That unfortunately has to do with the Speed Tree implementation and the work of the content developer. If the content developer used additional leaf and branch colors rather than the common green for leaves and brown for branches, there would be more variety in the forests. Speed Trees also tend to be of a uniform height and shape so that they have that uniform size, color, and shape even in the distance.
We also have the drop off in LOD, which causes the Speed Trees to look spindly too soon so we don't quite have the cover that we should have in the Northeast. That jungle in the Southern Tier of New York in the Susquehanna Valley area to be exact, Western Pennsylvania, and all across the rest of New England can never, ever, be modeled as it should, otherwise, we will melt our computers.
But as I've said before, my complaint about the Speed Trees is they are too perfect and too uniform, and usually too big. They work well for those locations where we want perfect trees such as in towns along the fancy boulevards and avenues, in parks, and in the yards and gardens of big fancy estates. The Speed Trees as we have them just don't work well outside of those areas, and the billboard/flip board trees, just don't cut it either because they may have the right shape, but look like paper and mylar strips glued and stapled together a star-shape to represent trees. They looked bad before, but with the new lighting in TRS19, they glow in the dark.
What we need is a nice low-cost tree library that is compatible with the existing Speed Tree enging that is easy to use so we can all create trees that suit our needs. Maybe N3V could develop a random tree generator based on specific parameters, such as region, type, season, and let the system generate a procedural forest for our routes.