Googling gives me figures of 100 trees/acre or less as sparse forest, and 200 trees/acre or more as "overgrown" in the real world.
I don't play 3.5E, but I think this is where you went wrong. My search for average forest trees/acre yields a much wider variation of figures, based on factors like individual species represented, average tree diameter, and the age of the stand. Here's a quote from a pdf on "Forest Thinning" put out by
http://4hforestryinvitational.org:
An adequately stocked stand will have fast-growing trees of good form. For example, consider an even-aged hardwood stand. There may be well over 10,000 seedlings per acre in the first 5 years. This number decreases to about 1,000 as the stand matures to a point when the trees average 5-11 inches DBH (pole-sized). Most of the young trees naturally die as other trees out-compete them for sunlight. When trees average more than 11 inches DBH (saw-timber size), the number of trees declines to 500 trees per acre, and eventually to 150 trees per acre in very mature woodlands.
The 100-200 trees per acre figure you quote appears to come from this website:
http://www.sbcounty.gov/calmast/sbc/html/healthy_forest.asp. Note this article is about preventing wildfires in Southern California and that 100-200 trees per acre is considered overstocked, whereas 40-60 trees per acre is considered healthy when seen through the lens of fire-prevention that the article presents. If you have another source for these figures I'd be interested.
My point here is that figures for average trees/acre in a forest vary depending on the particulars, and in most cases they are expressing an ideal based on whatever forest management goals the author has in mind, whether it's maximizing productivity of the land for harvesting lumber, preventing fires, or some other goal. Assuming the forests in your fantasy world may or may not be under similar management regimens, the figures available may or may not be relevant. When all this is taken together, I think the 871 to 1,742 trees/acre yielded by following the SRD rules are within the realm of plausibility. I also think that what qualifies as a forest in the real world hasn't been well defined.
Again, I don't play 3.5E, but when I design outdoor encounters for my 5E games, I use a homebrew system that randomizes terrain features based on the prevailing terrain. So, for example, if an encounter takes place in a forest, I generate terrain for each quadrant of the map from a table that includes the following results: featureless, dense woods, light woods, dense brush, rough ground, slope, hill, marsh, stream, or pond. I mention this because what I think the 3.5 SRD system is lacking is an acknowledgement that what we call a forest in the real world is actually made up of many terrain elements that merge in complex and varying ways. I also think that keeping that in mind may help with the dissatisfaction you seem to be expressing with the 3.5E system. A forest isn't just made up of wooded areas. There are meadows, floodplains, rockfalls, etc., where very few trees grow but are nevertheless considered part of the forest.