No matter where you put the few cubic kilometers of water that pumped hydro requires, you need to put it somewhere. The ecosystem of that place will then be displaced or destroyed.
> It does not need more than the tiniest fraction of a cubic km of water.
One cubic km is one billion tons. Each ton is ~10kJ per meter lifted, and let's assume you can separate your reservoirs by 100m vertically.
That would give a total system capacity of 1e9 tons x 1e5 J/(ton x meter) x 1e3 meters, or 1e17 Joules.
A kilowatt hour is 3.6e6 Joules, so this gives us ~3e10 KWh. If we're imperfect in reclaiming that energy, we'll actually end up with 1-2e10 KWh.
US annual electricity consumption was ~4e12 KWh/year in 2018 (or ~1e10 KWh/day), so a 1km^3 * 100m installation is probably larger than required to provide a backup for replacing our entire electrical generation infrastructure with intermittent sources 99.9% of the time. Not by a huge amount though! Less than an order of magnitude, IMO.
> A reservoir can increase scarce habitat for migratory birds, which is commonly much more valuable than your typical bald hilltop.
How much is the level going to fluctuate on a daily basis, though? Does it really form useful habitat?
There are plenty of hills much more than 100m high. People will favor using the best places for it.
Efficiency of pumped hydro is always better than 70%, round-trip, and commonly better than 85%.
There is no reason to put all your storage in one place, and great reasons not to. A site moving 0.0001 km³ is useful for utility-scale storage. Of course, pumped hydro will be only one storage method among many.
Birds mostly use the top of the water. Probably, the lower pond is easiest to tailor for multiple use, but "floatovoltaics" are probably better in the upper one.
