the ‘hole’ in the middle of a fibre).