I used to farm 450 Swaledale ewes and 80 North Country Mules on 1000 acres of Northumberland moorland. About 300 acres was environmentally globally important mire, but dangerous to humans and livestock so was fenced off. Of the remaining 700 acres, about 50 was improved to varying degrees, some of it even mowable; about 500 was the roughest, wettest grazing you can get, molinia mostly, but supported a hen harrier winter roost and 5 long-eared owls in summer, and breeding populations of lapwings, curlews, snipe and grouse (including one black grouse spotted one time only) plus a variety of rare wetland plants; and the other 150 was in between the two, being rough but with drainage "grips" which meant the ground could be excellent grazing in summer, a plethora of different species of plants on it.
Understocking would lead to almost monoculture molinia over time, so I do get what you mean about understocking being at least as bad as overstocking. But... overstocking when it was very wet (and up there, it often rained almost non-stop from October to March) made it hell on earth for the sheep and everyone trying to look after them.
Specific comments derived from my experience up there :
1. It needed cattle (appropriate native types) to get that ground back and then to help keep it right. (A very typical story in English farming... Natural England paid the previous farmer to take all the cattle off and keep the sheep numbers down, then at the end of that 10 year scheme, when it a lot of it had become overgrown with molinia, they paid the next farmer to put cattle back on...

)
2. Winters were very tough on the livestock and the humans looking after them. Cattle needed to be housed for at least 4 months, sometimes 6, or they made too much mess of the ground as it was so very wet. Hoggs (homebred replacements which we didn't put in lamb until their second year) needed cake to support them growing and getting through their first winter. The previous farmer used to send his hoggs away to easier ground for their first winter, and I think that may have been the better option; it kept their headcount off the ground over winter when it could support fewer, and they grew on better in easier circumstances. The in-lamb ewes really need the ground to be as good as possible to get them and their unborn lambs through the winter, so reducing the headcount over winter would help them too. One year, instead of breeding our own, we bought in two batches of excellently bred 2 and 3-crop draft ewes from hard farms, but it wasn't a great success. That sort of sheep and that sort of ground, they are hefted really; it took the incomers more than 12 months to settle in and they never really thrived.
3. Because it was a hard farm, our 3- and 4-crop draft ewes were sought after by easier but still upland farms, where they would do well for another couple of crops at least. Farmers who knew how to manage them and the transition would pay a decent price, so long as we only drafted good ewes and culled any we couldn't stand behind. But drafting at 4 crops latest meant we needed a lot of replacement ewes, so to maintain our 450 Swales to the tups each year, we would have to retain more than 100 good ewe lambs each year, which meant breeding more than a quarter of the Swales pure rather than to the Blue-faced Leicester for our main product, the North Country Mule ewe lamb.
4. We had permission to feed silage on the areas of moorland which had the drainage grips, provided we moved the ring for each big bale to reduce poaching. The sheep thrived on the home-produced silage (haylage really) but even moving the ring every bale, the underlying ground was damaged by the silage - you could see the scars on the satellite pictures, and the ground would not cope with waterlogging as well for a couple of years after. (Hay was much kinder to the ground, but it was a wet area and it wasn't always possible to make hay. Plus if it had been a hard lambing, and a wet spring / early summer, we might struggle to shut the hay fields up early enough to get hay.)
Sorry, that's all rather jumbled, but I am trying to say that ramping the numbers up may cause a lot of knock-on effects.